基于重构相空间AD-SVM的短期电力负荷预测  

Short-term Power Load Forecasting of AD-SVM Based on Phase Space Reconstruction

作　　者：黎武[1] 冯平[1] 李九林[1] 李树光 

机构地区：[1]后勤工程学院机械电气工程系 [2]91740部队

出　　处：《后勤工程学院学报》2014年第6期72-77,96,共7页Journal of Logistical Engineering University

摘　　要：利用重庆市九龙坡区电网2009年7月1日0:00—10月8日4:00 99 d共2 380个历史电力负荷数据,分析其特点和规律。将构建混沌理论的平均位移(AD)法和支持向量机(SVM)相结合,提出了一种新的短期电力负荷预测模型。通过仿真计算,将结果与神经网络法预测结果进行对比,可得新方法能较好反应数据变化趋势,并且具备较好的拟合能力,能够提高负荷预测精度。在实际短期电力负荷预测中,可优先选用平均位移法与支持向量机相结合的新方法。The characteristics and rules of the data of 2 380 historical power load in Jiulongpo district of Chongqing, from 0:00 July 1, 2009 to 4:00 October 8 were obtained and their characteristics and regularity were analyzed. Based on the average displacement (AD) method of chaos theory and the support vector machine(SVM), a new short?term power load forecasting model is built. Through simulation, the results of the new model and load forecasting method of chaotic neural networks is analyzed and compared. The results show that the support vector machine reflects the trend of data better and thus improves the precision of load forecasting. Therefore, in practical short?term power load forecasting, the new approach based on the average displacement method and the support vector machine can be chosen preferentially.

关 键 词：短期电力负荷 预测 重构相空间 支持向量机 神经网络 

分 类 号：TM74[电气工程—电力系统及自动化]