基于时间序列分析的负荷预测方法的比较研究

A Comparative Study of Load Forecasting Methods based on Time Series Analysis

电力系统负荷预测是电力系统规划和经济运行的基础.趋势移动平均方法和指数平滑方法可以对区域内不同种类负荷时间序列建模,趋势移动平均方法是对负荷时间序列历史数据进行处理,建立趋势分量以及季节分量模型,对残差进行平稳性检验后建立ARMA模型;指数平滑方法是建立Holt-Winter季节乘法模型.Eviews、Excel等软件可获得负荷未来值,并对模型预测功能进行评价,分析两种时间序列方法的差异性,结果表明采用时间序列分析的方法可以取得较好的预测效果.

Load forecasting of power systems plays an important role in power system planning and economic operation.Trend moving averaging method and exponential smoothing method can be used to model different types of load time series in a region, process load time series historical data through Autoregressive Moving Average with trend method, and establish trend component and seasonal models. ARMA model is built after stationarity of residuals is tested. Exponential smoothing method is then applied to build Holt-Winter seasonal multiplication model. EVIEWS, EXCEL and other software are used to obtain future load value and evaluate the model prediction function. The results show that better prediction results can be achieved by using time series analysis method.