基于LWCA-SVM模型对洪泽湖饮用水源地二河闸断面水质的预测分析

Prediction and Analysis of Water Quality of Drinking Water Source of Erhezha in Hongze Lake Based on LWCA-SVM Model

二河闸位于洪泽湖出湖口,是二河上游重要的饮用水安全监测断面,为了对其富营养化指示因子TN和TP以及波动幅度较大的DO指标准确的预测,提出了基于领导者策略狼群搜索算法和支持向量机模型的饮用水水源地水质预测模型。对初始狼群进行了敏感性分析,得到狼群数量超过150时,模型的寻优效果最佳;利用LWCA-SVM模型对水质进行预测,得到DO、TN和TP的MSE、MAPE和Pearson系数依次分别为0.791、6.79%和0.931,1.32×10-4、0.5%和0.907,4.49×10-5、5.0%和0.903,说明基于LWCA-SVM的二河饮用水源地水质预测模型预报精度高,推广适应能力强,为二河水质预测提供了一种新方法。

Erhezha is located in Hongze Lake outlet which is the important monitoring section of drinking water safety in the upper reaches of Erhe River. In order to accurately predict the eutrophication factors TN and TP, and the DO with large fluctuation range in Erhezha, this paper puts forward wolves search algorithm based on the leader strategy and the drinking water quality prediction model supporting vector machine （SVM） model. The initial wolves are analyzed. When the number of initial wolves is more than 150, the parameteroptimization of the model is better. The LWCA-SVM model is used to predict water quality. The MSE, MAPE and Pearson coefficients of DO, TN and TP are 0.791, 6.79% and 0.931, 1.32×10^-4, 0.5% and 0.907, 4.49×10^-5, 5.0% and 0.903. The results show that the prediction model in the water quality of Erhe drinking water source based on LWCA-SVM has high precision and strong adaptability, which provides a new method for the prediction of Erhe water quality.