基于修正的时间序列模型对冲击负荷的预测

电力负荷预测的实质是对电力市场需求的预测,是利用以往的历史数据资料找出电力负荷的变化规律,进而预测负荷在未来时期的变化趋势.由于经济、气候以及工业生产等诸多因素的约束和限制,电力负荷预测精度很难提高.一个好的实用的电力负荷预测模型则要求既能充分利用负荷的历史数据,又能灵活方便地综合考虑其他多种相关因素的影响.提出了回归与自回归模型相结合的时间序列混合回归预测模型,它的待估参数由BP神经网络进行修正,经实例验证,预测效果良好.

Forecasting of Load Impact Based on Modified Models for Time Series

The forecast of electric load is in essence a prediction of the demand in market for electricity, which is carried out through forecasting of the future trends in change of electric load, based on the regularity in change of electric load obtained according to historical data. Due to the influence of a number of factors such as economy, climate, industrial production, and so on, it is hard to improve the degree of accuracy in forecast of electric load. As a feasible and practical model for forecasting electric load, not only must it make full use of the historical data, it is also supposed to take into consideration of all other relevant factors in a flexible and convenient manner. This paper provides a time series mixed regressive forecasting model, which integrates regressive models and autoregressive models and in which the expectant parameters are corrected through BP neural network. Utilization of the model in practice shows that this model has worked well in forecasting electric load.