Genetic-based multilayered perception for Taiwan power system short-term load forecasting

A new approach using a genetic algorithm based multilayered perceptron (GA-MLP) to solve short-term load forecasting problems is proposed in this paper. The merit of genetic algorithms is that they prevent learning stagnation in the neural network learning process. The proposed method is tested on the load data from Taiwan Power Company. The results demonstrate the feasibility of the proposed approach for applications.