广东省电力负荷非线性预报与分析  

Nonlinear forecasting of electric load in Guangdong Province

作　　者：樊福梅[1] 曾智建 梁平[1] 

机构地区：[1]华南理工大学电力学院,广东广州510640 [2]广东省广电集团有限公司,广东广州510600

出　　处：《广东电力》2004年第3期1-4,共4页Guangdong Electric Power

基　　金：国家自然科学基金资助项目(50379015)

摘　　要：电力负荷预报是电力系统的重要工作之一。广东电力年负荷为非线性时间序列 ,根据时间序列的理论与方法 ,在分析广东电力历史负荷非线性时间序列运动轨迹的基础上 ,发现其电力负荷数据可由时间序列的典型分解式表示 ,选择期望值为噪声项的概率模型 ,并采用最小二乘法拟合趋势项 ,再根据参数估计和数据拟合优度的检验结果对时间序列未来取值进行预报。对广东电力负荷时间序列R/S分形分析进一步表明了预报值的合理性 ,也预示了今后 2 0～Load forecasting is an important task in the electric power system. As the annual electric load of Guangdong Province is a nonlinear time series, according to the theory and method of time series, and by analyzing the move trace of the nonlinear time series of Guangdongs historical electric load, we find out that the annual electric load of Guangdong can be showed by a typical resolution equation of the time series. We choose the expectation value as a probability model of noise item and use the least square method to fit the tendency item, then forecast the future time series according to the parameter estimation and check results of the optimum grade about the data fitting. The R/S fractal analysis of Guangdong load time series further shows that the forecasting value is reasonable and the electric load in Guangdong will maintain a sustained and stable increase in the following 20 to 30 years.

关 键 词：电力负荷 非线性 时间序列 预报 最小二乘法:分形布朗运动模型 R/S分析 

分 类 号：TB114[理学—概率论与数理统计] O212.1[理学—数学]