A DTW and BPN Based Approach for Modeling Nonlinear Load

Abstract

It is known that a direct application of artificial neural network (ANN) for modeling high power, nonlinear, and impact load may lead to inaccuracies. This paper proposes an approach that the characteristic of the high power, nonlinear and time varying load is modeled by advanced ANN, which is implemented by using back propagation network (BPN) and dynamic time warping (DTW) algorithms. The model of the nonlinear load can be accurately established, the consumption of the voltage, reactive power and the precision of the model can be obtained. Then the type of the reactive power compensation devices and the capacity of the reactive power compensation equipments can be effectively determined.