城市生活垃圾产量的变权重组合预测

Variable weights combination forecasting of municipal solid waste output

基于预测显著增加的不确定性因素,在预测时采用单一模型进行预测通常难以达到较为理想的预测效果.选择作为中国东北城市的哈尔滨市为分析对象,以1992～2005年的数据为基础,将多元线性回归模型、GM（1,1）模型、三次指数平滑法这三种单项预测模型进行变权组合,预测哈尔滨市2006-2010年的生活垃圾产量.单一模型的局限性得到明显改善,有效地集结了更多的有用信息,组合预测模型的预测精度得到明显提高,改善了预测结果.

The uncertainty factors of prediction is increasing dramatically, so it is often difficult to achieve ideal forecasting effect using the single forecasting model. In this paper, Harbin, northeastern city of China, was chosen as the analysis object. Variable weights combination was performed based on the data from 1992 to 2005, using multiple linear regression model, GM （ 1, 1 ） model and three exponential smoothing as the single forecasting model. Then Harbin domestic waste output in 2006 - 2010 using prediction to forecast of was predicted. The limitations of single model were improved, and more useful information was gathered effectively. Results showed that the prediction precision of combination forecast model has been obviously improved.