School of Business and Administration, North China Electric Power University, Baoding 071003, China

---

Abstract—Comparison with the classical BP neural network, the generalized regression neural network requires not periodic training process but a smoothing parameter. The model has steady and fast speed, and meanwhile, the connection weight of different neurons is not necessary to be adjusted in the training process. The paper establishes the index system of GRNN forecasting model, and then uses Bayes theory to reduce them, which will be inputting variables of GRNN model. The method is testified to get higher speed and accuracy by simulation of actual data and comparison to classical BP neural network.

Index Terms—Bayes, load forecasting, generalized regression neural network

[PDF]

Cite: Yanmei Li and Jingmin Wang, "The Load Forecasting Model Based on Bayes- GRNN," Journal of Software vol. 7, no. 6, pp. 1273-1280, 2012.