25-30 September 200, Abu Dhabi, UAE Output Voltage Control of a Wind Generation Scheme using Neural Networks Mohammed Abdulla Abdulsada, Furat A. Abbas, Fathi R. Abusief and A. A. Hassan 2 Faculty of Engineering, Omar Al-Mukhtar University, Tobruk, Libya 2 Faculty of Engineering, Omar Al-Mukhtar University, Darna, Libya mohammed_alsaraj73@yahoo.com Abstract This paper presents the use of excitation capacitance variation for controlling of the output voltage of Self-Excited Induction Generator (SEIG) driven by wind turbine and supplies static load. The effects of rotor speed, load impedance and the excitation capacitance variations on the terminal voltage of the SEIG are discussed. An adaptive controller scheme based on Radial Basic Function Neural Network (RBFNN) is proposed to predict the suitable value of regulator capacitance for maintaining a constant output voltage of the SEIG. A programmable high speed controller (PHSC) is used to switch ON the required capacitor for providing the predicted capacitance. Computer simulation results are presented to demonstrate the performance of the SEIG with the proposed control scheme. The results proved that the proposed scheme is able to regulate the terminal voltage in spite of the wind speed and load variations. Keywords: wind generation, output voltage control, neural networks.. Introduction The wind energy market has grown because of the environmental advantages of harnessing a clean and inexhaustible energy source and because of the economic incentives supplied by several governments [].According to the American Wind Energy Association, the installed capacity of wind grew at an average rate of 29% per year. At the end of 2009, the worldwide installed capacity of wind energy was over 59 MW. The prediction capacity for 200 is over 203 MW [2].However, there are still many unsolved challenges in expanding wind power. The standard controls as well as recently developed advanced controls for variable-speed, horizontal axis wind turbines have been investigated [3].Most of the wind turbines are equipped with self-excited induction generators (SEIG). They are simple and rugged in construction and offer impressive efficiency under varying operating conditions. Characteristics of these generators like the over speed capability make them suitable for wind turbine application [4].Recent advancements in Power Electronics have made it possible to regulate the SEIG in many ways, which has resulted in an increased interest in the use of SEIG for power generation with wind power [5]. The use of artificial neural networks (ANNs) is the most powerful approach in artificial intelligence. One of the most important features of neural networks is their ability to learn and improve their operation using a set of examples named training data set [6]. ANNs have been used in diverse applications and play an important role in modelling and prediction of the performance and control of wind energy processes [7-0]. The need for adaptive regulating capacitance value comes from the fact that the wind turbine operates over a wide range of operating conditions, which means that the terminal voltage of the induction generator is not constant. Changing the value of regulator capacitance with the change of operating conditions (wind speed and loading conditions) can regulate the induction generator terminal voltage. In this paper, an adaptive controller scheme based on Radial Basic Function Neural Network (RBFNN) is proposed to predict the suitable value of regulator capacitance for maintaining a constant output voltage of the SEIG. A programmable high speed controller (PHSC) is used to switch ON the required capacitor for providing the predicted capacitance. Computer simulation results are presented to demonstrate the performance of the SEIG with the proposed control scheme. W E: Wind Energy Applications
25-30 September 200, Abu Dhabi, UAE 2. Effect of Excitation Capacitance on Amplitude of Terminal Voltage The steady-state operation of the SEIG may be analyzed by using the conventional equivalent circuit representation. The output voltage or terminal voltage V t is expressed in terms of air-gap voltage E and given by []: A E V t = () B 2 + D 2 Where : A 2 2 = ( X L X c a) + 3 ( X c R L ) B = ( R RL a + X L X ca + X X ca X X L a ), D = 2 2 ( R X L a X c R RL X c + RL X a ) The induction machine used as the SEIG in this investigation has the specification and parameters which given in appendix. Eq. () is used to demonstrate the effect of excitation capacitance on the terminal voltage of induction generator. Fig. shows the variation of terminal voltage V t around a desired value (20 volts) against the variation of rotor speed at different values of excitation capacitance and the load resistance is kept constant at 70 ohms. Fig. 2 shows the variations of terminal voltage versus the excitation capacitance at different values of rotor speed. Terminal voltage (V) C=54 uf C=76 uf C=200 uf 50 0 70 30 400 600 800 2000 Rotor speed (rpm) Figure Terminal voltage variation against rotor speed at different excitation capacitances N=600 rpm N=700 rpm N=800 rpm Terminal voltage (V) 60 0 60 40 90 240 Capacitance (uf) Figure 2 Terminal voltage variation versus excitation capacitance at different rotor speeds It can be noted that smaller values of excitation capacitance are required to maintain constant terminal voltage at high speeds and vice versa. At fixed rotor speed, the terminal voltage rises if larger values of excitation capacitance are used. W E: Wind Energy Applications 2
25-30 September 200, Abu Dhabi, UAE Fig. 3 shows the variation of terminal voltage Vt against the load impedance for different values of excitation capacitance and the speed is kept constant at 700 rpm. With fixed load impedance, the terminal voltage increases if larger excitation capacitor is used instead of a smaller one. terminal voltage (V) 40 20 00 80 60 C=54 uf C=76 uf C=200 uf 20 40 60 80 00 Load impedance (ohm) Figure 3 Terminal voltage variation against load impedance at different values of excitation capacitance 3. Adaptive Excitation Capacitor Based on Neural Network for SEIG The proposed RBFNN consists of three layers as shown in Fig. 4; the input, hidden and output layers. The input layer has two inputs, rotor speed and load impedance. The output layer has one output, which represent the prediction values of excitation capacitance. Figure 4 Proposed radial basic function neural network The rotor speed is changed in steps of 20 from 600 rpm to 900 rpm. The load impedance is increased gradually from 50 Ω to 00Ω. The calculated values of excitation capacitance are used as an output target of the proposed neural network. The results of the training are shown in Fig. 5.a. To test the generalization capabilities of the neural network, 96 operating conditions (rather than the training points) are used. The results of the test are depicted in Fig. 5.b, which shows that the RBFNN is able to predict the capacitor value for new operating conditions. 4. Implementation of the Proposed System Fig. 6 shows the proposed wind turbine - SEIG voltage control scheme. A fixed excitation capacitor and regulator excitation capacitor banks are connected in parallel at the stator terminal of the induction generator. A regulator capacitor is used to stabilize the SEIG terminal voltage for a wide range of operating conditions while the fixed capacitor is responsible for voltage build up. In this scheme, the induction generator, regulator capacitor and neural network are interfaced to the programmable high speed controller as shown in Fig. 6. In this figure, voltage and current sensors are used to measure the load impedance. The load impedance is equal the load voltage divided by the load current. A speed sensor is used to measure the generator s rotor speed. The RBFNN algorithm is used to adapt the desired regulator capacitor values for different operating conditions (different load impedances and rotor speeds). The desired values of the regulator capacitance, which meet most of the expected operating conditions, are stored in a programmable high speed controller. A ladder program is used in the controller to compare the predicted value of the regulator capacitance with the desired one to decide which capacitor must be ON. W E: Wind Energy Applications 3
25-30 September 200, Abu Dhabi, UAE The firing angle of the thyristor is zero or π, i.e., the thyristor acts as a switch to turn on the required capacitor. The firing signal of the thyristor is controlled by the PHSC. (a) (b) Figure 5 Predicted excitation capacitance at different operation condition (a) Training results, (b) Testing results Figure 6 Schematic diagram of the proposed system 5. Results and Discussions A comparison results between adaptive and constant excitation capacitors are investigated and explained in this section. Firstly, assuming that the excitation capacitance value of the SEIG is kept constant at 74 µf and the rotor speed and the terminal load impedance are changed simultaneously for a specific period. As expected, the terminal voltage is not constant in this case, because it depends mainly on the rotor speed and the load impedance. Therefore, to keep the terminal voltage constant, an adaptation scheme is used to adapt the excitation capacitance value, when the rotor speed and load impedance are changed. This scheme is based on the neural network and programmable high speed controller. Fig. 7.a show the changed rotor speed in step from 700 rpm to 750 rpm, then it is changed from 750 rpm to 660 rpm. Also, the load impedance is changed simultaneously from 50 Ω to 80 Ω and then changed to 60 Ω as shown in Fig. 7.b. Fig. 8 shows the variation of the excitation capacitance value with the simultaneous variation of the rotor speed and load impedance in comparison with the constant one (74 µf). Fig. 9 shows the corresponding variation of the terminal voltage of the SEIG with and without adaptive capacitance value. It is found that the value of the adapted capacitance swings between 48 µf and 85 µf to W E: Wind Energy Applications 4
25-30 September 200, Abu Dhabi, UAE maintain the terminal voltage constant at rated value (20 V), with the corresponding variation of the rotor speed from 660 rpm to 750 rpm and the load impedance from 50 Ω to 80 Ω. Generally, the proposed RBFNN is able to control successfully the terminal voltage of the SEIG by adapting the excitation capacitance for a wide range of operating conditions. Rotor speed (rpm) 800 750 700 650 600 0 2 3 Figure 7.a Variation of the rotor speed Load impedance 00 80 60 40 20 0 (ohm) 0 2 3 Figure 7.b Variation of the load impedance adaptive capacitance constant capacitance Capacitance (uf) 200 50 00 0 2 3 Figure 8 Variation of the excitation capacitance constant capacitance adaptive capacitance Terminal voltage (V) 40 30 20 0 0 2 3 Figure 9 Variation of the terminal voltage W E: Wind Energy Applications 5
25-30 September 200, Abu Dhabi, UAE 6. Conclusions In this paper, the output voltage of SEIG driven by wind turbine and supplies static load is controlled. A neural adaptive controller is used to control the generator terminal voltage at any operating condition. The use of an adaptive regulator capacitance value is motivated by the fact that the wind turbine generator operates over a wide range of operating conditions, and hence no single capacitance value is sufficient for regulating the terminal voltage. The RBFNN is used to predict the suitable value of regulator capacitor for any operating condition. Simulation results are presented to investigate the variation of terminal voltage when the rotor speed and load impedance are changed simultaneously with and without adapting the value of regulator capacitance. To maintain the terminal voltage of the SEIG constant at a desired value, large values of regulator capacitance are needed at low speed and small values of regulator capacitance are needed at high load impedance values and vice versa. 7. References [] K. Heinloth (2006) Energy Technologies Subvolume C: Renewable Energy, Springer-Verlag Berlin Heidelberg. [2] World Wind Energy Association. World wind energy installed capacity. http://www.wwindea.org. Accessed April / 5 / 200. [3] Jason H. Laks, L. Y. Pao and A. Wright (2009) Control of wind turbines: Past, present, and future, American Control Conference 2009 (ACC09), St. Louis, Missouri, USA, pp. 2096-203, June 0-2. http://ecee.colorado.edu/~pao/anonftp/lakspaowright_acc09.pdf [4] Sathyajith Mathew (2006) Wind Energy Fundamentals, Resource Analysis and Economics, Springer-Verlag Berlin Heidelberg, Germany. [5] Frede Blaabjerg and Zhe Chen (2006) Power Electronics For Modern Wind Turbines, Morgan & Claypool Publishers series,usa. [6] M.N. Cirstea, A. Dinu, J. Khor and M. McCormick (2002) Neural and Fuzzy Logic Control of Drives and Power Systems, Elsevier Ltd., Oxford. [7] Soteris A. Kalogirou (200) Artificial neural networks in renewable energy systems applications: a review, Renewable and Sustainable Energy Reviews (5), PP. 373 40. [8] Raja Singh Khela, Raj Kumar Bansal, K. S. Sandhu and Ashok Kumar Goel (2006) Application of Artificial Neural Network for Analysis of Self-Excited Induction Generator, Journal of Computer Science and Technology (JCS&T), October (6), No. 2, PP. 73-79. [9] M. Carolin Mabel and E. Fernandez (2008) Analysis of wind power generation and prediction using ANN: A case study, Renewable Energy 33 (5), pp. 986-992. [0] L. Rajaji and C. Kumar (2009) Neural network controller based induction generator for wind turbine applications, Indian Journal of Science and Technology, Feb. 2009, Vol.2 No. 2, PP. 70-74. [] R. M. Hilloowala (992) Control and interface of renewable energy systems, Ph. D. Thesis, The University of New Brunswick, Canada. Appendix: Specification and parameters of SEIG Induction machine: Rating: 3-phase, 2 kw, 20 V, 0 A, 4-pole, 740 rpm. Parameters: R = 0.62 Ω, R 2 = 0.566 Ω, L = L 2 = 0.05874 H, L m = 0.054 H. Self Excitation Capacitor: Rating: 76 µf / phase, 350 V, 8 A. Air gap voltage (E ) Variation of air gap voltage (E ) with magnetizing reactance at rated frequency induction machine; Xm < 82.292 E = 344.4.6Xm 95.569> Xm 82.292 E = 465.2 3.077Xm 08.00> Xm 95.569 E = 579.897 4.278Xm Xm 08.00 E = 0 W E: Wind Energy Applications 6