Control of a Shunt Active Power Filter with Neural Networks—Theory and Practical Results

This paper presents theoretical studies and practical results obtained with a four-wire shunt active power filter fully controlled with neural networks. The paper is focused on a current compensation method based on adaptive linear elements (adalines), which are powerful and easy-to-use neural networks. The reader will find here an introduction about these networks, an explanatory section about the achievement of Fourier series with adalines, and the full description of an adaline-based selective current compensator. The paper also brings a quick discussion about the use of a feedforward neural network in the current controller of the active filter, as well as simulation and experimental results obtained with the prototype of an active power filter.

[1]  Martin T. Hagan,et al.  Neural network design , 1995 .

[2]  M.G. Villalva,et al.  Current controller with artificial neural network for 3-phase 4-wire active filter , 2004, 2004 IEEE 35th Annual Power Electronics Specialists Conference (IEEE Cat. No.04CH37551).

[3]  Simon Haykin,et al.  Neural Networks: A Comprehensive Foundation , 1998 .

[4]  M. Rukonuzzaman,et al.  Adaptive neural network based harmonic current compensation in active power filter , 2001, IJCNN'01. International Joint Conference on Neural Networks. Proceedings (Cat. No.01CH37222).

[5]  J. R. Vazquez,et al.  Active power filter control using neural network technologies , 2003 .

[6]  M.M.A. Salama,et al.  Modular approach to active power-line harmonic filtering , 1998, PESC 98 Record. 29th Annual IEEE Power Electronics Specialists Conference (Cat. No.98CH36196).

[7]  Bernard Widrow,et al.  Neural nets for adaptive filtering and adaptive pattern recognition , 1988, Computer.

[8]  Bernard Widrow,et al.  30 years of adaptive neural networks: perceptron, Madaline, and backpropagation , 1990, Proc. IEEE.