针对电力负荷预测中单一模型不能充分利用数据信息和对其内在规律考虑不完全的问题,文中采用基于递归等权的组合预测模型,通过灰色关联度法对多个单一模型进行筛选,并确定参与组合的模型.再由递归等权法实现了对参与组合的各单一模型的变权重处理,有效地考虑各单一模型的预测好坏的变化.最后,通过对某地区最大负荷进行预测,对比单一模型与递归等权组合预测模型的预测误差.结果表明,递归等权组合预测模型比各单一预测模型的误差都小,从而验证了该模型能有效提高电力系统负荷预测能力,其精度高、结果可靠.%For solving the problem that a single model can not take full advantage of data information and consider internal law of data fully, a recursive right combination forecasting model was proposed in the paper. The single model selection in recursive right combination forecasting model can be achieved through the gray correlation analysis, and the weights of the single model can be also solved through recursive right method, which can effectively considering the variation of predict quality of the single models. Finally, the forecast deviations of the recursive right combination forecasting model and the single models are compared through a prediction of maximum load for a region. The results indicated that recursive right combination forecasting model has lower error than the single models, and can effectively improve the power system load forecasting capability, accuracy and reliability.
展开▼
