A New Method of Electric Power Load Forecasting

In order to solve the situations in which the neural networks can not reflect the volatility of the power load in time or some problems of the time cumulative effect, the process neural network is established in this paper. Aiming at the lower speed in the process of neural network training, the problems of convergence uneasily and so on, we design the accelerate disturbance particle swarm to optimize each parameters of the process neural network. The algorithm uses basis function to expand the network input and connection weights, and then use the accelerate disturbance particle swarm optimization algorithm to optimize the factors. If doing like this, we can lower the complexity of the neural network and ensure the global of the algorithm optimization. When finishing the predicting of the practical power load by the new way, we found that the method of using acceleration disturbance particle swarm to optimize neural network can improve the accuracy and timeliness of the load forecasting effectively.