Published in IET Electric Power Applications Received on 13th February 2008 Revised on 20th May 2008 doi: 10.1049/iet-epa:20080038 ISSN 1751-8660 Recurrent wavelet neural network controller with improved particle swarm optimisation for induction generator system L.-T. Teng 1 F.-J. Lin 2 H.-C. Chiang 3 J.-W. Lin 1 1 Department of Electrical Engineering, National Dong Hwa University, Hualien 974, Taiwan, Republic of China 2 Department of Electrical Engineering, National Central University, Chungli 320, Taiwan, Republic of China 3 Department of Electrical Engineering, National United University, Miaoli 360, Taiwan, Republic of China E-mail: linfj@ee.ncu.edu.tw Abstract: A recurrent wavelet neural network (RWNN) controller with improved particle swarm optimisation (IPSO) is proposed to control a three-phase induction generator (IG) system for stand-alone power application. First, the indirect field-oriented mechanism is implemented for the control of the IG. Then, an AC/DC power converter and a DC/AC power inverter are developed to convert the electric power generated by a three- phase IG from variable frequency and variable voltage to constant frequency and constant voltage. Moreover, two online trained RWNNs using backpropagation learning algorithm are introduced as the regulating controllers for both the DC-link voltage of the AC/DC power converter and the AC line voltage of the DC/AC power inverter. Furthermore, an IPSO is adopted to adjust the learning rates to further improve the online learning capability of the RWNN. Finally, some experimental results are provided to demonstrate the effectiveness of the proposed IG system. 1 Introduction The recent evolution of power-electronics technologies has aided the advancement of variable-speed wind-turbine generation systems [1–5]. In spite of the additional cost of power electronics and control circuits, the total energy capture in a variable-speed wind-turbine system is larger when compared with that in the conventional one, resulting in lower life-cycle cost. Moreover, the pulse- width-modulation (PWM) converters not only can be used as a variable capacitor but also can supply the needed reactive power to load and to minimise the harmonic current and imbalance in the generator current. On the other hand, the variable-speed wind-turbine-driven IG systems display highly resonant, nonlinear and time-varying dynamics subject to wind turbulence and operating temperature of the IG. Furthermore, there is an appreciable amount of fluctuation in the magnitude and frequency of the generator terminal voltage owing to varying rotor speeds governed by the wind velocity and the pulsating input torque from the wind turbine. The phenomena of fluctuation are objectionable to some sensitive loads. Therefore the employment of PWM converters with advanced control methodologies to control the wind- turbine-driven IG systems is necessary [6, 7]. In addition, for the research of wind energy conversion systems, the developments of wind-turbine emulators are also necessary [8, 9]. Since the neural network control can tune the connective weights by the learning process to export an ideal control signal [10, 11], the amount of research on neural network control has considerably increased in the past decade. It has been proven that neural network can approximate a wide range of nonlinear functions to any desired degree of accuracy under certain conditions. Moreover, the characteristics of fault tolerance, parallelism and learning suggest that it is useful for implementing real-time control for nonlinear dynamical systems. Furthermore, wavelets are well suited to depict functions with local nonlinearities and variations owing to their intrinsic properties of finite support and self-similarity [12, 13]. With these specific IET Electr. Power Appl., 2009, Vol. 3, Iss. 2, pp. 147–159 147 doi: 10.1049/iet-epa:20080038 & The Institution of Engineering and Technology 2009 www.ietdl.org