Using an adaptive self-tuning approach to forecast power loads

To overcome the difficulty in forecasting power loads precisely, an adaptive self-tuning approach is proposed. Using an RBF neural network as a predictor and building an evaluator to evaluate the output of the predictor, then, according to the evaluator's judgment, the predictor adjusts its structure and weight by using critical self-learning mechanics. The predictor can keep the same pattern as the current power loads state and power loads can be forecasted precisely. The simulation of practical date indicated that this method is effective.