基于小波分解的工程材料价格组合预测方法

Combination Forecast Method of Engineering Material Price Based on Wavelet Decomposition

为准确地估计工程材料价格在建设期内的变化情况,考虑到材料价格时间序列具有长期性、周期性和波动性的特点,本文建立了ES-BPANN模型对其进行预测.在预测模型中,引入小波变换方法对价格序列进行分解,针对每个分量的实际特征建立相应的独立预测模型,并综合每个子模型的预测情况作为最终预测结果.选用福州市2007-2017年C25现浇混凝土信息价进行实例验证,结果表明本文模型的预测精度高于单一预测模型,能够为工程投标报价方案制定提供一定指导.

In order to estimate the change of engineering material price accurately during the construction period,this paper fully considers the long-term trend,periodicity and volatility of material price time series,and the ES-BPANN model is established to predict the price.In the forecasting model,this paper considers introducing wavelet transform to decompose the price series,and establishes corresponding independent prediction models according to the characteristics of decomposition components respectively.The material price is forecasted by synthesizing the forecasting results of each component sub-model.The data of information price of C25 cast-in-situ concrete in Fuzhou from 2007 to 2017 are used to test and verify the model.The results show that the model has higher accuracy than single prediction model and can provide some guidance for the formulation of project bidding quotation proposals.