专家系统与BP神经网络相结合的短期负荷预测  

Short-term Power Load Forecasting Based on Combination of Expert System and BP Neural Network

作　　者：李莉[1] 孔哲峰 李鹏 张海蓉 刘崇新[1] 

机构地区：[1]西安交通大学电气工程学院,陕西西安710049 [2]咸阳供电局,陕西咸阳712000 [3]陕西省电力公司,陕西西安710048

出　　处：《陕西电力》2009年第1期22-27,共6页Shanxi Electric Power

摘　　要：研究了专家系统结合神经网络BP算法在短期电力负荷预测中的应用。对神经网络BP算法进行改进,使用BP算法对咸阳电网实际负荷数据进行预测,并将预测值与实际负荷值进行比较,总结其长期的发展变化规律。同时汲取有关专家学者和专业预测人员的经验知识,形成一系列的规则集,从而模拟人类专家的决策过程进行推理和判断,形成一个专家系统,以此来改进采用单一BP算法进行预测的种种不足。结果表明,经验知识越成熟,推理规则越完备,对提高预测精度越有利,对神经网络BP算法的预测值进一步修正的可能性越大。The application of expert system combined with BP algorithm used in neural network in short-term power load forecasting is researched. BP algorithm for the neural network is analyzed and applied to load forecasting of Xianyang power grid, the actual values are compared with the forecasting data to summarize its long-term development law. At the same time, the experience of professional knowledge is learned from some experts ＆ scholars concerned as well as professional forecasters to generate a series of rules sets, which helps to infer ＆ judge by simulating the decision process of human experts and form an expert system, to improve the deficiency in forecasting by simple BP algorithm. The results show that prediction accuracy became better and better with increasing maturation ＆ perfection of experience knowledge ＆ inference rule, and predictive value obtained by the neural network BP algorithm has much more possibility to be further corrected.

关 键 词：负荷预测 神经网络BP算法 专家系统 智能决策 

分 类 号：TM715[电气工程—电力系统及自动化]