Design of an adaptive dynamic load shedding algorithm using neural network in the steelmaking cogeneration facility

Abstract

A new adaptive dynamic under frequency load shedding scheme for a large industrial power system with large cogeneration units is presented. The adaptive LD- method with variable load shedding amount based on the disturbance magnitude is applied to have a minimum load shedding and a proper frequency recovery for different disturbances. To increase the speed of the load shedding scheme and to have an optimum response at different loading conditions, the artificial neural network (ANN) algorithm is developed. The Levenberg–Marquardt algorithm has been used for designed feed-forward neural network training. To prepare the training data set for the designed ANN, transient stability analysis has been performed to determine the minimum load shedding in the industrial power system at various operation scenarios. The ANN inputs are selected to be total in-house power generation, total load demand and initial frequency decay, while the minimum amount of load shedding at each step is selected for the output neurons. The proposed method is applied to the Mobarakeh steelmaking company (M.S.C) at different loading conditions. The performance of the presented ANN load shedding algorithm is demonstrated by the LD- method. Numerical results show the effectiveness of the proposed method.