基于集对分析的径流组合预测模型

Runoff Combination Forecasting Model Based on SPA

采用人工智能BP模型、小波BP模型及GA-BP模型对径流进行预测,然后将径流的实测值系列A和上述3种模型的预测值Bi建立集对H(A,Bi),利用集对分析的同、异、反特性进行联系度计算,据此确定径流预测模型的相对隶属度,并对隶属度进行归一化处理,得到上述3种模型的权重,再依据此权重建立相应的径流组合预测模型。应用1950~1975年小浪底水库的资料,对径流组合预测模型进行模拟,结果显示其预测精度明显高于单个模型的预测精度。

In this paper,the first use of artificial intelligence BP model,wavelet BP model and GA-BP model to predict runoff,then A series of runoff Found and the prediction value of the three models Biwere established set of H( A,Bi). The identity,difference and opposition characteristics of the set were used to calculate the contact degree,accordingly determine the relative membership degree runoff forecasting model,membership and normalized to give the right of the three models of weight,based on the weight of this combination to establish the appropriate runoff combination forecasting model. Application Xiaolangdi reservoir 1950 ~ 1975 years of data to simulate the runoff combination forecasting model. The results show that the prediction accuracy was significantly higher than its prediction accuracy of a single model.