A Novel Particle Swarm Grey Neural Network Model for Power Load Risk Forecasting

In order to establish a high accuracy forecasting model for short-term electric power load, this paper made a change to grey differential equation utilizing the fundamental theorem of discrete time function. Through mapping the parameters of the equation into the BP neural network, giving the corresponding parameters when the sequence sample of load was converged in the network. In this case, optimizing the deviation of gray neural network step by step utilizing the quickly hunt ability of overall situation of the particle swarm optimized model and establishing the gray neural network forecasting model-PGNN with less deviation based on particle swarm optimization. Finally, the model's effectiveness and accuracy were examined through a case study. The result by computer simulation suggested that the new model had a high accuracy for forecasting.