Short-Term Power Load Forecasting Based on a Combination of VMD and ELM

Size: px
Start display at page:

Download "Short-Term Power Load Forecasting Based on a Combination of VMD and ELM"

Transcription

1 Pol. J. Evro. Stud. Vol. 7, No. 5 (08), DOI: 0.544/pjoes/7844 ONLINE PUBLICATION DATE: Orgal Research Short-Term Power Load Forecastg Based o a Combato of VMD ad ELM We L*, Cogx Qua, Xuyag Wag, Shu Zhag Departmet of Ecoomcs ad Maagemet, North Cha Electrc Power Uversty, Baodg, Cha Receved: 3 August 07 Accepted: 3 October 07 Abstract Accurate short-term power load forecastg s becomg more ad more mportat for the stable operato ad mproved ecoomc beefts of electrc power systems. However, whe affected by varous factors the power load shows o-lear ad o-statoary characterstcs. I order to forecast power load precsely, we propose a extreme learg mache (ELM) combed wth varatoal mode decomposto (VMD), as a ew hybrd tme seres forecastg model. I the frst stage, sce decomposed modes ad hdde layer odes have great fluece o predcto accuracy, a three-dmesoal relatoshp has bee establshed to determe them advace. I the secod stage, usg VMD, the tme seres of power load s decomposed to predetermed modes that are the used to costruct trag parts ad forecast outputs. The every dvdual mode s tae as a put data to the ELM. Evetually, the thrd stage, the fal forecasted power load data s obtaed by aggregatg the forecastg results of all the modes. To testfy the forecastg performace of the proposed model, a fve-mute power load data Hebe of Cha s used for smulato, ad comprehesve evaluato crtera s proposed for quattatve error evaluato. Smulato results demostrate that the proposed model performs better tha some prevous methods. Keywords: short-term power load forecastg, extreme learg mache (ELM), varatoal mode decomposto (VMD) Itroducto I recet years, short-term power load forecastg s playg a creasgly mportat part the stable operato of electrcal power systems, whch s essetal for the developmet of prevetve mateace plas [-]. Wth creasgly maret-oreted electrc power producto ad cosumpto, more accurate short-term power load forecastg s eeded for matag the secure ad stable operato of the electrcal power grd, whch ca also promote the sustaable developmet of *e-mal: cepulw@6.com the electrcty dustry [3]. Therefore, t s dspesable to develop power load forecastg techques to acheve a more accurate smulato result [4-5]. Thus far, short-term power load forecastg methods maly clude covetoal methods, tellget methods, ad hybrd forecastg methods. Covetoal methods maly clude may mathematcal statstcs methods such as multple lear regresso aalyss [6], the grey-forecastg model [7], autoregressve tegrated movg average (ARIMA) [8-9], state space model [0], box-jes model [], geeral expoetal smoothg [], ad so o. Although these methods have the advatage of capturg lear characterstcs, they caot accurately forecast short-term power load

2 44 L W., et. al. showg o-lear ad o-statoary characterstcs. To capture olear characterstcs ad obta accurate predcto results, may tellget methods such as artfcal eural etwor (ANN) [3], support vector regresso (SVM) [4-5], ad tellget evolutoary algorthms [6] have bee proposed to forecast shortterm power loads. ANN ad SVM are well-ow models, ad they have obvous predcto advatages compared wth covetoal methods. Tag to accout the fact that each sgle forecastg model has ts ow predcto shortcomgs, a creasg umber of researchers are adoptg hybrd forecastg methods because they ca avod the shortcomgs of each sgle forecastg model ad deal wth the complex problem more effectvely. ARIMA ad Grey predcto models, as typcal represetatves of covetoal methods, have bee wdely used ad appled to solve may forecastg problems. The ma dea of these models s to loo for the tred of data chages, accordg to the tred equato, to forecast future data [7]. They are sutable to smooth ad lear tme seres predcto, ot good at graspg the tred of olear sequece chages. Wth the rapd developmet of the artfcal tellgece method, artfcal tellgece forecastg techology has bee gradually replacg covetoal forecastg techology the feld of power load forecastg. Bac propagato (BP) eural etwor, as a commo ANN, has the ablty to lear by tself, ad t has bee proposed by may researchers to apply to the short-term power load forecastg problem [8]. But the BP eural etwor s easly trapped local mmum value [9]. Hece, may optmzato algorthms have bee proposed, but the computato burde s greatly creased. Compared wth the BP eural etwor, the SVM algorthm has advatages of good robustess ad strog geeralzato ablty, avodg fallg to local mmum [0]. But the trag tme of the SVM creases expoetally by creasg the dmeso of put vector. A ew learg algorthm for the sglehdde-layer feed-forward eural etwor (SLFN) called the extreme learg mache (ELM) has recetly bee proposed []. I the learg process of ELM, sce ELM requres oly a sgle-pass trag stage wthout ay terato for weght adjustmet, t has very fast learg processg []. Ad t s able to solve the problem of stoppg crtera, learg rate, learg epochs, ad local mmum [3]. I recet years, the ELM has bee appled varous felds ad has become a sgfcat method olear modelg [4-6]. Cosderg the o-statoary ad chaotc propertes of short-term power load data, the tme seres decomposto techque s appled to decompose the orgal power load sequece to several subsequeces. Emprcal mode decomposto (EMD), based o Hlbert-Huag trasform (HHT), ca effectvely extract the compoets of the basc mode from o-lear or o-statoary tme seres. Owg to ts attractve features, t has bee appled to may ecoomc ad facal ssues such as predctg crude ol prces [7], electrcty prces [8], ad so o. But EMD has may dsadvatages, cludg lac of exact mathematcal model, terpolato choce, ad sestvty to both ose ad samplg [9]. The a ew multresoluto called varatoal mode decomposto (VMD) was troduced as a alteratve to the EMD algorthm to overcome ts lmts. VMD has the ablty to separate toes of smlar frequeces, cotrary to EMD [30]. Ad much lterature [3-33] has come to the cocluso that VMD was more effectve modelg ad forecastg ecoomc ad facal data compared wth EMD. The ths paper, hybrd ELM-VMD tellgece s proposed for forecastg short-term power load. I partcular, the umber of modes decomposed by the VMD model maes a great fluece o aalyss, ad the umber of hdde layer odes plays a mportat role predcto accuracy. Ad both of them must be predetermed. So, how to determe the umber of modes ad the umber of hdde layer odes s a hghlght of ths paper. The pot s specfcally made the remader of ths paper. After determg the umber of modes ad the umber of hdde layer odes, the VMD s appled to decompose power load data to several modes that have specfc sparsty propertes. The ELM s costructed to forecast these modes dvdually. Ad all of these forecastg values are aggregated to produce the fal power load forecastg results. Evetually, by cotrast, the proposed method s compared wth other alteratve models such as the GM (,), ARIMA, sgle BP, sgle SVM, sgle ELM, hybrd BP-EMD, hybrd BP-VMD, ad ELM- EMD models. Smulato results demostrate that the proposed model performs more effectvely ad obtas more accurate predcto results. Materal ad Methods Varatoal Mode Decomposto (VMD) The purpose of the VMD s to decompose the orgal sgal to K dscrete umber of modes that have specfc sparsty propertes whle producg the ma sgal. Thus, each mode u s requred to be mostly compact aroud a ceter pulsato ω determed alog wth the decomposto. Compared wth EMD, VMD has the ablty to separate toes of smlar frequeces. Meawhle, the VMD model searches for a umber of modes ad ther respectve ceter frequeces, such that the bad-lmted modes reproduce the put sgal exactly or a least-squares sese. The VMD algorthm to assess the badwdth of a oe-dmesoal sgal s as follows [4]: ) For each mode u, compute the assocated aalytc sgal by meas of the Hlbert trasform to obta a ulateral frequecy spectrum.

3 Short-Term Power Load Forecastg Based ) For each mode, shft the mode s frequecy spectrum to basebad (arrow frequecy) by mxg wth a expoetal tued to the respectve estmated ceter frequecy. 3) Estmate the badwdth through Gaussa smoothess of the demodulated sgal; for example, the squared L-orm of the gradet. The, the decomposto process s realzed by solvg the followg optmzato problem [4]: ω + ω u ω d ˆ ( ) + ω 0 = + uˆ ( ω) dω 0 ˆ ˆ λ + ( ω) = λ ( ω) + τ f ( ω) uˆ + ( ω) (5) (6) j jω m = t δ ( t) * u ( t) e u, ω + πt Subject to u = f t () where f s the sgal to be decomposed, u s ts mode, ω represets the frequecy, δ s the Drac dstrbuto, t represets tme scrpt, * deotes covoluto, j =, ad s the umber of modes. I ths framewor, the mode u wth hgh-order expresses low frequecy compoets. To address the costraed varatoal problem oted above, a quadratc pealty factor α ad Lagrage multplers λ(t) are troduced. The combato of these two terms beefts from both cosderable covergece propertes of the quadratc pealty at fte weght ad the strct eforcemet of the costrat by the Lagraga multpler. Therefore, the above optmzato problem s chaged to a ucostraed oe as below [4]: () where represets the terato umber ad τ devotes the tme step of the dual ascet. Ad the cocrete procedures of the VMD algorthm are llustrated Fg.. Extreme Learg Mache The extreme learg mache (ELM) was a ew method for learg the sgle-layer feed-forward eutral etwor (SLFNs), whch was frst troduced by Huag et al. []. It teds to provde superor geeralzato performace ad extremely fast learg speed by j L( { u},{ ω}, λ ) = α t σ ( t) + * u ( t) e πt e jw t + f ( t) u ( t) + λ( t), f ( t) u ( t) The optmzato methodology deoted as the alterate drecto method of multplers (ADMM) s the used to obta the saddle pot of the augmeted Lagraga by updatg u, ω, ad λ alterately. The covergece crtero of the algorthm + ( u ˆ uˆ / uˆ ) < ε s, where ε s the covergece tolerace. So, the fal updated equatos are gve as follows [4]: (3) ˆ ˆ λ ( ω ) f uˆ u + + ( ω) ( ω) ( ω) + < > ω = + α( ω ω ) uˆ ( ) (4) Fg.. Cocrete procedures of the VMD algorthm.

4 46 L W., et. al. where H ( W, W,..., W, b, b,..., b,x,x,..., X ) = l l N f ( W, b, X)... f ( Wl, bl, X) M M M L f ( W, b, X N ) f ( Wl, bl, X N ) N l T η η= M, T η l l m T T t = M T t N N m. Fg.. Typcal structure of the ELM. radomly choosg the put weghts ad aalytcally determg the output weghts [4]. Ad oly the umber of hdde layer euros s set, whch case the uque optmal soluto wll be obtaed. The typcal structure of ELM s show Fg., whch cludes put, hdde, ad output layers. Assumg that the etwor has p put layer euros ad l hdde layer euros, for N arbtrary dstct samples m ( X, t ) R R the SLFN wth hdde euros ca be descrbed as: l j= f ( W, b, X ) η = o =,,..., N j j j where W j = [W, W,...,W ] T s the put weght vector coectg the put odes ad the j th hdde ode, η j = [η j, η j,...,η j ] T s the weght vector coectg the jth hdde euro ad output euro, X s the th trag example, o j = [o, o,...,o ] T s the output vector, b j s the bas of the j th hdde ode, ad f(.) deotes the actvato fucto of the hdde euro. If the above ELM ca approxmate the N samples wth a zero error, the we ca obta: N = o y = 0 Hece parameters η j, b j, ad W j meet the codtos: l j= (7) (8) f ( Wj, bj, X ) η j = t =,,..., N (9) The above equato ca be descrbed as aother form: Hη = T (0) Dfferet from the tradtoal fucto approxmato theory, the put weghts ad hdde bases of the ELM are radomly geerated. Thus, trag the ELM s approprately equal to fd a least-square soluto of the lear fucto Hη = T[]. H ˆ η T = m Hη T The the soluto of the above form s η + ηˆ=h T () () where H + represets the Moore-Perose geeralzed verse of the hdde layer s output matrx H. Evaluato Crtero To quattatvely measure the performace of the proposed model, mea absolute percetage error (MAPE), root mea square error (RMSE), ad the coeffcet of determato (R ) are used to calculate forecastg accuracy. The overall fttg effect of the proposed model ca be evaluated by RMSE ad R. The MAPE s also oe mportat dcator, ad the smaller the MAPE values, the closer the predcted power load tme seres values to those of the actual value. The deftos of the above-metoed crtera are: y ˆ y MAPE = y = RMSE = (y yˆ ) R = (3) (4) ( y ˆ y ) = = (y % ˆ y ) = (5)

5 Short-Term Power Load Forecastg Based y ad ˆ where y represet the actual ad forecastg power load values at tme, respectvely; y~ s mea value of the predcted value; ad s the umber of the all load data. Approaches of ELM-VMD Model It s worth otg that both the umber of hdde layer odes l ad the umber of decomposed subsgal K play a mportat role predcto accuracy. Thus, ths paper t maes sese to fd ther optmal combatos, whch ths secto dscusses detal. Sce the umber of sub-sgals decomposed by VMD requres t to be gve advace, the rage of chage vares depedg o the umber of IMFs obtaed by EMD. For stace, f the umber of IMFs s, the parameter K used to determe the umber of decomposto to obta by EMD s 3,,,, +, +, ad +3 [34]. I order to avod mssg the optmal umber of hdde layer odes, ths paper the umber of hdde layer odes s set wth a wde rage from 5 to 00. I ths paper, predetermed trag samples are dvded to sub-trag samples ad subtestg samples. MAPE, a bechmar for evaluatg predcto accuracy, s used to determe the optmal combatos. Thus, there s a three-dmesoal relatoshp amog them. The specfc steps are preseted as follows: ) Determe sub-trag samples ad sub-testg samples. ) Apply VMD to decompose the orgal power load data to K statoary modes wth dfferet frequeces. 3) A rollg forecastg process s studed, ad ELM s used to forecast each mode decomposed by VMD; order to mprove the forecastg accuracy of the ELM, all the orgal power load seres ad decomposed sub-seres must be ormalzed prmarly as follows: Fg. 3. Forecastg procedures of the ELM-VMD forecastg model. (6) where x s orgal sample data, represets ormalzed sample data, ad x max ad x m represet the maxmum ad mmum of the data 4) The fal power load forecastg result s obtaed by aggregatg all de-ormalzed values of the ELM outputs. 5) Calculate MAPE ad obta l ad K wth the mmum value of MAPE; the l ad K are determed. I fact, parameters l ad K are obtaed after 7 96 = 67 calculatos. The, usg the parameters already obtaed, repeat steps (), (3), ad (4) to forecast short-term power load. The flowchart of the forecastg procedure s show Fg. 3. Results ad Dscusso Descrpto of Data To verfy the effectveess of the proposed ELM-VMD model, fve-mute load data from Hebe Provce Cha was used ths case study. The power load data obtaed from oe day, wth a total of 88 data pots, was selected as expermetal samples. A rollg forecastg process s studed whch the prevous half-hour data s used to forecast the ext data. That s, the frst sx load data pots are used to forecast the seveth data, amely the last fve load data (x 6, x 5, x 4, x 3, x, x ) are used as the put varables of the ELM, ad the output varable s x. The detals are show Fg. 4.

6 48 L W., et. al. Fg. 4. A rollg-forecastg mechasm for short-term power load forecastg: a) for trag set ad b) for testg set. I ths paper, the prevous hours of data, cludg 5 data pots, were used as the trag set, ad the remag 36 data pots were used as the testg set. To determe the parameters l ad K, the prevous 8 hours of the trag set data were selected as sub-trag set ad the remag data pots were used as the sub-testg set. All computatos are mplemeted a MATLAB (05a) evromet o a computer wth Itel core 5,.6 to.3 GHz CPU, ad 4 GB of RAM. Fg. 5. Orgal short-term power load tme seres. Determe Parameters l ad K The fluctuato of short-term power load s severe, as s show Fg. 5. Thus, EMD s exploted to decrease the o-statoary characterstcs of the orgal shortterm power load tme seres. The results are show Fg. 6, whch the short-term power load data s decomposed to sx dfferet IMFs ad oe resdue. It ca be clearly see that the frequeces of IMF ad IMF are too hgh, whch maly dcates the radom formato of power load. The frequeces of IMF3 to IMF6 are moderate, whch actually are the perodc compoets of the orgal short-term power load seres. Ad the resdue represets the log-term varato tedecy of the orgal power load seres. The, parameter K the umber of decomposed modes vares from 3 to 9, accordg to the umber of IMFs decomposed by EMD. Sce the decomposed modes are ormalzed betwee 0 ad, the sgmod s selected as actvato fucto ths paper. The subtrag power load data s used to tra ELM, ad the sub-testg sample s used to calculate MAPE. Because the mmum value s ot easy to observe the mage, the verse of MAPE s troduced to have

7 Short-Term Power Load Forecastg Based Fg. 6. EMD decomposto results for short-term power load tme seres; from top to bottom: frst IMF to resdue. a better represetato ad uderstadg of obtaed results, whch s show Fg. 7. It s clearly observed that the verse of MAPE reaches a maxmum wth 8 decomposed modes together wth 6 hdde layer odes, amely achevg the best performace ths case. The other coclusos we ca draw are that wth the crease of decomposto levels, the predcto accuracy s bascally creasg. After 8 decomposed levels, the predcto remas almost costat. Therefore, the decomposto of sgal to 8 levels s the best choce, whch smply creases the computatoal burde wth more decomposto levels. Ad wth the umber of hdde layer odes creasg, the predcto error frst creases ad the decreases. We ca observe that the predcto effect wll eep well whe the hdde odes rage approxmately from 0 to 50. The the orgal short-term power load seres was decomposed to 8 modes by VMD (Fg. 8). Fgure 8 shows that the frst ad secod modes prmarly represet the log-term tred of power load seres. Whle the last two modes, wth hgh frequecy, capture short varatos of power load seres. Fg. 7. Results of determg parameters usg MAPE as error dex. Short-Term Power Load Forecastg Results To verfy the effectveess of the ELM as forecastg part ths paper, may forecastg models exstg other lterature such as GM (,), ARIMA, BP eutral etwor, ad SVM are appled as cotrast. The the forecastg results are obtaed by usg orgal power load data that has ot bee decomposed. The parameters of BP eutral etwor are set as follows. Obvously, the put layer ode ad output layer ode were set to 6 ad, respectvely. As a geeral rule of thumb, the hdde layer ode was set to 0 ths paper. The maxmum umber of trag steps was set to 00, the learg rate was set to 0., ad the error precso was set to As for SVM model, the sgmod erel fucto s selected to map the orgal feature space to a hgh-dmesoal space. The epslo loss fucto of epslo-svm s set to 0.0. Owg to the radomess of predcto results for BP eural etwor, the BP eural etwor rus 0 tmes ad the average value of the predcted results s used as the fal result. The, the obtaed results are show Table ad the absolute error values are depcted Fg. 9 for the above-metoed forecastg tools. The forecastg errors of ARIMA ad GM (,) are too large, so ther absolute error values are ot show the above Fgure 9 for the rederg effects of the whole mage. As ca be see from Table, the ELM has the best forecastg accuracy amog all forecastg tools. The forecastg accuracy of GM (,) ad ARIMA s far lower tha other forecastg tools because they are sutable for stable ad smooth tme seres. Compared wth them, tellgece algorthms such as BP eutral etwor, SVM, ad ELM are able to eep o learg, the they ca deal wth o-lear ad o-statoary tme seres more effectvely. The MAPE, RMSE, ad R of the ELM are superor to the BP eutral etwor because the heret characterstcs of the BP eutral etwor may fall

8 50 L W., et. al. Fg. 8. VMD decomposto results for short-term power load tme seres; from top to bottom: frst mode to eghth. to local optmum ad the hdde odes of BP eural etwor hghly deped o tral ad error procedure. The SVM usually has very good forecastg performace whe the put ad output dmesos are hgh, but the SVM does ot wor well ths rollg predcto. Ad t eve performs worse tha the BP eutral etwor. From the stadpot of computg tme usg ELM, the computg tme of the forecastg process s s, whch s very short amog all appled forecastg tools. Although the computg tme of GM (,) s rather short ( s), t has large forecastg errors. What s more, the SVM has the logest forecastg tme of s. If SVM combed wth the VMD s proposed, the forecastg tme wll be approxmately 44 8 = 35 s. Compared wth the ELM-VMD, ts forecastg tme s much greater tha the computg tme of the ELM-VMD (approxmately = 0.48 s). Thus, the SVM model combed wth other sgal decomposed techques was ot tae to cosderato because of ts too-log calculato tme. Moreover, to demostrate the superorty of applyg VMD to sgal aalyss, EMD as a sgal processg tool s troduced to provde comparso results. Whe t comes to the ELM-EMD model, to be cosstet wth the proposed method, usg sub-trag set ad sub-testg set to ga the predetermed optmal hdde odes. The result s show as Fg. 0. Thus, ths codto the hdde odes of the ELM were set to. Whe t comes to the BP-VMD model, the predetermed umber of decomposto K rages from 3 to 9, the BP eural etwor also rus 0 tmes ad obtas average value. The obtaed forecastg errors of hybrd forecastg methods are gve Table. It ca be see that the MAPE ad RMSE of the hybrd ELM-VMD model are the smallest ad the goodess of ft reaches 0.998, whch s the best of all models. As Tables ad show, the hybrd models have better predcto accuracy tha the correspodg sgle format. Ths dcates that the forecastg performace wll be mproved after Table. Error statstcs for sgle forecastg model. Forecastg model Error dex MAPE (%) RMSE (MW) R (MW) Computg tme (s) GM(,) ARIMA BP SVM ELM (proposed model)

9 Short-Term Power Load Forecastg Based... 5 Fg. 9. Absolute error of sgle models the short-term power load forecastg. orgal data are decomposed by sgal processg tools. It s worth otg that the BP-EMD model has worse performace tha the sgle BP model because the BP eutral etwor may fall to the local mmum value. The forecastg fte IMFs ad resdue are added to obta the fal result, ad the forecastg error may crease. As show Table, the VMD-based approach performs better tha the EMD-based approach terms of MAPE ad RMSE, whch llustrates the advatage of the VMD-based approach for forecastg short-term power load. Oe possble reaso s that the EMD has the dsadvatage of mode mxg, whle the EMD has the ablty to separate toes of smlar frequeces. Table also showed that the ELM-VMD model outperformed the BP-VMD model terms of MAE, RMSE, ad R for all values of K. The BP-VMD model performs better tha the ELM-EMD model terms of MAPE ad RMSE, wth K = 6,7,8,9. Ths also meas that sgal preprocessg has great fluece o predcto accuracy. The absolute error values are also depcted Fg.. For dverse decomposed modes, t has dfferet forecastg results of the BP-VMD. The case of the best performace wth K = 8, as the represeted case of all cases for the BP-VMD model s show Fg.. I summary, the expermetal results fully demostrate the effcecy of ELM power load forecastg ad the advacemet of VMD sgal processg. At the same tme, t cofrms that establshg three-dmesoal relatoshps to determe coeffcets l ad K s a good dea to obta Fg. 0. Result of determg hdde odes usg MAPE as error dex.

10 5 L W., et. al. Table. Statstcs error for all hybrd forecastg models. Error dex Hybrd model MAPE (%) RMSE (MW) R (MW) BP-EMD K = K = K = BP-VMD K = K = K = K = ELM-EMD ELM-VMD (proposed hybrd model) accuracy forecastg results. Ad t also dcates that the proposed model s effectve ad effcet for shortterm power load forecastg. I fact, the forecastg process tme of ELM-VMD s approxmately 0.48s, ad ts computatoal tme s much smaller tha other hybrd models. Coclusos I ths paper, a ew hybrd short-term power load forecastg model based o ELM ad VMD has bee proposed. Frst of all, t uses a sub-trag set ad subtestg set, establshg three-dmesoal relatoshps, to obta the umber of varatoal modes ad hdde odes of ELM. The, cosderato of the volatlty of power load, the orgal load seres s decomposed to predetermed modes. Next, rollg predcto s studed step by step ad the ELM s appled to forecast dvdual modes. Fally, the dfferet forecastg seres are recostructed to get the fal forecastg results. Expermets wth dfferet statstcal crtera (MAPE, RMSE, R ) clearly testfy that the ELM- VMD model acheved the lowest forecastg error. Fg.. Absolute error of all hybrd models short-term power load forecastg.

11 Short-Term Power Load Forecastg Based... Compared wth other hybrd or sgle models, t has a very fast forecastg process ad saves sgfcat computatoal tme, whch meas that the ELM-VMD model ca be used as a very promsg methodology for short-term power load predcto. I ths paper, several other coclusos ca be acqured as follows: A) Establshg three-dmesoal relatoshps to obta the umber of varatoal modes ad hdde odes of ELM, as a hghlght of ths paper, acheves fe performace. B) As a ew adaptve multresoluto techque, VMD has better performace ad t s more robust for aalyzg osy sgals tha EMD. C) The tellget algorthm s more sutable for predctg the o-statoary ad olear tme seres tha covetoal method. D) Compared wth other forecastg algorthms the hybrd models, the ELM has the lowest forecastg error ad the shortest computato tme. As a result, the proposed hybrd model ths paper s smple to mplemet ad t provdes fast learg ad covergece whe the sample sze s large. The whole process from selectg optmal parameters to obtag the fal forecastg result s easy to uderstad. Thus, all of these reasos mae t more sutable for forecastg short-term power load ad eve other tme seres. However, ths paper, oly the oe-step-ahead forecastg model s costructed, wth regards to future research drectos, to costruct the mult-step-ahead forecastg model should also be tae to accout. Besdes, short-term power load forecastg s equvalet to tme seres forecastg ths paper, ad other factors such as temperature, wd power, ad precptato are excluded from the proposed hybrd model. Thus, a future study should verfy the superorty of the proposed model mult-step-ahead forecastg ad corporate those fluecg factors to develop the comprehesve hybrd forecastg model. Acowledgemets The curret wor s supported by the Natoal Socal Scece Foudato of Cha (grat No. 5BGL45), the Natoal Natural Scece Foudato of Cha (grat No ), the Fudametal Research Fuds for the Cetral Uverstes (No. 06MS5), ad the Phlosophy ad Socal Scece Research Base of Hebe Provce. Refereces. Sastad A.H., McMeam S., Sue A., Barbose G.L., Goldma C.A. Modelg a aggressve eergy-effcecy scearo log-rage load forecastg for electrc power trasmsso plag. Appled Eergy, 8 (3), 65, L C., L, S., Lu Y. A least squares support vector mache model optmzed by moth-flame optmzato algorthm for aual power load forecastg. Appled Itellgece 45 (4), 66, Wag J., L L., Nu D., Ta Z. A aual load forecastg model based o supportvector regresso wth dfferetal evoluto algorthm. Appled Eergy, 94 (6), 65, Nu D., Wag Y., Dua C., Xg M. A New Shortterm Power Load Forecastg Model Based o Chaotc Tme Seres ad SVM. Joural of Uversal Computer Scece, 5 (3), 76, Zhou X., Zhou X., Zhag Z.M., Tetzers M.M. Short-term power load forecastg usg grey corelato cotest modelg. Expert Systems wth Applcatos, 39 (), 773, Hog T., Gu M., Bara M.E., Wlls H.L. Modelg ad forecastg hourly electrc load by multple lear regresso wth teractos. Power & Eergy Socety Geeral, Meetg, -8, L W., Ha Z.H. Applcato of mproved grey predcto model for power load forecastg. Iteratoal Coferece o Computer Supported Cooperatve Wor Desg, 6, Ne H., Lu G., Lu X., Wag Y. Hybrd of ARIMA ad SVMs for Short-Term Load Forecastg. Eergy Proceda, 6 (5), 455, Nazaro J., Jurczu A., Zalews W. ARIMA models load modellg wth clusterg approach. Power Tech, IEEE Russa, -6, Dordoat V., Koopma S.J., Ooms M., Dessertae A., Collect J. A hourly perodc state space model for modellg Frech atoal electrcty load. Iteratoal Joural of Forecastg, 4 (4), 566, Vähäyla P., Haoe E., Léma P. Short-term forecastg of grd load usg Box-Jes techques. Iteratoal Joural of Electrcal Power & Eergy Systems, (), 9, Chrstaase W.R. Short-term load forecastg usg geeral expoetal smoothg. IEEE Trasactos o Power Apparatus & Systems, 90 (), 900, Xa F., Fa L. Applcato of Artfcal Neural Networ (ANN) for Predcto of Power Load. Sprger Berl Hedelberg, 5, 673, Wu Q. Power load forecasts based o hybrd PSO wth Gaussa ad adaptve mutato ad W v -SVM. Expert Systems wth Applcatos, 37 (), 94, Lu B., Yag R. A ovel method based o PCA ad LS- SVM for power load forecastg. Iteratoal Coferece o Electrc Utlty Deregulato & Restructurg & Power Techologes, 759, Tzafestas S., Tzafestas E. Computatoal Itellgece Techques for Short-Term Electrc Load Forecastg. Joural of Itellgece & Robotc Systems, 3 (-3), 7, Heg J., Wag C., Zhao X., Wag J. A Hybrd Forecastg Model Based o Emprcal Mode Decomposto ad the Cucoo Search Algorthm: A Case Study for Power Load. Mathematcal Problems Egeerg, L L.J., Huag W. A Short-Term Power Load Forecastg Method Based o BP Neural Networ. Appled Mechacs & Materals, , 647, Hog W.C. Chaotc partcle swarm optmzato algorthm a support vector regresso electrc load

12 54 L W., et. al. forecastg model. Eergy Coverso & Maagemet, 50 (), 05, Nu D., Da S. A Short-Term Load Forecastg Model wth a Modfed Partcle Swarm Optmzato Algorthm ad Least Squares Support Vector Mache Based o the Deosg Method of Emprcal Mode Decomposto ad Grey Relatoal Aalyss. Eerges, 0, 07.. Huag G.B., Zhu Q.Y., Sew C.K. Extreme learg mache: Theory ad applcatos. Neurocomputg, 70 (), 489, Abdoos A.A. A ew tellget method based o combato of VMD ad ELM for short term wd power forecastg. Neurocomputg, 03,, Fgueredo E.M.N., Ludermr T.B. Ivestgatg the use of alteratve topologes operformace of the PSO-ELM. Neurocomputg, 7 (3), 4, Xa M., Zhag Y., Weg L., Ye X. Fasho retalg forecastg based o extreme learg mache wth adaptve metrcs of puts. Kowledge-Based Systems, 36 (6), 53, Zhag C., Zhou J., L C., Fu W., Peg T. A compoud structure of ELM based o feature selecto ad parameter optmzato usg hybrd bactracg search algorthm for wd speed forecastg. Eergy Coverso & Maagemet, 43, 360, Wag X., Zhag H., Guo X. Demad Forecastg Models of Toursm Based o ELM. Seveth Iteratoal Coferece o Measurg Techology & Mechatrocs Automato, 36, He K., Zha R., Wu J., La K. Multvarate EMD- Based Modelg ad Forecastg of Crude Ol Prce. Sustaablty, 8 (4), 367, He K., Yu L., Tag L. Electrcty prce forecastg wth a BED (Bvarate EMD Deosg) methodology. Eergy, 9, 60, Dragomretsy K., Zosso D. Varatoal Mode Decomposto. IEEE Trasactos o Sgal Processg, 6 (3), 53, Lahmr S. Comparg Varatoal ad Emprcal Mode Decomposto Forecastg Day-Ahead Eergy Prces. IEEE Systems Joural, 99,, Su G., Che T., We Z., Su Y., Zag H. A Carbo Prce Forecastg Model Based o Varatoal Mode Decomposto ad Spg Neural Networs. Eerges, 9 (), 54, Yag W., Peg Z., We K., Sh P., Ta W. Superortes of varatoal mode decomposto over emprcal mode decomposto partcularly tmefrequecy feature extracto ad wd turbe codto motorg. Let Reewable Power Geerato, (4), 443, Huag G.B., Zhu Q.Y., Sew C.K. Extreme learg mache: a ew learg scheme of feedforward eural etwors. IEEE Iteratoal Jot Coferece o Neural Networs,, 985, Lahmr S. A Varatoal Mode Decomposto Approach for Aalyss ad Forecastg of Ecoomc ad Facal Tme Seres. Expert Systems wth Applcatos, 55, 68, 06.

Research on SVM Prediction Model Based on Chaos Theory

Research on SVM Prediction Model Based on Chaos Theory Advaced Scece ad Techology Letters Vol.3 (SoftTech 06, pp.59-63 http://dx.do.org/0.457/astl.06.3.3 Research o SVM Predcto Model Based o Chaos Theory Sog Lagog, Wu Hux, Zhag Zezhog 3, College of Iformato

More information

Functions of Random Variables

Functions of Random Variables Fuctos of Radom Varables Chapter Fve Fuctos of Radom Varables 5. Itroducto A geeral egeerg aalyss model s show Fg. 5.. The model output (respose) cotas the performaces of a system or product, such as weght,

More information

Solving Constrained Flow-Shop Scheduling. Problems with Three Machines

Solving Constrained Flow-Shop Scheduling. Problems with Three Machines It J Cotemp Math Sceces, Vol 5, 2010, o 19, 921-929 Solvg Costraed Flow-Shop Schedulg Problems wth Three Maches P Pada ad P Rajedra Departmet of Mathematcs, School of Advaced Sceces, VIT Uversty, Vellore-632

More information

Analysis of Variance with Weibull Data

Analysis of Variance with Weibull Data Aalyss of Varace wth Webull Data Lahaa Watthaacheewaul Abstract I statstcal data aalyss by aalyss of varace, the usual basc assumptos are that the model s addtve ad the errors are radomly, depedetly, ad

More information

Study on a Fire Detection System Based on Support Vector Machine

Study on a Fire Detection System Based on Support Vector Machine Sesors & Trasducers, Vol. 8, Issue, November 04, pp. 57-6 Sesors & Trasducers 04 by IFSA Publshg, S. L. http://www.sesorsportal.com Study o a Fre Detecto System Based o Support Vector Mache Ye Xaotg, Wu

More information

Bayes Estimator for Exponential Distribution with Extension of Jeffery Prior Information

Bayes Estimator for Exponential Distribution with Extension of Jeffery Prior Information Malaysa Joural of Mathematcal Sceces (): 97- (9) Bayes Estmator for Expoetal Dstrbuto wth Exteso of Jeffery Pror Iformato Hadeel Salm Al-Kutub ad Noor Akma Ibrahm Isttute for Mathematcal Research, Uverst

More information

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions

Estimation of Stress- Strength Reliability model using finite mixture of exponential distributions Iteratoal Joural of Computatoal Egeerg Research Vol, 0 Issue, Estmato of Stress- Stregth Relablty model usg fte mxture of expoetal dstrbutos K.Sadhya, T.S.Umamaheswar Departmet of Mathematcs, Lal Bhadur

More information

Introduction to local (nonparametric) density estimation. methods

Introduction to local (nonparametric) density estimation. methods Itroducto to local (oparametrc) desty estmato methods A slecture by Yu Lu for ECE 66 Sprg 014 1. Itroducto Ths slecture troduces two local desty estmato methods whch are Parze desty estmato ad k-earest

More information

Kernel-based Methods and Support Vector Machines

Kernel-based Methods and Support Vector Machines Kerel-based Methods ad Support Vector Maches Larr Holder CptS 570 Mache Learg School of Electrcal Egeerg ad Computer Scece Washgto State Uverst Refereces Muller et al. A Itroducto to Kerel-Based Learg

More information

A New Method for Decision Making Based on Soft Matrix Theory

A New Method for Decision Making Based on Soft Matrix Theory Joural of Scetfc esearch & eports 3(5): 0-7, 04; rtcle o. JS.04.5.00 SCIENCEDOMIN teratoal www.scecedoma.org New Method for Decso Mag Based o Soft Matrx Theory Zhmg Zhag * College of Mathematcs ad Computer

More information

Comparison of Dual to Ratio-Cum-Product Estimators of Population Mean

Comparison of Dual to Ratio-Cum-Product Estimators of Population Mean Research Joural of Mathematcal ad Statstcal Sceces ISS 30 6047 Vol. 1(), 5-1, ovember (013) Res. J. Mathematcal ad Statstcal Sc. Comparso of Dual to Rato-Cum-Product Estmators of Populato Mea Abstract

More information

Feature Selection: Part 2. 1 Greedy Algorithms (continued from the last lecture)

Feature Selection: Part 2. 1 Greedy Algorithms (continued from the last lecture) CSE 546: Mache Learg Lecture 6 Feature Selecto: Part 2 Istructor: Sham Kakade Greedy Algorthms (cotued from the last lecture) There are varety of greedy algorthms ad umerous amg covetos for these algorthms.

More information

Analysis of Lagrange Interpolation Formula

Analysis of Lagrange Interpolation Formula P IJISET - Iteratoal Joural of Iovatve Scece, Egeerg & Techology, Vol. Issue, December 4. www.jset.com ISS 348 7968 Aalyss of Lagrage Iterpolato Formula Vjay Dahya PDepartmet of MathematcsMaharaja Surajmal

More information

Chapter 14 Logistic Regression Models

Chapter 14 Logistic Regression Models Chapter 4 Logstc Regresso Models I the lear regresso model X β + ε, there are two types of varables explaatory varables X, X,, X k ad study varable y These varables ca be measured o a cotuous scale as

More information

Journal of Chemical and Pharmaceutical Research, 2014, 6(7): Research Article

Journal of Chemical and Pharmaceutical Research, 2014, 6(7): Research Article Avalable ole www.jocpr.com Joural of Chemcal ad Pharmaceutcal Research, 04, 6(7):4-47 Research Artcle ISSN : 0975-7384 CODEN(USA) : JCPRC5 Predcto of CNG automoble owershp by usg the combed model Ku Huag,

More information

A COMBINED FORECASTING METHOD OF WIND POWER CAPACITY WITH DIFFERENTIAL EVOLUTION ALGORITHM

A COMBINED FORECASTING METHOD OF WIND POWER CAPACITY WITH DIFFERENTIAL EVOLUTION ALGORITHM Joural of Theoretcal ad Appled Iformato Techology 0 th Jauary 03. Vol. 47 No. 005-03 JATIT & LLS. All rghts reserved. ISSN: 99-8645 www.jatt.org E-ISSN: 87-395 A COMBINED FORECASTING METHOD OF WIND POWER

More information

ABOUT ONE APPROACH TO APPROXIMATION OF CONTINUOUS FUNCTION BY THREE-LAYERED NEURAL NETWORK

ABOUT ONE APPROACH TO APPROXIMATION OF CONTINUOUS FUNCTION BY THREE-LAYERED NEURAL NETWORK ABOUT ONE APPROACH TO APPROXIMATION OF CONTINUOUS FUNCTION BY THREE-LAYERED NEURAL NETWORK Ram Rzayev Cyberetc Isttute of the Natoal Scece Academy of Azerbaa Republc ramrza@yahoo.com Aygu Alasgarova Khazar

More information

Unimodality Tests for Global Optimization of Single Variable Functions Using Statistical Methods

Unimodality Tests for Global Optimization of Single Variable Functions Using Statistical Methods Malaysa Umodalty Joural Tests of Mathematcal for Global Optmzato Sceces (): of 05 Sgle - 5 Varable (007) Fuctos Usg Statstcal Methods Umodalty Tests for Global Optmzato of Sgle Varable Fuctos Usg Statstcal

More information

Systematic Selection of Parameters in the development of Feedforward Artificial Neural Network Models through Conventional and Intelligent Algorithms

Systematic Selection of Parameters in the development of Feedforward Artificial Neural Network Models through Conventional and Intelligent Algorithms THALES Project No. 65/3 Systematc Selecto of Parameters the developmet of Feedforward Artfcal Neural Network Models through Covetoal ad Itellget Algorthms Research Team G.-C. Vosakos, T. Gaakaks, A. Krmpes,

More information

Summary of the lecture in Biostatistics

Summary of the lecture in Biostatistics Summary of the lecture Bostatstcs Probablty Desty Fucto For a cotuos radom varable, a probablty desty fucto s a fucto such that: 0 dx a b) b a dx A probablty desty fucto provdes a smple descrpto of the

More information

Source-Channel Prediction in Error Resilient Video Coding

Source-Channel Prediction in Error Resilient Video Coding Source-Chael Predcto Error Reslet Vdeo Codg Hua Yag ad Keeth Rose Sgal Compresso Laboratory ECE Departmet Uversty of Calfora, Sata Barbara Outle Itroducto Source-chael predcto Smulato results Coclusos

More information

Comparing Different Estimators of three Parameters for Transmuted Weibull Distribution

Comparing Different Estimators of three Parameters for Transmuted Weibull Distribution Global Joural of Pure ad Appled Mathematcs. ISSN 0973-768 Volume 3, Number 9 (207), pp. 55-528 Research Ida Publcatos http://www.rpublcato.com Comparg Dfferet Estmators of three Parameters for Trasmuted

More information

A New Family of Transformations for Lifetime Data

A New Family of Transformations for Lifetime Data Proceedgs of the World Cogress o Egeerg 4 Vol I, WCE 4, July - 4, 4, Lodo, U.K. A New Famly of Trasformatos for Lfetme Data Lakhaa Watthaacheewakul Abstract A famly of trasformatos s the oe of several

More information

Part 4b Asymptotic Results for MRR2 using PRESS. Recall that the PRESS statistic is a special type of cross validation procedure (see Allen (1971))

Part 4b Asymptotic Results for MRR2 using PRESS. Recall that the PRESS statistic is a special type of cross validation procedure (see Allen (1971)) art 4b Asymptotc Results for MRR usg RESS Recall that the RESS statstc s a specal type of cross valdato procedure (see Alle (97)) partcular to the regresso problem ad volves fdg Y $,, the estmate at the

More information

A Method for Damping Estimation Based On Least Square Fit

A Method for Damping Estimation Based On Least Square Fit Amerca Joural of Egeerg Research (AJER) 5 Amerca Joural of Egeerg Research (AJER) e-issn: 3-847 p-issn : 3-936 Volume-4, Issue-7, pp-5-9 www.ajer.org Research Paper Ope Access A Method for Dampg Estmato

More information

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model

Lecture 7. Confidence Intervals and Hypothesis Tests in the Simple CLR Model Lecture 7. Cofdece Itervals ad Hypothess Tests the Smple CLR Model I lecture 6 we troduced the Classcal Lear Regresso (CLR) model that s the radom expermet of whch the data Y,,, K, are the outcomes. The

More information

L5 Polynomial / Spline Curves

L5 Polynomial / Spline Curves L5 Polyomal / Sple Curves Cotets Coc sectos Polyomal Curves Hermte Curves Bezer Curves B-Sples No-Uform Ratoal B-Sples (NURBS) Mapulato ad Represetato of Curves Types of Curve Equatos Implct: Descrbe a

More information

Median as a Weighted Arithmetic Mean of All Sample Observations

Median as a Weighted Arithmetic Mean of All Sample Observations Meda as a Weghted Arthmetc Mea of All Sample Observatos SK Mshra Dept. of Ecoomcs NEHU, Shllog (Ida). Itroducto: Iumerably may textbooks Statstcs explctly meto that oe of the weakesses (or propertes) of

More information

Cubic Nonpolynomial Spline Approach to the Solution of a Second Order Two-Point Boundary Value Problem

Cubic Nonpolynomial Spline Approach to the Solution of a Second Order Two-Point Boundary Value Problem Joural of Amerca Scece ;6( Cubc Nopolyomal Sple Approach to the Soluto of a Secod Order Two-Pot Boudary Value Problem W.K. Zahra, F.A. Abd El-Salam, A.A. El-Sabbagh ad Z.A. ZAk * Departmet of Egeerg athematcs

More information

Application of Calibration Approach for Regression Coefficient Estimation under Two-stage Sampling Design

Application of Calibration Approach for Regression Coefficient Estimation under Two-stage Sampling Design Authors: Pradp Basak, Kaustav Adtya, Hukum Chadra ad U.C. Sud Applcato of Calbrato Approach for Regresso Coeffcet Estmato uder Two-stage Samplg Desg Pradp Basak, Kaustav Adtya, Hukum Chadra ad U.C. Sud

More information

MULTIDIMENSIONAL HETEROGENEOUS VARIABLE PREDICTION BASED ON EXPERTS STATEMENTS. Gennadiy Lbov, Maxim Gerasimov

MULTIDIMENSIONAL HETEROGENEOUS VARIABLE PREDICTION BASED ON EXPERTS STATEMENTS. Gennadiy Lbov, Maxim Gerasimov Iteratoal Boo Seres "Iformato Scece ad Computg" 97 MULTIIMNSIONAL HTROGNOUS VARIABL PRICTION BAS ON PRTS STATMNTS Geady Lbov Maxm Gerasmov Abstract: I the wors [ ] we proposed a approach of formg a cosesus

More information

Multiple Regression. More than 2 variables! Grade on Final. Multiple Regression 11/21/2012. Exam 2 Grades. Exam 2 Re-grades

Multiple Regression. More than 2 variables! Grade on Final. Multiple Regression 11/21/2012. Exam 2 Grades. Exam 2 Re-grades STAT 101 Dr. Kar Lock Morga 11/20/12 Exam 2 Grades Multple Regresso SECTIONS 9.2, 10.1, 10.2 Multple explaatory varables (10.1) Parttog varablty R 2, ANOVA (9.2) Codtos resdual plot (10.2) Trasformatos

More information

ESS Line Fitting

ESS Line Fitting ESS 5 014 17. Le Fttg A very commo problem data aalyss s lookg for relatoshpetwee dfferet parameters ad fttg les or surfaces to data. The smplest example s fttg a straght le ad we wll dscuss that here

More information

Block-Based Compact Thermal Modeling of Semiconductor Integrated Circuits

Block-Based Compact Thermal Modeling of Semiconductor Integrated Circuits Block-Based Compact hermal Modelg of Semcoductor Itegrated Crcuts Master s hess Defese Caddate: Jg Ba Commttee Members: Dr. Mg-Cheg Cheg Dr. Daqg Hou Dr. Robert Schllg July 27, 2009 Outle Itroducto Backgroud

More information

TESTS BASED ON MAXIMUM LIKELIHOOD

TESTS BASED ON MAXIMUM LIKELIHOOD ESE 5 Toy E. Smth. The Basc Example. TESTS BASED ON MAXIMUM LIKELIHOOD To llustrate the propertes of maxmum lkelhood estmates ad tests, we cosder the smplest possble case of estmatg the mea of the ormal

More information

CHAPTER VI Statistical Analysis of Experimental Data

CHAPTER VI Statistical Analysis of Experimental Data Chapter VI Statstcal Aalyss of Expermetal Data CHAPTER VI Statstcal Aalyss of Expermetal Data Measuremets do ot lead to a uque value. Ths s a result of the multtude of errors (maly radom errors) that ca

More information

A tighter lower bound on the circuit size of the hardest Boolean functions

A tighter lower bound on the circuit size of the hardest Boolean functions Electroc Colloquum o Computatoal Complexty, Report No. 86 2011) A tghter lower boud o the crcut sze of the hardest Boolea fuctos Masak Yamamoto Abstract I [IPL2005], Fradse ad Mlterse mproved bouds o the

More information

An Indian Journal FULL PAPER ABSTRACT KEYWORDS. Trade Science Inc. Research on scheme evaluation method of automation mechatronic systems

An Indian Journal FULL PAPER ABSTRACT KEYWORDS. Trade Science Inc. Research on scheme evaluation method of automation mechatronic systems [ype text] [ype text] [ype text] ISSN : 0974-7435 Volume 0 Issue 6 Boechology 204 Ida Joural FULL PPER BIJ, 0(6, 204 [927-9275] Research o scheme evaluato method of automato mechatroc systems BSRC Che

More information

Reliability evaluation of distribution network based on improved non. sequential Monte Carlo method

Reliability evaluation of distribution network based on improved non. sequential Monte Carlo method 3rd Iteratoal Coferece o Mecatrocs, Robotcs ad Automato (ICMRA 205) Relablty evaluato of dstrbuto etwork based o mproved o sequetal Mote Carlo metod Je Zu, a, Cao L, b, Aog Tag, c Scool of Automato, Wua

More information

Objectives of Multiple Regression

Objectives of Multiple Regression Obectves of Multple Regresso Establsh the lear equato that best predcts values of a depedet varable Y usg more tha oe eplaator varable from a large set of potetal predctors {,,... k }. Fd that subset of

More information

An Improved Differential Evolution Algorithm Based on Statistical Log-linear Model

An Improved Differential Evolution Algorithm Based on Statistical Log-linear Model Sesors & Trasducers, Vol. 59, Issue, November, pp. 77-8 Sesors & Trasducers by IFSA http://www.sesorsportal.com A Improved Dfferetal Evoluto Algorthm Based o Statstcal Log-lear Model Zhehuag Huag School

More information

Lecture 8: Linear Regression

Lecture 8: Linear Regression Lecture 8: Lear egresso May 4, GENOME 56, Sprg Goals Develop basc cocepts of lear regresso from a probablstc framework Estmatg parameters ad hypothess testg wth lear models Lear regresso Su I Lee, CSE

More information

Investigating Cellular Automata

Investigating Cellular Automata Researcher: Taylor Dupuy Advsor: Aaro Wootto Semester: Fall 4 Ivestgatg Cellular Automata A Overvew of Cellular Automata: Cellular Automata are smple computer programs that geerate rows of black ad whte

More information

best estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best

best estimate (mean) for X uncertainty or error in the measurement (systematic, random or statistical) best Error Aalyss Preamble Wheever a measuremet s made, the result followg from that measuremet s always subject to ucertaty The ucertaty ca be reduced by makg several measuremets of the same quatty or by mprovg

More information

An Introduction to. Support Vector Machine

An Introduction to. Support Vector Machine A Itroducto to Support Vector Mache Support Vector Mache (SVM) A classfer derved from statstcal learg theory by Vapk, et al. 99 SVM became famous whe, usg mages as put, t gave accuracy comparable to eural-etwork

More information

Newton s Power Flow algorithm

Newton s Power Flow algorithm Power Egeerg - Egll Beedt Hresso ewto s Power Flow algorthm Power Egeerg - Egll Beedt Hresso The ewto s Method of Power Flow 2 Calculatos. For the referece bus #, we set : V = p.u. ad δ = 0 For all other

More information

MEASURES OF DISPERSION

MEASURES OF DISPERSION MEASURES OF DISPERSION Measure of Cetral Tedecy: Measures of Cetral Tedecy ad Dsperso ) Mathematcal Average: a) Arthmetc mea (A.M.) b) Geometrc mea (G.M.) c) Harmoc mea (H.M.) ) Averages of Posto: a) Meda

More information

Chapter 5 Properties of a Random Sample

Chapter 5 Properties of a Random Sample Lecture 6 o BST 63: Statstcal Theory I Ku Zhag, /0/008 Revew for the prevous lecture Cocepts: t-dstrbuto, F-dstrbuto Theorems: Dstrbutos of sample mea ad sample varace, relatoshp betwee sample mea ad sample

More information

Generating Multivariate Nonnormal Distribution Random Numbers Based on Copula Function

Generating Multivariate Nonnormal Distribution Random Numbers Based on Copula Function 7659, Eglad, UK Joural of Iformato ad Computg Scece Vol. 2, No. 3, 2007, pp. 9-96 Geeratg Multvarate Noormal Dstrbuto Radom Numbers Based o Copula Fucto Xaopg Hu +, Jam He ad Hogsheg Ly School of Ecoomcs

More information

STRONG CONSISTENCY FOR SIMPLE LINEAR EV MODEL WITH v/ -MIXING

STRONG CONSISTENCY FOR SIMPLE LINEAR EV MODEL WITH v/ -MIXING Joural of tatstcs: Advaces Theory ad Alcatos Volume 5, Number, 6, Pages 3- Avalable at htt://scetfcadvaces.co. DOI: htt://d.do.org/.864/jsata_7678 TRONG CONITENCY FOR IMPLE LINEAR EV MODEL WITH v/ -MIXING

More information

Dimensionality Reduction and Learning

Dimensionality Reduction and Learning CMSC 35900 (Sprg 009) Large Scale Learg Lecture: 3 Dmesoalty Reducto ad Learg Istructors: Sham Kakade ad Greg Shakharovch L Supervsed Methods ad Dmesoalty Reducto The theme of these two lectures s that

More information

Bayesian Classification. CS690L Data Mining: Classification(2) Bayesian Theorem: Basics. Bayesian Theorem. Training dataset. Naïve Bayes Classifier

Bayesian Classification. CS690L Data Mining: Classification(2) Bayesian Theorem: Basics. Bayesian Theorem. Training dataset. Naïve Bayes Classifier Baa Classfcato CS6L Data Mg: Classfcato() Referece: J. Ha ad M. Kamber, Data Mg: Cocepts ad Techques robablstc learg: Calculate explct probabltes for hypothess, amog the most practcal approaches to certa

More information

Taylor s Series and Interpolation. Interpolation & Curve-fitting. CIS Interpolation. Basic Scenario. Taylor Series interpolates at a specific

Taylor s Series and Interpolation. Interpolation & Curve-fitting. CIS Interpolation. Basic Scenario. Taylor Series interpolates at a specific CIS 54 - Iterpolato Roger Crawfs Basc Scearo We are able to prod some fucto, but do ot kow what t really s. Ths gves us a lst of data pots: [x,f ] f(x) f f + x x + August 2, 25 OSU/CIS 54 3 Taylor s Seres

More information

Lecture 16: Backpropogation Algorithm Neural Networks with smooth activation functions

Lecture 16: Backpropogation Algorithm Neural Networks with smooth activation functions CO-511: Learg Theory prg 2017 Lecturer: Ro Lv Lecture 16: Bacpropogato Algorthm Dsclamer: These otes have ot bee subected to the usual scruty reserved for formal publcatos. They may be dstrbuted outsde

More information

ANALYSIS ON THE NATURE OF THE BASIC EQUATIONS IN SYNERGETIC INTER-REPRESENTATION NETWORK

ANALYSIS ON THE NATURE OF THE BASIC EQUATIONS IN SYNERGETIC INTER-REPRESENTATION NETWORK Far East Joural of Appled Mathematcs Volume, Number, 2008, Pages Ths paper s avalable ole at http://www.pphm.com 2008 Pushpa Publshg House ANALYSIS ON THE NATURE OF THE ASI EQUATIONS IN SYNERGETI INTER-REPRESENTATION

More information

Correlation and Regression Analysis

Correlation and Regression Analysis Chapter V Correlato ad Regresso Aalss R. 5.. So far we have cosdered ol uvarate dstrbutos. Ma a tme, however, we come across problems whch volve two or more varables. Ths wll be the subject matter of the

More information

STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. x, where. = y - ˆ " 1

STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS. x, where. = y - ˆ  1 STATISTICAL PROPERTIES OF LEAST SQUARES ESTIMATORS Recall Assumpto E(Y x) η 0 + η x (lear codtoal mea fucto) Data (x, y ), (x 2, y 2 ),, (x, y ) Least squares estmator ˆ E (Y x) ˆ " 0 + ˆ " x, where ˆ

More information

13. Parametric and Non-Parametric Uncertainties, Radial Basis Functions and Neural Network Approximations

13. Parametric and Non-Parametric Uncertainties, Radial Basis Functions and Neural Network Approximations Lecture 7 3. Parametrc ad No-Parametrc Ucertates, Radal Bass Fuctos ad Neural Network Approxmatos he parameter estmato algorthms descrbed prevous sectos were based o the assumpto that the system ucertates

More information

Principal Components. Analysis. Basic Intuition. A Method of Self Organized Learning

Principal Components. Analysis. Basic Intuition. A Method of Self Organized Learning Prcpal Compoets Aalss A Method of Self Orgazed Learg Prcpal Compoets Aalss Stadard techque for data reducto statstcal patter matchg ad sgal processg Usupervsed learg: lear from examples wthout a teacher

More information

CHAPTER 2. = y ˆ β x (.1022) So we can write

CHAPTER 2. = y ˆ β x (.1022) So we can write CHAPTER SOLUTIONS TO PROBLEMS. () Let y = GPA, x = ACT, ad = 8. The x = 5.875, y = 3.5, (x x )(y y ) = 5.85, ad (x x ) = 56.875. From equato (.9), we obta the slope as ˆβ = = 5.85/56.875., rouded to four

More information

Open Access Study on Optimization of Logistics Distribution Routes Based on Opposition-based Learning Particle Swarm Optimization Algorithm

Open Access Study on Optimization of Logistics Distribution Routes Based on Opposition-based Learning Particle Swarm Optimization Algorithm Sed Orders for Reprts to reprts@bethamscece.ae 38 The Ope Automato ad Cotrol Systems Joural, 05, 7, 38-3 Ope Access Study o Optmzato of Logstcs Dstrbuto Routes Based o Opposto-based Learg Partcle Swarm

More information

Ranking Bank Branches with Interval Data By IAHP and TOPSIS

Ranking Bank Branches with Interval Data By IAHP and TOPSIS Rag Ba Braches wth terval Data By HP ad TPSS Tayebeh Rezaetazaa Departmet of Mathematcs, slamc zad Uversty, Badar bbas Brach, Badar bbas, ra Mahaz Barhordarahmad Departmet of Mathematcs, slamc zad Uversty,

More information

Bayes (Naïve or not) Classifiers: Generative Approach

Bayes (Naïve or not) Classifiers: Generative Approach Logstc regresso Bayes (Naïve or ot) Classfers: Geeratve Approach What do we mea by Geeratve approach: Lear p(y), p(x y) ad the apply bayes rule to compute p(y x) for makg predctos Ths s essetally makg

More information

Module 7: Probability and Statistics

Module 7: Probability and Statistics Lecture 4: Goodess of ft tests. Itroducto Module 7: Probablty ad Statstcs I the prevous two lectures, the cocepts, steps ad applcatos of Hypotheses testg were dscussed. Hypotheses testg may be used to

More information

Support Vector Regression Model Based on Empirical Mode Decomposition and Auto Regression for Electric Load Forecasting

Support Vector Regression Model Based on Empirical Mode Decomposition and Auto Regression for Electric Load Forecasting Eerges 3, 6, 887-9; do:.339/e64887 Artcle OPEN ACCESS eerges ISSN 996-73 www.mdp.com/joural/eerges Support Vector Regresso Model Based o Emprcal Mode Decomposto ad Auto Regresso for Electrc Load Forecastg

More information

Dynamic Analysis of Axially Beam on Visco - Elastic Foundation with Elastic Supports under Moving Load

Dynamic Analysis of Axially Beam on Visco - Elastic Foundation with Elastic Supports under Moving Load Dyamc Aalyss of Axally Beam o Vsco - Elastc Foudato wth Elastc Supports uder Movg oad Saeed Mohammadzadeh, Seyed Al Mosayeb * Abstract: For dyamc aalyses of ralway track structures, the algorthm of soluto

More information

Nonlinear Blind Source Separation Using Hybrid Neural Networks*

Nonlinear Blind Source Separation Using Hybrid Neural Networks* Nolear Bld Source Separato Usg Hybrd Neural Networks* Chu-Hou Zheg,2, Zh-Ka Huag,2, chael R. Lyu 3, ad Tat-g Lok 4 Itellget Computg Lab, Isttute of Itellget aches, Chese Academy of Sceces, P.O.Box 3, Hefe,

More information

13. Artificial Neural Networks for Function Approximation

13. Artificial Neural Networks for Function Approximation Lecture 7 3. Artfcal eural etworks for Fucto Approxmato Motvato. A typcal cotrol desg process starts wth modelg, whch s bascally the process of costructg a mathematcal descrpto (such as a set of ODE-s)

More information

Bounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy

Bounds on the expected entropy and KL-divergence of sampled multinomial distributions. Brandon C. Roy Bouds o the expected etropy ad KL-dvergece of sampled multomal dstrbutos Brado C. Roy bcroy@meda.mt.edu Orgal: May 18, 2011 Revsed: Jue 6, 2011 Abstract Iformato theoretc quattes calculated from a sampled

More information

Unsupervised Learning and Other Neural Networks

Unsupervised Learning and Other Neural Networks CSE 53 Soft Computg NOT PART OF THE FINAL Usupervsed Learg ad Other Neural Networs Itroducto Mture Destes ad Idetfablty ML Estmates Applcato to Normal Mtures Other Neural Networs Itroducto Prevously, all

More information

A Sensitivity-Based Adaptive Architecture Pruning Algorithm for Madalines

A Sensitivity-Based Adaptive Architecture Pruning Algorithm for Madalines Advaced Scece ad echology etters, pp.84-9 http://dx.do.org/0.457/astl.06.3.35 A Sestvty-Based Adaptve Archtecture Prug Algorthm for Madales Sa J, Pg Yag, Shumg Zhog, J Wag, Jeog-Uk Km Jagsu Egeerg Ceter

More information

Analyzing Fuzzy System Reliability Using Vague Set Theory

Analyzing Fuzzy System Reliability Using Vague Set Theory Iteratoal Joural of Appled Scece ad Egeerg 2003., : 82-88 Aalyzg Fuzzy System Relablty sg Vague Set Theory Shy-Mg Che Departmet of Computer Scece ad Iformato Egeerg, Natoal Tawa versty of Scece ad Techology,

More information

A new type of optimization method based on conjugate directions

A new type of optimization method based on conjugate directions A ew type of optmzato method based o cojugate drectos Pa X Scece School aj Uversty of echology ad Educato (UE aj Cha e-mal: pax94@sacom Abstract A ew type of optmzato method based o cojugate drectos s

More information

ENGI 3423 Simple Linear Regression Page 12-01

ENGI 3423 Simple Linear Regression Page 12-01 ENGI 343 mple Lear Regresso Page - mple Lear Regresso ometmes a expermet s set up where the expermeter has cotrol over the values of oe or more varables X ad measures the resultg values of aother varable

More information

Multivariate Transformation of Variables and Maximum Likelihood Estimation

Multivariate Transformation of Variables and Maximum Likelihood Estimation Marquette Uversty Multvarate Trasformato of Varables ad Maxmum Lkelhood Estmato Dael B. Rowe, Ph.D. Assocate Professor Departmet of Mathematcs, Statstcs, ad Computer Scece Copyrght 03 by Marquette Uversty

More information

CS 2750 Machine Learning. Lecture 8. Linear regression. CS 2750 Machine Learning. Linear regression. is a linear combination of input components x

CS 2750 Machine Learning. Lecture 8. Linear regression. CS 2750 Machine Learning. Linear regression. is a linear combination of input components x CS 75 Mache Learg Lecture 8 Lear regresso Mlos Hauskrecht mlos@cs.ptt.edu 539 Seott Square CS 75 Mache Learg Lear regresso Fucto f : X Y s a lear combato of put compoets f + + + K d d K k - parameters

More information

Some Notes on the Probability Space of Statistical Surveys

Some Notes on the Probability Space of Statistical Surveys Metodološk zvezk, Vol. 7, No., 200, 7-2 ome Notes o the Probablty pace of tatstcal urveys George Petrakos Abstract Ths paper troduces a formal presetato of samplg process usg prcples ad cocepts from Probablty

More information

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity

ECONOMETRIC THEORY. MODULE VIII Lecture - 26 Heteroskedasticity ECONOMETRIC THEORY MODULE VIII Lecture - 6 Heteroskedastcty Dr. Shalabh Departmet of Mathematcs ad Statstcs Ida Isttute of Techology Kapur . Breusch Paga test Ths test ca be appled whe the replcated data

More information

Non-uniform Turán-type problems

Non-uniform Turán-type problems Joural of Combatoral Theory, Seres A 111 2005 106 110 wwwelsevercomlocatecta No-uform Turá-type problems DhruvMubay 1, Y Zhao 2 Departmet of Mathematcs, Statstcs, ad Computer Scece, Uversty of Illos at

More information

Multiple Choice Test. Chapter Adequacy of Models for Regression

Multiple Choice Test. Chapter Adequacy of Models for Regression Multple Choce Test Chapter 06.0 Adequac of Models for Regresso. For a lear regresso model to be cosdered adequate, the percetage of scaled resduals that eed to be the rage [-,] s greater tha or equal to

More information

LINEARLY CONSTRAINED MINIMIZATION BY USING NEWTON S METHOD

LINEARLY CONSTRAINED MINIMIZATION BY USING NEWTON S METHOD Jural Karya Asl Loreka Ahl Matematk Vol 8 o 205 Page 084-088 Jural Karya Asl Loreka Ahl Matematk LIEARLY COSTRAIED MIIMIZATIO BY USIG EWTO S METHOD Yosza B Dasrl, a Ismal B Moh 2 Faculty Electrocs a Computer

More information

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution:

{ }{ ( )} (, ) = ( ) ( ) ( ) Chapter 14 Exercises in Sampling Theory. Exercise 1 (Simple random sampling): Solution: Chapter 4 Exercses Samplg Theory Exercse (Smple radom samplg: Let there be two correlated radom varables X ad A sample of sze s draw from a populato by smple radom samplg wthout replacemet The observed

More information

Generalized Convex Functions on Fractal Sets and Two Related Inequalities

Generalized Convex Functions on Fractal Sets and Two Related Inequalities Geeralzed Covex Fuctos o Fractal Sets ad Two Related Iequaltes Huxa Mo, X Su ad Dogya Yu 3,,3School of Scece, Bejg Uversty of Posts ad Telecommucatos, Bejg,00876, Cha, Correspodece should be addressed

More information

Arithmetic Mean and Geometric Mean

Arithmetic Mean and Geometric Mean Acta Mathematca Ntresa Vol, No, p 43 48 ISSN 453-6083 Arthmetc Mea ad Geometrc Mea Mare Varga a * Peter Mchalča b a Departmet of Mathematcs, Faculty of Natural Sceces, Costate the Phlosopher Uversty Ntra,

More information

Beam Warming Second-Order Upwind Method

Beam Warming Second-Order Upwind Method Beam Warmg Secod-Order Upwd Method Petr Valeta Jauary 6, 015 Ths documet s a part of the assessmet work for the subject 1DRP Dfferetal Equatos o Computer lectured o FNSPE CTU Prague. Abstract Ths documet

More information

Bootstrap Method for Testing of Equality of Several Coefficients of Variation

Bootstrap Method for Testing of Equality of Several Coefficients of Variation Cloud Publcatos Iteratoal Joural of Advaced Mathematcs ad Statstcs Volume, pp. -6, Artcle ID Sc- Research Artcle Ope Access Bootstrap Method for Testg of Equalty of Several Coeffcets of Varato Dr. Navee

More information

Chapter 5. Curve fitting

Chapter 5. Curve fitting Chapter 5 Curve ttg Assgmet please use ecell Gve the data elow use least squares regresso to t a a straght le a power equato c a saturato-growthrate equato ad d a paraola. Fd the r value ad justy whch

More information

Lecture 9: Tolerant Testing

Lecture 9: Tolerant Testing Lecture 9: Tolerat Testg Dael Kae Scrbe: Sakeerth Rao Aprl 4, 07 Abstract I ths lecture we prove a quas lear lower boud o the umber of samples eeded to do tolerat testg for L dstace. Tolerat Testg We have

More information

A Multi-Entry Simulated and Inversed Function Approach. for Alternative Solutions

A Multi-Entry Simulated and Inversed Function Approach. for Alternative Solutions Iteratoal Mathematcal Forum,, 2006, o. 40, 2003 207 A Mult-Etry Smulated ad Iversed Fucto Approach for Alteratve Solutos Kev Wag a, Che Chag b ad Chug Pg Lu b a Computg ad Mathematcs School Joh Moores

More information

4. Standard Regression Model and Spatial Dependence Tests

4. Standard Regression Model and Spatial Dependence Tests 4. Stadard Regresso Model ad Spatal Depedece Tests Stadard regresso aalss fals the presece of spatal effects. I case of spatal depedeces ad/or spatal heterogeet a stadard regresso model wll be msspecfed.

More information

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions.

Ordinary Least Squares Regression. Simple Regression. Algebra and Assumptions. Ordary Least Squares egresso. Smple egresso. Algebra ad Assumptos. I ths part of the course we are gog to study a techque for aalysg the lear relatoshp betwee two varables Y ad X. We have pars of observatos

More information

Chapter 8. Inferences about More Than Two Population Central Values

Chapter 8. Inferences about More Than Two Population Central Values Chapter 8. Ifereces about More Tha Two Populato Cetral Values Case tudy: Effect of Tmg of the Treatmet of Port-We tas wth Lasers ) To vestgate whether treatmet at a youg age would yeld better results tha

More information

Transforms that are commonly used are separable

Transforms that are commonly used are separable Trasforms s Trasforms that are commoly used are separable Eamples: Two-dmesoal DFT DCT DST adamard We ca the use -D trasforms computg the D separable trasforms: Take -D trasform of the rows > rows ( )

More information

ECE 595, Section 10 Numerical Simulations Lecture 19: FEM for Electronic Transport. Prof. Peter Bermel February 22, 2013

ECE 595, Section 10 Numerical Simulations Lecture 19: FEM for Electronic Transport. Prof. Peter Bermel February 22, 2013 ECE 595, Secto 0 Numercal Smulatos Lecture 9: FEM for Electroc Trasport Prof. Peter Bermel February, 03 Outle Recap from Wedesday Physcs-based devce modelg Electroc trasport theory FEM electroc trasport

More information

OPTIMAL LAY-OUT OF NATURAL GAS PIPELINE NETWORK

OPTIMAL LAY-OUT OF NATURAL GAS PIPELINE NETWORK 23rd World Gas Coferece, Amsterdam 2006 OPTIMAL LAY-OUT OF NATURAL GAS PIPELINE NETWORK Ma author Tg-zhe, Ne CHINA ABSTRACT I cha, there are lots of gas ppele etwork eeded to be desged ad costructed owadays.

More information

On Modified Interval Symmetric Single-Step Procedure ISS2-5D for the Simultaneous Inclusion of Polynomial Zeros

On Modified Interval Symmetric Single-Step Procedure ISS2-5D for the Simultaneous Inclusion of Polynomial Zeros It. Joural of Math. Aalyss, Vol. 7, 2013, o. 20, 983-988 HIKARI Ltd, www.m-hkar.com O Modfed Iterval Symmetrc Sgle-Step Procedure ISS2-5D for the Smultaeous Icluso of Polyomal Zeros 1 Nora Jamalud, 1 Masor

More information

X ε ) = 0, or equivalently, lim

X ε ) = 0, or equivalently, lim Revew for the prevous lecture Cocepts: order statstcs Theorems: Dstrbutos of order statstcs Examples: How to get the dstrbuto of order statstcs Chapter 5 Propertes of a Radom Sample Secto 55 Covergece

More information

CSE 5526: Introduction to Neural Networks Linear Regression

CSE 5526: Introduction to Neural Networks Linear Regression CSE 556: Itroducto to Neural Netorks Lear Regresso Part II 1 Problem statemet Part II Problem statemet Part II 3 Lear regresso th oe varable Gve a set of N pars of data , appromate d by a lear fucto

More information

Research Article A New Derivation and Recursive Algorithm Based on Wronskian Matrix for Vandermonde Inverse Matrix

Research Article A New Derivation and Recursive Algorithm Based on Wronskian Matrix for Vandermonde Inverse Matrix Mathematcal Problems Egeerg Volume 05 Artcle ID 94757 7 pages http://ddoorg/055/05/94757 Research Artcle A New Dervato ad Recursve Algorthm Based o Wroska Matr for Vadermode Iverse Matr Qu Zhou Xja Zhag

More information

Simple Linear Regression

Simple Linear Regression Statstcal Methods I (EST 75) Page 139 Smple Lear Regresso Smple regresso applcatos are used to ft a model descrbg a lear relatoshp betwee two varables. The aspects of least squares regresso ad correlato

More information