基于Theil不等系数的IOWA算子和马尔科夫链的电量组合模型研究

摘要

通过引进Thei1不等系数、IOWA算子和马尔科夫链的概念,建立了一种基于Thei1不等系数的IOWA算子和马尔科夫链的电量组合模型.从相关性的角度对该模型进行了研究,同时采用IOWA算子,使组合模型的各个时间点上的权系数只与预测精度有关,与预测方法无关,较好地反映了负荷发展实际情况.利用马尔科夫链定性推出组合模型中各单项模型在待预测时点上的预测精度状态,进而得到待预测时点上的组合模型的权系数.以陕西省某市1994 ～ 2009年年用电量为样本,通过指数平滑法、回归法和灰色模型法分别建立模型,然后利用基于Theil不等系数的IOWA算子和马尔科夫链的组合模型进行权系数的求解,实证分析表明该模型使预测精度得到了明显提高,具有良好的预测效果.%A new combination forecasting model is proposed based on Theil coefficient and the Induced Ordered Weighted Averaging (IOWA) Operator and Markov chain (MC). The combined forecasting model is constructed by improving the Theil coefficient, meanwhile, by use of IOWA the weight coefficient is bound up with the forecasted accuracy, rather than the forecasting method, which reflects the reality of load. By use of MC the forecasted accuracy condition of each forecasting method at the forecasted time point can be qualitatively surmised, thus its weight coefficient at the forecasted time point can be determined. This paper respectively makes use of Regression, Exponential Smoothing and Grey forecasting to construct models based on the annual electricity consumption of one city in Shaanxi province from 1994 to 2009. Then the combination forecasting model based on IOWA operator and MC is used to solve the weight coefficients. The result illustrates that the new model can improve forecasting precision effectively and possesses better practicability.