Performance Evaluation of Acoustic Scene Classification Using DNN-GMM and Frame-Concatenated Acoustic Features

Size: px
Start display at page:

Download "Performance Evaluation of Acoustic Scene Classification Using DNN-GMM and Frame-Concatenated Acoustic Features"

Transcription

1 Proceeing of APSIPA Annual Summit an Conference 2017 Performance Evaluation of Acoutic Scene Claification Uing NN-GMM an Frame-Concatenate Acoutic Feature Gen Takahahi, Takehi Yamaa, Nobutaka Ono an Shoji Makino Univerity of Tukuba, Japan National Intitute of Informatic / SOKENAI, Japan g.takahahi@mmlab.c.tukuba.ac.jp Abtract e previouly propoe a metho of acoutic cene claification uing a eep neural network-gauian mixture moel (NN-GMM) an frame-concatenate acoutic feature. It wa ubmitte to the etection an Claification of Acoutic Scene an Event (CASE) 2016 Challenge an wa ranke eighth among 49 algorithm. In the propoe metho, acoutic feature in temporally itant frame were concatenate to capture their temporal relationhip. The experimental reult inicate that the claification accuracy i improve by increaing the number of concatenate frame. On the other han, the frame concatenation interval, which i the interval with which the frame ue for frame concatenation are electe, i another important parameter. In our previou metho, the frame concatenation interval wa fixe to 100 m. In thi paper, we optimize the number of concatenate frame an the frame concatenation interval for the previouly propoe metho. A a reult, it wa confirme that the claification accuracy of the metho wa improve by 2.61% in comparion with the reult ubmitte to the CASE I. INTROUCTION Attempt have been mae to automatically recognize human behavior an urrouning circumtance. Thi technology i applicable to monitoring elerly people, the auto-tagging of multimeia content an life-log collection. Acoutic event etection an acoutic cene claification have been focue on a funamental technologie ue in thee ytem. Acoutic event etection etect acoutic event incluing oun ignal an their timetamp, where acoutic event are a ingle oun emitte from one oun ource, uch a oor opening, coughing an moue clicking. On the other han, acoutic cene claification claifie the place or ituation where the acoutic oun wa recore. The length of an acoutic oun i typically on the orer of 10, an type of oun inclue thoe of bue, park an crow. In thi reearch, we focu on acoutic cene claification. The etection an Claification of Acoutic Scene an Event (CASE) 2013 [1] i a workhop on acoutic event etection an acoutic cene claification. Tak involving acoutic cene claification an acoutic event etection are prepare at the CASE In the metho propoe for acoutic cene claification [2][3], mel frequency pectral coefficient (MFCC), the mel frequency pectrum an recurrence quantification analyi (RQA) have been ue a feature an a Gauian mixture moel (GMM) an a upport vector machine (SVM) have been ue a claifier. The reulting claification accuracy wa at mot 70%. On the other han, eep neural network (NN), which have a multilayer neural network, have recently been actively invetigate. In general, a NN ten to fall into a local olution an require an unrealitic learning time. However, a pre-training metho [4] that give appropriate initial value an high-pee computation on a graphic proceing unit (GPU) have been etablihe. Becaue of thi, NN are now being applie in variou claification problem. For peech recognition, a NN-hien Markov moel (HMM), which combine a NN an an HMM, ha been propoe [4]. The probability itribution in an HMM i approximately repreente by a GMM. On the other han, it i preciely repreente by the NN in a NN-HMM. It wa reporte that the performance of peech recognition i markely improve uing a NN-HMM [4]. Previouly, we propoe a metho of acoutic cene claification uing a NN-GMM an frame-concatenate acoutic feature [5]. It wa ubmitte to the CASE 2016 Challenge [6] an wa ranke eighth among 49 algorithm. In the propoe metho, feature in temporally itant frame were concatenate to capture their temporal relationhip. The experimental reult inicate that the claification accuracy i improve by increaing the number of concatenate frame. On the other han, the frame concatenation interval, which i the interval with which the frame ue for frame concatenation are electe, i another important parameter. In our metho, the frame concatenation interval wa fixe to 100 m. In thi paper, we optimize the number of concatenate frame an the frame concatenation interval of the previouly propoe metho. II. PROPOSE METHO A. Proce flow of the propoe metho Figure 1 how the proce flow of the previouly propoe metho. e aume two input channel becaue the acoutic ata in the CASE 2016 ha two channel. Firt, we compute acoutic feature (the MFCC an it firt an econ ifference) in each time frame for both the left an right channel. MFCC are a repreentation of frequency characteritic coniering human auitory characteritic. The

2 Proceeing of APSIPA Annual Summit an Conference 2017 Fig. 1. Proce flow. Fig. 2. Example of a NN-GMM. feature ue in our metho are imilar a thoe in the baeline ytem of the CASE Next, we concatenate acoutic feature in each frame an channel, which we will ecribe in Sect. II-C. Finally, we perform acoutic cene claification by inputting the high-imenional acoutic feature obtaine in thi manner into the NN-GMM. e ecribe the NN- GMM below. B. NN-GMM Figure 2 how an example of a NN-GMM. A NN- GMM conit of multilayere neural network an GMM correponing to iniviual acoutic cene. A NN-GMM i baically the ame a a NN-HMM but a GMM (a onetate HMM without a tate tranition) i ue intea of an HMM. Acoutic cene claification i the problem of claifying an acoutic cene ŝ from a time erie of feature X. ŝ i expree a ŝ = argmax P ( k X), (1) k where k i an iniviual acoutic cene. By tranforming P ( k X) in thi equation uing Baye theorem, the following equation i obtaine: P ( k X) = P (X k) P (X) P ( k). (2) Since P (X) i inepenent of k, P (X) can be regare a a contant. Alo, if we aume that the probability of appearance of each acoutic cene ha a uniform itribution, we alo can ignore P ( k ). Therefore, Equation (1) become ŝ = argmax P (X k ). (3) k A GMM i often ue in acoutic cene claification a a moel for olving Equation (3). Let x t an t k be the feature vector an the tate at time frame t, repectively, then P (X k ) can be obtaine uing a GMM a follow: P (X k ) = t P (x t t k)p ( t k t 1 k ). (4) Thi repreent the probability that the erie of feature X i generate from the GMM of the acoutic cene k. It i generally ifficult to obtain the true itribution of the output probability P (x k ); thu, the GMM repreent it by the following Gauian mixture itribution: p(x k ) = I π i N(x µ i, Σ i ), (5) i=1 where N( ) i a normal itribution, an µ i an Σ i are the mean an variance of the normal itribution, repectively. Alo, I i the number of itribution ue for mixing an π i i the weight for each itribution. A NN-GMM repreent thi output probability uing a NN intea of a Gauian mixture itribution. To expre the output probability uing a NN, P (x t t k ) in Equation (4) i tranforme by Baye theorem to obtain p(x t t k) = P (t k x t) P ( t k ) P (x t). (6) Since P (x t ) i inepenent of k, P (x t ) can be regare a a contant. Alo, if we aume that the probability of appearance of each acoutic cene ha a uniform itribution, P ( t k ) can alo be ignore. The NN an GMM are integrate by etting the input vector of the NN to x t in Equation (6) an each noe of the output layer to P ( t k x t). e ecribe the learning metho of the NN-GMM aopte in thi paper. Firt, we perform upervie learning of each acoutic cene moel uing a conventional GMM. Next, uing thi GMM, we make training ata for NN coniting of the feature in each frame of each training ata an the tate number of the GMM. In the cae of acoutic cene claification, one acoutic oun file correpon to one acoutic cene, allowing thi tep to be implifie. Then, the initial parameter of the NN are etermine by unupervie pre-training [7]. In the pre-training, each layer i regare a a retricte Boltzmann machine (RBM) an the training i performe uing the contrative ivergence (C) metho. Finally, we perform fine-tuning, which i upervie training

3 Proceeing of APSIPA Annual Summit an Conference 2017 TABLE I OVERVIE OF THE EVELOPMENT ATASET AN THE EVALUATION ATASET. Fig. 3. Example of the time frame concatenation (n = 3 an m = 100 m). # of cene 15 # of oun ata of 1170 evelopment ataet =15 cene 78 ata # of oun ata of 390 evaluation ataet (=15 cene 26 ata) ata length 30 # of channel 2 (left an right) ampling frequency 44.1 khz quantization bit 16 bit uing training ata. In the fine-tuning, we a a oftmax layer initialize uing a ranom ee an perform backpropagation uing the tochatic graient ecent (SG) metho. C. Feature concatenation In the cae of acoutic cene claification, the relationhip between oun that are more temporally itant than thoe involve in peech recognition i coniere to be involve in the claification. Therefore, improvement in the claification accuracy can be expecte by concatenating uch feature. Furthermore, we concatenate the feature of the left an right channel in each time frame to capture the patial information. e ue the above feature a the input into the NN-GMM. Here we ecribe the concrete proce of frame concatenation. e concatenate acoutic feature in each frame with thoe in everal temporally itant frame before an after the frame for both the left an right channel. There are two important parameter in thi proce. One i the number of concatenate frame n an the other i the frame concatenation interval m. e concatenate n frame before an after each frame at an interval of m m, incluing the current frame. Figure 3 how an example of the time frame concatenation. hen n an m are et to 3 an 100 m, the three frame are concatenate a hown in Figure 3. On the other han, when m i change to 20 m, the eleven frame mut be concatenate. The imenion of the feature then increae about by four time. Thi woul have a negative effect to train the claifier. By uing n an m, we can capture the relationhip of temporally itant frame without increaing the imenion of the feature ignificantly. A. Experimental conition III. EVALUATION In thi experiment, we evaluate the effectivene of our metho uing the evelopment ataet an evaluation ataet provie by the CASE 2016 Challenge. Table I give an overview of thi ataet. The ataet contain 15 acoutic cene, for example, bu, retaurant an park. Each cene ha 78 oun ata in the evelopment ataet an 26 oun ata in the evaluation ataet, each of which i a tereo ignal with a uration of 30. The evaluation ataet ha open oun ata, which are ifferent from the oun ata of the evelopment ataet. The ampling frequency i 44.1 khz an the number of quantization bit i 16. TABLE II CONITIONS OF THE ACOUSTIC FEATURES AN THE NN-GMM. 20th-orer MFCC + + feature (60 imenion) n frame 2 channel frame length 40 m frame perio 20 m # of concatenate frame n 1, 3, 5, 7 frame concatenation 20, 100, 200, interval m (m) 500, 1000, 2000 # of hien layer 2, 3, 4, 5 imenion of hien layer 256, 512, 1024, 2048 imenion of input layer 120 n imenion of output layer 15 Table II how the conition of the acoutic feature an the NN-GMM. The feature are bae on a total of 60 imenion of 20th-orer MFCC an their firt an econ ifference. The time frame length an frame perio in the frame analyi are 40 an 20 m, repectively, which are the ame a thoe ue in the baeline ytem of the CASE 2016 Challenge. By the concatenation of n frame an two channel, the imenion of the feature become 20 3 n (frame) 2 (ch). The frame ue for frame concatenation are electe at m m interval. In thi experiment, the number of concatenate frame n i et to 1, 3, 5 or 7, an the frame concatenation interval m i et to 20, 100, 200, 500, 1000 or 2000 m. The number of hien layer of the NN (excluing the input layer an oftmax layer) i et to 2, 3, 4 or 5, an the imenion of each hien layer i contant an et to 256, 512, 1024 or e performe the pre-training proceing on the RBM of the hien layer uing the C-1 metho an on the RBM of the input layer uing the C-2 metho. e et the learning rate of the RBM to 0.4, the learning rate of the NN to an the ropout rate to 0.0 on the bai of the reult of a preliminary experiment. In the training of the claifier, we firt performe four fol cro-valiation (ata open, acoutic cene cloe) on the evelopment ataet an optimize the number of layer of the NN an the imenion of the hien

4 Proceeing of APSIPA Annual Summit an Conference 2017 Y Y Z E Y Y, Z & y Fig. 4. Reult of the CASE 2016 Challenge [9]. TABLE III CLASSIFICATION ACCURACY FOR EACH COMBINATION OF n AN m, ITH THE NUMBER OF HIEN LAYERS AN THE IMENSION OF THE HIEN LAYERS OPTIMIZE FOR THE EVELOPMENT ATASET IN PARENTHESES. 20 m (3, 1024) 100 m 200 m 500 m 1000 m 2000 m (2, 512) (3, 2048) (3, 512) (2, 512) (2, 256) (3, 1024) (3, 512) (3, 2048) (2, 256) layer, with a fixe ranom ee for initialization of the oftmax layer. Then, uing the parameter, we traine the claifier of which we ue all the acoutic ata in the evelopment ataet for the training, with each of ten ifferent ranom ee for initialization of the oftmax layer. e evaluate the effectivene of the metho uing the traine claifier an the evaluation ataet. e ue the Kali toolkit [8] to buil our ytem. B. Reult Table III how the reult of the experiment. The column are the time frame concatenation interval m an the row are the number of concatenate frame n. Each claification accuracy i the average of the claification accuracie obtaine for the ten ifferent ranom ee for initialization of the oftmax layer. The tanar eviation of the claification accuracie obtaine for the ten ifferent ranom ee wa at mot 1.1% in each combination of n an m. The number of hien layer an the imenion of the hien layer optimize for evelopment ataet are written in parenthee. e can ee that the claification accuracy of our metho varie with n an m. The highet average accuracy of 86.97% wa obtaine when n = 3 an m = 500 m, which i 2.38% higher than that when n = 1 (no frame concatenation). Thi emontrate the effect of uing the concatenate acoutic feature. In Table III, claification accuracie of more than 85.0%, written in TABLE IV CLASSIFICATION ACCURACY FOR EACH SOUN SCENE USING THE BASELINE SYSTEM, THE PREVIOUSLY PROPOSE METHO (n = 1) AN THE OPTIMIZE METHO (n = 3, m = 500 MS). cene metho Previouly propoe metho 1 Optimize metho 3, 500 reiential_area 83.46% 87.31% city_center 85.00% 86.15% beach 89.62% 90.38% park 91.92% 91.92% home 88.46% 86.54% foret_path 95.77% 95.00% bu 96.54% % grocery_tore 89.23% 97.31% café/retaurant 43.46% 58.85% car % % train 47.69% 57.31% tram % % metro_tation 88.46% 85.77% office % % library 69.23% 66.54% italic, were obtaine when n wa 5 or le an m wa 500 m or le. On the other han, the claification accuracy wa ignificantly reuce when n or m wa too high. Next, Figure 4 ummarize the reult of the CASE 2016 Challenge [9]. The orange, green an yellow bar in thi figure repreent the baeline ytem of the CASE 2016 Challenge, the propoe metho (n = 5, m = 100 m) ubmitte to the CASE 2016 Challenge (note that the feature are irreveribly compree) an the optimize metho (n = 3, m = 500 m), repectively. The claification accuracy of the optimize metho (n = 3, m = 500 m) wa improve by 9.77% compare with the baeline ytem. Thi illutrate the effect of uing both NN-GMM an concatenate acoutic feature. The bet claification accuracy of our metho (n = 3, m = 500 m) in the ten ranom ee wa 88.21% an wa improve by 2.61% compare with that of the propoe metho (n = 5 an m = 100 m), an wa ranke thir among the 49 algorithm.

5 Proceeing of APSIPA Annual Summit an Conference 2017 ata in both oun cene are imilar. Therefore, improvement of the claification accuracy for thee oun cene i require in the future. IV. CONCLUSION In thi paper, we optimize the number of concatenate frame an the frame concatenation interval for the previouly propoe metho. e carrie out an experiment uing the evelopment ataet an the evaluation ataet of the CASE 2016 Challenge. It wa foun that by etting the number of concatenate frame n to 5 or le an the frame concatenation interval m to 500 m or le, high claification accuracy wa obtaine relatively tably. Alo, the claification accuracy of oun cene having large temporal change wa improve. REFERENCES Fig. 5. Confuion matrix of the optimize metho (n = 3, m = 500 m), which i generate by counting the claification reult obtaine for all the ten ranom ee. Finally, Table IV how the claification accuracy for each oun cene. The column an row how the oun cene an the metho, repectively. Our metho (n = 3, m = 500 m) i greatly improve for grocery tore, cafe/retaurant an train compare with the previouly propoe metho (n = 1). The oun in grocery tore, cafe/retaurant an train are lou an untationary with large temporal change, incluing muic an peech. Figure 5 i the confuion matrix of the optimize metho (n = 3, m = 500 m), which i generate by counting the claification reult obtaine for all the ten ranom ee. The column an row how the target oun cene an the claifie reult, repectively. From Figure 5, many claification error for cafe/retaurant, train an library can be oberve. For example, the oun ata in cafe/retaurant are often incorrectly claifie a grocery tore, ince the oun [1]. Giannouli, E. Beneto,. Stowell, M. Roignol, M. Lagrange, M. Plumbley, etection an claification of acoutic cene an event: An IEEE AASP challenge, Proc. ASPAA 2013, Oct [2] G. Roma,. Nogueira, P. Herrera, Recurrence quantification analyi feature for auitory cene claification, Proc. ASPAA 2013, Oct [3] J. Nam Z. Hyung, K. Lee, Acoutic cene claification uing pare feature learning an elective max-pool by event etection, Proc. ASPAA 2013, Oct [4] G. E. Hinton, L. eng,. Yu, G. E. ahl, A. Mohame, N. Jaitly, A. Senior, V. Vanhoucke, P. Nguyen, T. N. Sainath, B. Kingbury, eep neural network for acoutic moeling in peech recognition: The hare view of four reearch group, IEEE Signal Proceing Magazine, Vol. 29, No. 6, pp , [5] G. Takahahi, T. Yamaa, S. Makino, N. Ono, Acoutic cene claification uing eep neural network an frame-concatenate acoutic feature, CASE 2016 Challenge Technical Report, Sep. 2016, technical report/tak1/takahahi 2016 tak1.pf. [6] [7] G. E. Hinton, A practical guie to training retricte Boltzmann machine, Momentum, Vol. 9, No. 1, [8] [9]

2.0 ANALYTICAL MODELS OF THERMAL EXCHANGES IN THE PYRANOMETER

2.0 ANALYTICAL MODELS OF THERMAL EXCHANGES IN THE PYRANOMETER 2.0 ANAYTICA MODE OF THERMA EXCHANGE IN THE PYRANOMETER In Chapter 1, it wa etablihe that a better unertaning of the thermal exchange within the intrument i neceary to efine the quantitie proucing an offet.

More information

Cross-View Action Recognition via a Transferable Dictionary Pair

Cross-View Action Recognition via a Transferable Dictionary Pair Cro- Action Recognition via a Tranferable Dictionary Pair 1 Jingjing heng 1 1 huolin Jiang Jonathon Phillip 1 Rama Chellappa Center for Automation 1 Reearch, UMIACS, Univerity of Univerity of Marylan,

More information

Module: 8 Lecture: 1

Module: 8 Lecture: 1 Moule: 8 Lecture: 1 Energy iipate by amping Uually amping i preent in all ocillatory ytem. It effect i to remove energy from the ytem. Energy in a vibrating ytem i either iipate into heat oun or raiate

More information

Clustering Methods without Given Number of Clusters

Clustering Methods without Given Number of Clusters Clutering Method without Given Number of Cluter Peng Xu, Fei Liu Introduction A we now, mean method i a very effective algorithm of clutering. It mot powerful feature i the calability and implicity. However,

More information

Saliency Modeling in Radial Flux Permanent Magnet Synchronous Machines

Saliency Modeling in Radial Flux Permanent Magnet Synchronous Machines NORPIE 4, Tronheim, Norway Saliency Moeling in Raial Flux Permanent Magnet Synchronou Machine Abtract Senorle control of Permanent Magnet Synchronou Machine i popular for everal reaon: cot aving an ytem

More information

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog

Chapter 2 Sampling and Quantization. In order to investigate sampling and quantization, the difference between analog Chapter Sampling and Quantization.1 Analog and Digital Signal In order to invetigate ampling and quantization, the difference between analog and digital ignal mut be undertood. Analog ignal conit of continuou

More information

DEEP NEURAL SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION. Shi-Xiong Zhang, Chaojun Liu, Kaisheng Yao and Yifan Gong

DEEP NEURAL SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION. Shi-Xiong Zhang, Chaojun Liu, Kaisheng Yao and Yifan Gong DEEP NEURAL SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION Shi-Xiong Zhang, Chaojun Liu, Kaiheng Yao and Yifan Gong Microoft Corporation, One Microoft Way, Redmond, WA 98052 zhahi, chaojunl, kaiheny, ygong@microoft.com

More information

User Selection Based Cooperative Spectrum Sensing for Cognitive Radio

User Selection Based Cooperative Spectrum Sensing for Cognitive Radio Uer Selection Bae Cooperative Spectrum Sening for Cognitive Raio Netork 1 Yang Ou, Yi-ming Wang 1 Soocho Univerity, Department of Electronic an Information Engineering, Suzhou, China *1 Univerity of Science

More information

Noise Reduction in DEXA Image based on System Noise Modeling

Noise Reduction in DEXA Image based on System Noise Modeling oie Reuction in DEXA Image bae on ytem oie Moeling J. W. Kwon 1,. I. Cho 1, Y. B. Ahn, Y. M. Ro 1*, 1 Image an Vieo ytem Lab., KAIT Dept o Electronic Engineering, Konkuk Univerity Image an Vieo ytem Lab.,

More information

Factor Analysis with Poisson Output

Factor Analysis with Poisson Output Factor Analyi with Poion Output Gopal Santhanam Byron Yu Krihna V. Shenoy, Department of Electrical Engineering, Neurocience Program Stanford Univerity Stanford, CA 94305, USA {gopal,byronyu,henoy}@tanford.edu

More information

Source slideplayer.com/fundamentals of Analytical Chemistry, F.J. Holler, S.R.Crouch. Chapter 6: Random Errors in Chemical Analysis

Source slideplayer.com/fundamentals of Analytical Chemistry, F.J. Holler, S.R.Crouch. Chapter 6: Random Errors in Chemical Analysis Source lideplayer.com/fundamental of Analytical Chemitry, F.J. Holler, S.R.Crouch Chapter 6: Random Error in Chemical Analyi Random error are preent in every meaurement no matter how careful the experimenter.

More information

Learning Multiplicative Interactions

Learning Multiplicative Interactions CSC2535 2011 Lecture 6a Learning Multiplicative Interaction Geoffrey Hinton Two different meaning of multiplicative If we take two denity model and multiply together their probability ditribution at each

More information

ERTH403/HYD503, NM Tech Fall 2006

ERTH403/HYD503, NM Tech Fall 2006 ERTH43/HYD53, NM Tech Fall 6 Variation from normal rawown hyrograph Unconfine aquifer figure from Krueman an e Rier (99) Variation from normal rawown hyrograph Unconfine aquifer Early time: when pumping

More information

Actuarial Models 1: solutions example sheet 4

Actuarial Models 1: solutions example sheet 4 Actuarial Moel 1: olution example heet 4 (a) Anwer to Exercie 4.1 Q ( e u ) e σ σ u η η. (b) The forwar equation correponing to the backwar tate e are t p ee(t) σp ee (t) + ηp eu (t) t p eu(t) σp ee (t)

More information

Optimal scheduling in call centers with a callback option

Optimal scheduling in call centers with a callback option Optimal cheuling in call center with a callback option Benjamin Legro, Ouali Jouini, Ger Koole To cite thi verion: Benjamin Legro, Ouali Jouini, Ger Koole. Optimal cheuling in call center with a callback

More information

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281 72 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 28 and i 2 Show how Euler formula (page 33) can then be ued to deduce the reult a ( a) 2 b 2 {e at co bt} {e at in bt} b ( a) 2 b 2 5 Under what condition

More information

FOURIER-BASED METHODS FOR THE SPECTRAL ANALYSIS OF MUSICAL SOUNDS. Sylvain Marchand

FOURIER-BASED METHODS FOR THE SPECTRAL ANALYSIS OF MUSICAL SOUNDS. Sylvain Marchand FOUIE-BAS METHOS FO THE SPECTAL ANALYSIS OF MUSICAL SOUNS Sylvain Marchand Univerity of Bret, Lab-STICC CNS UM 628, 292 Bret, Brittany, France ABSTACT When dealing with muical ound, the hort-time Fourier

More information

Compensation of backlash effects in an Electrical Actuator

Compensation of backlash effects in an Electrical Actuator 1 Compenation of backlah effect in an Electrical Actuator R. Merzouki, J. C. Caiou an N. M Siri LaboratoireeRobotiqueeVeraille 10-12, avenue e l Europe 78140 Vélizy e-mail: merzouki@robot.uvq.fr Abtract

More information

Chapter 3 : Transfer Functions Block Diagrams Signal Flow Graphs

Chapter 3 : Transfer Functions Block Diagrams Signal Flow Graphs Chapter 3 : Tranfer Function Block Diagram Signal Flow Graph 3.. Tranfer Function 3.. Block Diagram of Control Sytem 3.3. Signal Flow Graph 3.4. Maon Gain Formula 3.5. Example 3.6. Block Diagram to Signal

More information

POLITECNICO DI TORINO Repository ISTITUZIONALE

POLITECNICO DI TORINO Repository ISTITUZIONALE POLITECNICO DI TORINO Repoitory ISTITUZIONALE GNSS ignal acquiition in the preence of ign tranition Original GNSS ignal acquiition in the preence of ign tranition / KEWEN SUN; LO PRESTI L.; FANTINO M..

More information

Face Recognition using Discriminatively Trained Orthogonal Rank One Tensor Projections

Face Recognition using Discriminatively Trained Orthogonal Rank One Tensor Projections Face Recognition uing Dicriminatively Traine Orthogonal Rank One Tenor Proection Gang Hua, Paul A. Viola, Steven M. Drucker Microoft Live Lab Reearch One Microoft Way, Remon, WA 98052 {ganghua, viloa,

More information

The Hassenpflug Matrix Tensor Notation

The Hassenpflug Matrix Tensor Notation The Haenpflug Matrix Tenor Notation D.N.J. El Dept of Mech Mechatron Eng Univ of Stellenboch, South Africa e-mail: dnjel@un.ac.za 2009/09/01 Abtract Thi i a ample document to illutrate the typeetting of

More information

Microblog Hot Spot Mining Based on PAM Probabilistic Topic Model

Microblog Hot Spot Mining Based on PAM Probabilistic Topic Model MATEC Web of Conference 22, 01062 ( 2015) DOI: 10.1051/ matecconf/ 2015220106 2 C Owned by the author, publihed by EDP Science, 2015 Microblog Hot Spot Mining Baed on PAM Probabilitic Topic Model Yaxin

More information

PHASE-FIELD SIMULATION OF SOLIDIFICATION WITH DENSITY CHANGE

PHASE-FIELD SIMULATION OF SOLIDIFICATION WITH DENSITY CHANGE Proceeing of IMECE04 004 ASME International Mechanical Engineering Congre an Epoition November 3-0, 004, Anaheim, California USA IMECE004-60875 PHASE-FIELD SIMULATION OF SOLIDIFICATION WITH DENSITY CHANGE

More information

THE Z-TRANSFORM APPLIED TO BIRTH-DEATH MARKOV PROCESSES

THE Z-TRANSFORM APPLIED TO BIRTH-DEATH MARKOV PROCESSES he Zranform applied to birthdeath arov procee HE ZRASOR ALIED O BIRHDEAH ARKO ROCESSES H Department of Electronic and Communication niverity of DareSalaam O Box 3594 DareSalaam anania ABSRAC Birthdeath

More information

Conditional Random Fields For Speech and Language Processing

Conditional Random Fields For Speech and Language Processing Outline For Speech and Language Proceing Background Maximum Entropy model and CRF CRF Example SLaTe experiment with CRF Jeremy Morri 0/7/008 Background (CRF) Dicriminative probabilitic equence model Ued

More information

ON ITERATIVE FEEDBACK TUNING AND DISTURBANCE REJECTION USING SIMPLE NOISE MODELS. Bo Wahlberg

ON ITERATIVE FEEDBACK TUNING AND DISTURBANCE REJECTION USING SIMPLE NOISE MODELS. Bo Wahlberg ON ITERATIVE FEEDBACK TUNING AND DISTURBANCE REJECTION USING SIMPLE NOISE MODELS Bo Wahlberg S3 Automatic Control, KTH, SE 100 44 Stockholm, Sween. Email: bo.wahlberg@3.kth.e Abtract: The objective of

More information

RECURRENT SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION. Shi-Xiong Zhang, Rui Zhao, Chaojun Liu, Jinyu Li and Yifan Gong

RECURRENT SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION. Shi-Xiong Zhang, Rui Zhao, Chaojun Liu, Jinyu Li and Yifan Gong RECURRENT SUPPORT VECTOR MACHINES FOR SPEECH RECOGNITION Shi-Xiong Zhang, Rui Zhao, Chaojun Liu, Jinyu Li and Yifan Gong Microoft Corporation, One Microoft Way, Redmond, WA 98052 zhahi, ruzhao, chaojunl,

More information

Matching Feature Distributions for Robust Speaker Verification

Matching Feature Distributions for Robust Speaker Verification Matching Feature Ditribution for Robut Speaker Verification Marhalleno Skoan, Daniel Mahao Department of Electrical Engineering, Univerity of Cape Town Rondeboch, Cape Town, South Africa mkoan@crg.ee.uct.ac.za

More information

Fault Prediction in Electrical Valves Using Temporal Kohonen Maps

Fault Prediction in Electrical Valves Using Temporal Kohonen Maps Fault Preiction in Electrical Valve Uing Temporal Kohonen Map Luiz F. Gonçalve Euaro L. Schneier Renato Ventura B. Henrique Marcelo Lubazewki Univeriae Feeral o Rio Grane o Sul Departamento e Engenharia

More information

Lecture 4 Topic 3: General linear models (GLMs), the fundamentals of the analysis of variance (ANOVA), and completely randomized designs (CRDs)

Lecture 4 Topic 3: General linear models (GLMs), the fundamentals of the analysis of variance (ANOVA), and completely randomized designs (CRDs) Lecture 4 Topic 3: General linear model (GLM), the fundamental of the analyi of variance (ANOVA), and completely randomized deign (CRD) The general linear model One population: An obervation i explained

More information

Massachusetts Institute of Technology Dynamics and Control II

Massachusetts Institute of Technology Dynamics and Control II I E Maachuett Intitute of Technology Department of Mechanical Engineering 2.004 Dynamic and Control II Laboratory Seion 5: Elimination of Steady-State Error Uing Integral Control Action 1 Laboratory Objective:

More information

LOAD FREQUENCY CONTROL OF MULTI AREA INTERCONNECTED SYSTEM WITH TCPS AND DIVERSE SOURCES OF POWER GENERATION

LOAD FREQUENCY CONTROL OF MULTI AREA INTERCONNECTED SYSTEM WITH TCPS AND DIVERSE SOURCES OF POWER GENERATION G.J. E.D.T.,Vol.(6:93 (NovemberDecember, 03 ISSN: 39 793 LOAD FREQUENCY CONTROL OF MULTI AREA INTERCONNECTED SYSTEM WITH TCPS AND DIVERSE SOURCES OF POWER GENERATION C.Srinivaa Rao Dept. of EEE, G.Pullaiah

More information

THE EFFECT OF WIDE STIRRUP SPACING ON DIAGONAL COMPRESSIVE CAPACITY OF HIGH STRENGTH CONCRETE BEAMS

THE EFFECT OF WIDE STIRRUP SPACING ON DIAGONAL COMPRESSIVE CAPACITY OF HIGH STRENGTH CONCRETE BEAMS - Technical Paper - THE EFFECT OF WIDE STIRRUP SPACING ON DIAGONAL COMPRESSIVE CAPACITY OF HIGH STRENGTH CONCRETE BEAMS Patarapol TANTIPIDOK *1, Koji MATSUMOTO *2 an Junichiro NIWA *3 ABSTRACT To promote

More information

What lies between Δx E, which represents the steam valve, and ΔP M, which is the mechanical power into the synchronous machine?

What lies between Δx E, which represents the steam valve, and ΔP M, which is the mechanical power into the synchronous machine? A 2.0 Introduction In the lat et of note, we developed a model of the peed governing mechanim, which i given below: xˆ K ( Pˆ ˆ) E () In thee note, we want to extend thi model o that it relate the actual

More information

LTV System Modelling

LTV System Modelling Helinki Univerit of Technolog S-72.333 Potgraduate Coure in Radiocommunication Fall 2000 LTV Stem Modelling Heikki Lorentz Sonera Entrum O heikki.lorentz@onera.fi Januar 23 rd 200 Content. Introduction

More information

University Courses on Svalbard. AT-204 Thermo-Mechanical Properties of Materials, 3 vt, 9 ECTS EXAMINATION SUGGESTED SOLUTION (PROBLEM SETS 2 AND 3)

University Courses on Svalbard. AT-204 Thermo-Mechanical Properties of Materials, 3 vt, 9 ECTS EXAMINATION SUGGESTED SOLUTION (PROBLEM SETS 2 AND 3) Page 1 of 7 Univerity Coure on Svalbar AT-204 Thermo-Mechanical Propertie of Material, 3 vt, 9 ECTS EXAMINATION SUGGESTED SOLUTION (PROBLEM SETS 2 AND 3) May 29, 2001, hour: 09.00-13.00 Reponible: Sveinung

More information

Universality of Modified Pattern Decomposition Method for Satellite Sensors

Universality of Modified Pattern Decomposition Method for Satellite Sensors Unierality of Moifie Pattern Decompoition Metho for Satellite Senor Lifu Zhang *, Yauko Mituhita **, Shinobu Furumi **, Kanako Muramatu **, Noboru Fujiara **, Motomaa Daigo *** an Liangpei Zhang * * National

More information

A Study on Simulating Convolutional Codes and Turbo Codes

A Study on Simulating Convolutional Codes and Turbo Codes A Study on Simulating Convolutional Code and Turbo Code Final Report By Daniel Chang July 27, 2001 Advior: Dr. P. Kinman Executive Summary Thi project include the deign of imulation of everal convolutional

More information

Control Systems Engineering ( Chapter 7. Steady-State Errors ) Prof. Kwang-Chun Ho Tel: Fax:

Control Systems Engineering ( Chapter 7. Steady-State Errors ) Prof. Kwang-Chun Ho Tel: Fax: Control Sytem Engineering ( Chapter 7. Steady-State Error Prof. Kwang-Chun Ho kwangho@hanung.ac.kr Tel: 0-760-453 Fax:0-760-4435 Introduction In thi leon, you will learn the following : How to find the

More information

Introduction to Mechanism Design

Introduction to Mechanism Design 5 1 Introuction to Mechanim Deign 1.1 Dominant trategie an Nah equilibria In the previou lecture we have een example of game that amit everal Nah equilibria. Moreover, ome of thee equilibria correpon to

More information

MODELING OF NEGATIVE INFLUENCES AT THE SIGNAL TRANSMISSION IN THE OPTICAL MEDIUM. Rastislav Róka, Filip Čertík

MODELING OF NEGATIVE INFLUENCES AT THE SIGNAL TRANSMISSION IN THE OPTICAL MEDIUM. Rastislav Róka, Filip Čertík MODELING OF NEGATIVE INFLUENCES AT THE SIGNAL TRANSMISSION IN THE OPTICAL MEDIUM Ratilav Róka, Filip Čertík Intitute of Telecommunication, FEEIT, Slovak Univerity of Technology in Bratilava E-mail: ratilav.roka@tuba.k,

More information

Notes on Strategic Substitutes and Complements in Global Games

Notes on Strategic Substitutes and Complements in Global Games Note on Strategic Subtitute an Complement in Global Game Stephen Morri Cowle Founation, Yale Univerity, POBox 208281, New Haven CT 06520, U S A tephenmorri@yaleeu Hyun Song Shin Lonon School of Economic,

More information

Framework Model For Single Proton Conduction through Gramicidin

Framework Model For Single Proton Conduction through Gramicidin 2 Biophyical Journal Volume 80 January 200 2 30 Framework Moel For Single Proton Conuction through Gramiciin Mark F. Schumaker,* Régi Pomè, an Benoît Roux * Department of Pure an Applie Mathematic, Wahington

More information

DSTC Layering Protocols in Wireless Relay Networks

DSTC Layering Protocols in Wireless Relay Networks DSTC Layering Protocol in Wirele Relay etwork Pannir Selvam Elamvazhuthi, Parag Shankar Kulkarni, an Bikah Kumar Dey, Member, IEEE arxiv:8659v3 ci] 7 Mar 9 Abtract With multiple antenna at tranmitter an

More information

COPRIME ARRAYS AND SAMPLERS FOR SPACE-TIME ADAPTIVE PROCESSING

COPRIME ARRAYS AND SAMPLERS FOR SPACE-TIME ADAPTIVE PROCESSING COPRIE ARRAYS AND SAPLERS FOR SPACE-IE ADAPIVE PROCESSING Chun-Lin Liu 1 and P. P. Vaidyanathan 2 Dept. of Electrical Engineering, 136-93 California Intitute of echnology, Paadena, CA 91125, USA E-mail:

More information

Loss Less Image firmness comparision by DPCM and DPCM with LMS Algorithm

Loss Less Image firmness comparision by DPCM and DPCM with LMS Algorithm Lo Le Image firmne compariion by DPCM and DPCM with LMS Algorithm Pramod Kumar Rajput 1 and Brijendra Mihra 2 1,2 Department of Electronic & Communication, Nagaji Intitute of Technology and Management

More information

An estimation approach for autotuning of event-based PI control systems

An estimation approach for autotuning of event-based PI control systems Acta de la XXXIX Jornada de Automática, Badajoz, 5-7 de Septiembre de 08 An etimation approach for autotuning of event-baed PI control ytem Joé Sánchez Moreno, María Guinaldo Loada, Sebatián Dormido Departamento

More information

Real Sources (Secondary Sources) Phantom Source (Primary source) LS P. h rl. h rr. h ll. h lr. h pl. h pr

Real Sources (Secondary Sources) Phantom Source (Primary source) LS P. h rl. h rr. h ll. h lr. h pl. h pr Ecient frequency domain ltered-x realization of phantom ource iet C.W. ommen, Ronald M. Aart, Alexander W.M. Mathijen, John Gara, Haiyan He Abtract A phantom ound ource i a virtual ound image which can

More information

Gain and Phase Margins Based Delay Dependent Stability Analysis of Two- Area LFC System with Communication Delays

Gain and Phase Margins Based Delay Dependent Stability Analysis of Two- Area LFC System with Communication Delays Gain and Phae Margin Baed Delay Dependent Stability Analyi of Two- Area LFC Sytem with Communication Delay Şahin Sönmez and Saffet Ayaun Department of Electrical Engineering, Niğde Ömer Halidemir Univerity,

More information

DYNAMIC MODELS FOR CONTROLLER DESIGN

DYNAMIC MODELS FOR CONTROLLER DESIGN DYNAMIC MODELS FOR CONTROLLER DESIGN M.T. Tham (996,999) Dept. of Chemical and Proce Engineering Newcatle upon Tyne, NE 7RU, UK.. INTRODUCTION The problem of deigning a good control ytem i baically that

More information

Analysis of Step Response, Impulse and Ramp Response in the Continuous Stirred Tank Reactor System

Analysis of Step Response, Impulse and Ramp Response in the Continuous Stirred Tank Reactor System ISSN: 454-50 Volume 0 - Iue 05 May 07 PP. 7-78 Analyi of Step Repone, Impule and Ramp Repone in the ontinuou Stirred Tank Reactor Sytem * Zohreh Khohraftar, Pirouz Derakhhi, (Department of hemitry, Science

More information

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall Beyond Significance Teting ( nd Edition), Rex B. Kline Suggeted Anwer To Exercie Chapter. The tatitic meaure variability among core at the cae level. In a normal ditribution, about 68% of the core fall

More information

BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL

BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL J. Autral. Math. Soc. Ser. B 4(999), 332 349 BUBBLES RISING IN AN INCLINED TWO-DIMENSIONAL TUBE AND JETS FALLING ALONG A WALL J. LEE and J.-M. VANDEN-BROECK 2 (Received 22 April 995; revied 23 April 996)

More information

SMALL-SIGNAL STABILITY ASSESSMENT OF THE EUROPEAN POWER SYSTEM BASED ON ADVANCED NEURAL NETWORK METHOD

SMALL-SIGNAL STABILITY ASSESSMENT OF THE EUROPEAN POWER SYSTEM BASED ON ADVANCED NEURAL NETWORK METHOD SMALL-SIGNAL STABILITY ASSESSMENT OF THE EUROPEAN POWER SYSTEM BASED ON ADVANCED NEURAL NETWORK METHOD S.P. Teeuwen, I. Erlich U. Bachmann Univerity of Duiburg, Germany Department of Electrical Power Sytem

More information

Some Sets of GCF ϵ Expansions Whose Parameter ϵ Fetch the Marginal Value

Some Sets of GCF ϵ Expansions Whose Parameter ϵ Fetch the Marginal Value Journal of Mathematical Reearch with Application May, 205, Vol 35, No 3, pp 256 262 DOI:03770/jin:2095-26520503002 Http://jmredluteducn Some Set of GCF ϵ Expanion Whoe Parameter ϵ Fetch the Marginal Value

More information

Social Studies 201 Notes for March 18, 2005

Social Studies 201 Notes for March 18, 2005 1 Social Studie 201 Note for March 18, 2005 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

Estimating floor acceleration in nonlinear multi-story moment-resisting frames

Estimating floor acceleration in nonlinear multi-story moment-resisting frames Etimating floor acceleration in nonlinear multi-tory moment-reiting frame R. Karami Mohammadi Aitant Profeor, Civil Engineering Department, K.N.Tooi Univerity M. Mohammadi M.Sc. Student, Civil Engineering

More information

TRANSITION PROBABILITY MATRIX OF BRIDGE MEMBERS DAMAGE RATING

TRANSITION PROBABILITY MATRIX OF BRIDGE MEMBERS DAMAGE RATING TRANSITION PROBABILITY MATRIX OF BRIDGE MEMBERS DAMAGE RATING Hirohi Sato and Ryoji Hagiwara 2 Abtract Bridge member damage characteritic were tudied uing the inpection record. Damage can be claified into

More information

Jul 4, 2005 turbo_code_primer Revision 0.0. Turbo Code Primer

Jul 4, 2005 turbo_code_primer Revision 0.0. Turbo Code Primer Jul 4, 5 turbo_code_primer Reviion. Turbo Code Primer. Introduction Thi document give a quick tutorial on MAP baed turbo coder. Section develop the background theory. Section work through a imple numerical

More information

Evolutionary Algorithms Based Fixed Order Robust Controller Design and Robustness Performance Analysis

Evolutionary Algorithms Based Fixed Order Robust Controller Design and Robustness Performance Analysis Proceeding of 01 4th International Conference on Machine Learning and Computing IPCSIT vol. 5 (01) (01) IACSIT Pre, Singapore Evolutionary Algorithm Baed Fixed Order Robut Controller Deign and Robutne

More information

Sampling and the Discrete Fourier Transform

Sampling and the Discrete Fourier Transform Sampling and the Dicrete Fourier Tranform Sampling Method Sampling i mot commonly done with two device, the ample-and-hold (S/H) and the analog-to-digital-converter (ADC) The S/H acquire a CT ignal at

More information

Bogoliubov Transformation in Classical Mechanics

Bogoliubov Transformation in Classical Mechanics Bogoliubov Tranformation in Claical Mechanic Canonical Tranformation Suppoe we have a et of complex canonical variable, {a j }, and would like to conider another et of variable, {b }, b b ({a j }). How

More information

A Bayesian Graphical Model to Discover Latent Events from Twitter

A Bayesian Graphical Model to Discover Latent Events from Twitter A Bayeian Graphical Moel to Dicover Latent Event from Twitter Wei Wei Carnegie Mellon Univerity 5000 Forbe Ave Pittburgh, PA, USA weiwei@c.cmu.eu Kenneth Joeph Carnegie Mellon Univerity 5000 Forbe Ave

More information

Neural Machine Translation with Source Dependency Representation

Neural Machine Translation with Source Dependency Representation Neural Machine Tranlation with Source Depenency Repreentation Kehai Chen 1 Rui Wang 2 Maao Utiyama 2 Lemao Liu 3 Akihiro Tamura 4 Eiichiro Sumita 2 an Tieun Zhao 1 1 Harbin Intitute of Technology China

More information

REDUCED-ORDER models of synchronous machines

REDUCED-ORDER models of synchronous machines A Library of Secon-Orer Moel for Synchronou Machine Olaoluwapo Ajala Stuent Member IEEE Alejanro Domínguez-García Member IEEE Peter Sauer Life Fellow IEEE an Daniel Liberzon Fellow IEEE Abtract Thi paper

More information

Digital Control System

Digital Control System Digital Control Sytem - A D D A Micro ADC DAC Proceor Correction Element Proce Clock Meaurement A: Analog D: Digital Continuou Controller and Digital Control Rt - c Plant yt Continuou Controller Digital

More information

Confusion matrices. True / False positives / negatives. INF 4300 Classification III Anne Solberg The agenda today: E.g., testing for cancer

Confusion matrices. True / False positives / negatives. INF 4300 Classification III Anne Solberg The agenda today: E.g., testing for cancer INF 4300 Claification III Anne Solberg 29.10.14 The agenda today: More on etimating claifier accuracy Cure of dimenionality knn-claification K-mean clutering x i feature vector for pixel i i- The cla label

More information

SURFACE POTENTIAL BEHAVIOR IN ISFET BASED BIO-(CHEMICAL) SENSORS

SURFACE POTENTIAL BEHAVIOR IN ISFET BASED BIO-(CHEMICAL) SENSORS Armenian Journal of Phyic, 1, vol. 5, iue 4, pp. 194- SURFACE POTETIAL BEHAVIOR I ISFET BASED BIO-(CHEMICAL) SESORS A. V. SURMALYA Department of Semiconuctor Phyic & Microelectronic, Yerevan State Univerity,

More information

An inventory model with temporary price discount when lead time links to order quantity

An inventory model with temporary price discount when lead time links to order quantity 80 Journal of Scientific & Indutrial Reearch J SCI IN RES VOL 69 MARCH 00 Vol. 69 March 00 pp. 80-87 An inventory model with temporary price dicount when lead time link to order quantity Chih-Te Yang Liang-Yuh

More information

Social Studies 201 Notes for November 14, 2003

Social Studies 201 Notes for November 14, 2003 1 Social Studie 201 Note for November 14, 2003 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

EE Control Systems LECTURE 6

EE Control Systems LECTURE 6 Copyright FL Lewi 999 All right reerved EE - Control Sytem LECTURE 6 Updated: Sunday, February, 999 BLOCK DIAGRAM AND MASON'S FORMULA A linear time-invariant (LTI) ytem can be repreented in many way, including:

More information

Introduction to Laplace Transform Techniques in Circuit Analysis

Introduction to Laplace Transform Techniques in Circuit Analysis Unit 6 Introduction to Laplace Tranform Technique in Circuit Analyi In thi unit we conider the application of Laplace Tranform to circuit analyi. A relevant dicuion of the one-ided Laplace tranform i found

More information

Chapter 5 Consistency, Zero Stability, and the Dahlquist Equivalence Theorem

Chapter 5 Consistency, Zero Stability, and the Dahlquist Equivalence Theorem Chapter 5 Conitency, Zero Stability, and the Dahlquit Equivalence Theorem In Chapter 2 we dicued convergence of numerical method and gave an experimental method for finding the rate of convergence (aka,

More information

DESIGN OF CONTROLLERS FOR STABLE AND UNSTABLE SYSTEMS WITH TIME DELAY

DESIGN OF CONTROLLERS FOR STABLE AND UNSTABLE SYSTEMS WITH TIME DELAY DESIGN OF CONTROLLERS FOR STABLE AND UNSTABLE SYSTEMS WITH TIME DELAY P. Dotál, V. Bobál Department of Proce Control, Facult of Technolog, Toma Bata Univerit in Zlín Nám. T. G. Maarka 75, 76 7 Zlín, Czech

More information

Jan Purczyński, Kamila Bednarz-Okrzyńska Estimation of the shape parameter of GED distribution for a small sample size

Jan Purczyński, Kamila Bednarz-Okrzyńska Estimation of the shape parameter of GED distribution for a small sample size Jan Purczyńki, Kamila Bednarz-Okrzyńka Etimation of the hape parameter of GED ditribution for a mall ample ize Folia Oeconomica Stetinenia 4()/, 35-46 04 Folia Oeconomica Stetinenia DOI: 0.478/foli-04-003

More information

Lecture 10 Filtering: Applied Concepts

Lecture 10 Filtering: Applied Concepts Lecture Filtering: Applied Concept In the previou two lecture, you have learned about finite-impule-repone (FIR) and infinite-impule-repone (IIR) filter. In thee lecture, we introduced the concept of filtering

More information

S_LOOP: SINGLE-LOOP FEEDBACK CONTROL SYSTEM ANALYSIS

S_LOOP: SINGLE-LOOP FEEDBACK CONTROL SYSTEM ANALYSIS S_LOOP: SINGLE-LOOP FEEDBACK CONTROL SYSTEM ANALYSIS by Michelle Gretzinger, Daniel Zyngier and Thoma Marlin INTRODUCTION One of the challenge to the engineer learning proce control i relating theoretical

More information

Tuning of Extended Kalman Filter for Speed Estimation of PMSM Drive using Particle Swarm Optimization

Tuning of Extended Kalman Filter for Speed Estimation of PMSM Drive using Particle Swarm Optimization Tuning of Extene Kalman Filter for Spee Etimation of PMSM Drive uing Particle Swarm Optimization Mahewara Rao Chintala 1, Ramana Pilla 2, Ayya Rao SLV Tummala 3 1 M.Tech Scholar, Dept. of EEE, GMR Intitute

More information

Control of Delayed Integrating Processes Using Two Feedback Controllers R MS Approach

Control of Delayed Integrating Processes Using Two Feedback Controllers R MS Approach Proceeding of the 7th WSEAS International Conference on SYSTEM SCIENCE and SIMULATION in ENGINEERING (ICOSSSE '8) Control of Delayed Integrating Procee Uing Two Feedback Controller R MS Approach LIBOR

More information

Geometry and scale relationships in high. Reynolds number turbulence determined. from three-dimensional holographic. velocimetry

Geometry and scale relationships in high. Reynolds number turbulence determined. from three-dimensional holographic. velocimetry Phy. Flui, 12:941, 2 Geometry an cale relationhip in high Reynol number turbulence etermine where repreent patial filtering at a cale. τ mut be moele in term of the reolve (filtere) velocity from three-imenional

More information

Determination of the local contrast of interference fringe patterns using continuous wavelet transform

Determination of the local contrast of interference fringe patterns using continuous wavelet transform Determination of the local contrat of interference fringe pattern uing continuou wavelet tranform Jong Kwang Hyok, Kim Chol Su Intitute of Optic, Department of Phyic, Kim Il Sung Univerity, Pyongyang,

More information

Finding the location of switched capacitor banks in distribution systems based on wavelet transform

Finding the location of switched capacitor banks in distribution systems based on wavelet transform UPEC00 3t Aug - 3rd Sept 00 Finding the location of witched capacitor bank in ditribution ytem baed on wavelet tranform Bahram nohad Shahid Chamran Univerity in Ahvaz bahramnohad@yahoo.com Mehrdad keramatzadeh

More information

A FUNCTIONAL BAYESIAN METHOD FOR THE SOLUTION OF INVERSE PROBLEMS WITH SPATIO-TEMPORAL PARAMETERS AUTHORS: CORRESPONDENCE: ABSTRACT

A FUNCTIONAL BAYESIAN METHOD FOR THE SOLUTION OF INVERSE PROBLEMS WITH SPATIO-TEMPORAL PARAMETERS AUTHORS: CORRESPONDENCE: ABSTRACT A FUNCTIONAL BAYESIAN METHOD FOR THE SOLUTION OF INVERSE PROBLEMS WITH SPATIO-TEMPORAL PARAMETERS AUTHORS: Zenon Medina-Cetina International Centre for Geohazard / Norwegian Geotechnical Intitute Roger

More information

Electrical and Control Aspects of Offshore Wind Farms II (Erao II)

Electrical and Control Aspects of Offshore Wind Farms II (Erao II) ECN-C- -04-050 Electrical an Control Apect of Offhore Win Farm II (Erao II) Volume : Dynamic moel of win farm J.T.G. Pierik (ECN) J. Morren (TUD) E.J. Wiggelinkhuizen (ECN) S.W.H. e Haan (TUD) T.G. van

More information

Outage Probability of Wireless Ad Hoc Networks with Cooperative Relaying

Outage Probability of Wireless Ad Hoc Networks with Cooperative Relaying Outage Probability of Wirele Hoc Network with Cooperative elaying Mohommaali Mohammai, Himal. Suraweera, an Xiangyun Zhou eearch School of Engineering, The utralian National Univerity, utralia Singapore

More information

NUMERICAL STUDY OF CENTRAL NERVOUS SYSTEM, MUSCULO-SKELETAL SYSTEM, AND THEIR COUPLING TOWARD MOTOR DYSFUNCTION IN PARKINSON S DISEASE

NUMERICAL STUDY OF CENTRAL NERVOUS SYSTEM, MUSCULO-SKELETAL SYSTEM, AND THEIR COUPLING TOWARD MOTOR DYSFUNCTION IN PARKINSON S DISEASE Blucher Mechanical Engineering Proceeing May 14, vol. 1, num. 1 www.proceeing.blucher.com.br/evento/1wccm NUMERIAL STUDY OF ENTRAL NERVOUS SYSTEM, MUSULO-SKELETAL SYSTEM, AND THEIR OUPLING TOWARD MOTOR

More information

EE 508 Lecture 16. Filter Transformations. Lowpass to Bandpass Lowpass to Highpass Lowpass to Band-reject

EE 508 Lecture 16. Filter Transformations. Lowpass to Bandpass Lowpass to Highpass Lowpass to Band-reject EE 508 Lecture 6 Filter Tranformation Lowpa to Bandpa Lowpa to Highpa Lowpa to Band-reject Review from Lat Time Theorem: If the perimeter variation and contact reitance are neglected, the tandard deviation

More information

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get Lecture 25 Introduction to Some Matlab c2d Code in Relation to Sampled Sytem here are many way to convert a continuou time function, { h( t) ; t [0, )} into a dicrete time function { h ( k) ; k {0,,, }}

More information

Modeling of Transport and Reaction in a Catalytic Bed Using a Catalyst Particle Model.

Modeling of Transport and Reaction in a Catalytic Bed Using a Catalyst Particle Model. Excerpt from the Proceeding of the COMSOL Conference 2010 Boton Modeling of Tranport and Reaction in a Catalytic Bed Uing a Catalyt Particle Model. F. Allain *,1, A.G. Dixon 1 1 Worceter Polytechnic Intitute

More information

A BATCH-ARRIVAL QUEUE WITH MULTIPLE SERVERS AND FUZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH

A BATCH-ARRIVAL QUEUE WITH MULTIPLE SERVERS AND FUZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH Mathematical and Computational Application Vol. 11 No. pp. 181-191 006. Aociation for Scientific Reearch A BATCH-ARRIVA QEE WITH MTIPE SERVERS AND FZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH Jau-Chuan

More information

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL 98 CHAPTER DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL INTRODUCTION The deign of ytem uing tate pace model for the deign i called a modern control deign and it i

More information

An Image-encoded Mach-Zehnder Joint Transform Correlator for Polychromatic Pattern Recognition with Multi-level Quantized Reference Functions

An Image-encoded Mach-Zehnder Joint Transform Correlator for Polychromatic Pattern Recognition with Multi-level Quantized Reference Functions Proceeding of the International MultiConference of Engineer and Computer Scientit 008 Vol I IMECS 008, 19-1 March, 008, Hong Kong An Image-encoded Mach-Zehnder Joint Tranform Correlator for Polychromatic

More information

Homework #7 Solution. Solutions: ΔP L Δω. Fig. 1

Homework #7 Solution. Solutions: ΔP L Δω. Fig. 1 Homework #7 Solution Aignment:. through.6 Bergen & Vittal. M Solution: Modified Equation.6 becaue gen. peed not fed back * M (.0rad / MW ec)(00mw) rad /ec peed ( ) (60) 9.55r. p. m. 3600 ( 9.55) 3590.45r.

More information

THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER

THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER Proceeding of IMAC XXXI Conference & Expoition on Structural Dynamic February -4 Garden Grove CA USA THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER Yung-Sheng Hu Neil S Ferguon

More information

Horizontal Biaxial Loading Tests on Sliding Lead Rubber Bearing System

Horizontal Biaxial Loading Tests on Sliding Lead Rubber Bearing System Horizontal Biaxial Loading Tet on Sliding Lead Rubber Bearing Sytem M. Yamamoto, H. Hamaguchi & N. Kamohita Takenaka Reearch and Development Intitute, Japan. M. Kikuchi & K. Ihii Hokkaido Univerity, Japan.

More information

Alternate Dispersion Measures in Replicated Factorial Experiments

Alternate Dispersion Measures in Replicated Factorial Experiments Alternate Diperion Meaure in Replicated Factorial Experiment Neal A. Mackertich The Raytheon Company, Sudbury MA 02421 Jame C. Benneyan Northeatern Univerity, Boton MA 02115 Peter D. Krau The Raytheon

More information

The Dynamics of Learning Vector Quantization

The Dynamics of Learning Vector Quantization The Dynamic of Learning Vector Quantization Barbara Hammer TU Clauthal-Zellerfeld Intitute of Computing Science Michael Biehl, Anarta Ghoh Rijkuniveriteit Groningen Mathematic and Computing Science Introduction

More information

Lecture 8: Period Finding: Simon s Problem over Z N

Lecture 8: Period Finding: Simon s Problem over Z N Quantum Computation (CMU 8-859BB, Fall 205) Lecture 8: Period Finding: Simon Problem over Z October 5, 205 Lecturer: John Wright Scribe: icola Rech Problem A mentioned previouly, period finding i a rephraing

More information

Advanced Digital Signal Processing. Stationary/nonstationary signals. Time-Frequency Analysis... Some nonstationary signals. Time-Frequency Analysis

Advanced Digital Signal Processing. Stationary/nonstationary signals. Time-Frequency Analysis... Some nonstationary signals. Time-Frequency Analysis Advanced Digital ignal Proceing Prof. Nizamettin AYDIN naydin@yildiz.edu.tr Time-Frequency Analyi http://www.yildiz.edu.tr/~naydin 2 tationary/nontationary ignal Time-Frequency Analyi Fourier Tranform

More information