Abstract:

With the electricity consumption data in 2007—2014 as the training test sample,the mean square errors by the gray prediction method,the artificial neural network prediction method,and the  least squares support vector machine (LSSVM) prediction method based on neural network  are compared and analyzed.The results show that the a neural network-based LSSVM  method has the minimal error,and the prediction of the electricity consumption in an area of Anhui province from 2015 to 2017 is  close to the actual electricity consumption.

Key words: power load, prediction, medium and long term, artificial neural networks, least squares support vector machine