关键词: 中长期电力负荷, 模拟退火算法, 前馈型网络

Abstract: Due to the industrial structure adjustment, the change of resident consumption ability and pattern of consumption, and market-oriented and so on, long-term/mid-term power load forecasting for urban plans faces considerable difficulties. In the past two years, the methods that combined genetic algorithm and Back Propagation (BP) algorithm have been used for short-term power load forecasting rather than long-term/mid-term power load forecast of electricity. In this paper, a neural network prediction model with combination of genetic algorithm and BP neural network was established; the example in Nanchang was given to validate the accuracy of the algorithm, by comparing with the traditional BP neural network and Simulated Annealing (SA) prediction. Then the basic electricity load of Nanchang in the next dozens of years was forecasted and analyzed.

Key words: long-term/mid-term, Simulated Annealing (SA) algorithm, feed-forward type network