Application of Hopfield Neural Network for Harmonic Current Estimation and Shunt Compensation

Harmonics generated by nonlinear loads pollute the power system and affect the operation of equipment connected to it. Hence, harmonic mitigation is of prime concern to a power system engineer. Artificial Neural Network (ANN) is a nonlinear signal processing technique, which is built from interconnected elementary processors called neurons. In this article, a Hopfield Neural Network (HNN) based control algorithm for shunt compensator in a power distribution system is realized. The Hopfield network is modeled using energy minimization principle and consists of “n” interconnected neurons. The HNN is used to estimate different harmonic components present in distribution system operating with nonlinear loads. It also provides suitable control signals to the shunt compensator for compensation of various power quality issues such as power factor correction, load balancing, and harmonic reduction in the distribution system. Detailed experimental results are presented along with simulation studies on the prototype model developed in the laboratory and these results demonstrate the feasibility of the proposed method of control in DSTATCOM. The comparison of the HNN-based compensation technique with a popular and effective control algorithm based on Least Means Square (LMS) is also presented in this article.