基于多种神经网络的水沙预测模型研究  

Research on water and sediment prediction model based on multiple neural networks

作　　者：张明旺 侍克斌[1] 孔选昭 姚海坤 ZHANG Mingwang;SHI Kebin;KONG Xuanzhao;YAO Haikun(College of Water Conservancy and Civil Engineering,Xinjiang Agricultural University,Urumqi 830052,China)

机构地区：[1]新疆农业大学水利与土木工程学院,新疆乌鲁木齐830052

出　　处：《武汉大学学报（工学版）》2023年第9期1057-1065,共9页Engineering Journal of Wuhan University

基　　金：新疆水利科技专项资金项目(编号:T201801)。

摘　　要：为解决水沙变化机理中水库泥沙难以预测的问题,采用支持向量回归(support vector regression,SVR)、遗传算法-支持向量回归(genetic algorithm-support vector regression,GA-SVR)、粒子群优化-支持向量回归(particle swarmoptimization-supportvectorregression,PSO-SVR)、最小二乘支持向量回归(leastsquaressupportvector regression,LSSVR)等模型进行研究,并将模型应用于黑孜水库水文站的年径流量及年输沙量序列的预测分析。数据资料序列划分为85%的训练数据和15%的测试数据,使用均方根误差、相关系数、平均绝对百分比误差、纳什系数这4项指标来评价模型的精度,得到的结果可以为水库的短期调度和长期管理提供帮助。结果表明:LSSVR算法是4种模型中最优的,其误差最小,精度最高;构建的模型可为其他地区的水沙预测提供参考。In order to solve the problem that reservoir sediment is difficult to predict in the mechanism of water and sediment variation,this paper adopts support vector regression(SVR),genetic algorithm-support vector regression(GA-SVR),particle swarm optimization-support vector regression(PSO-SVR),least squares support vector regression(LSSVR)and other models for research,and applies the model to the series of annual runoff and annual sediment transport of the Heizi Reservoir hydrological station for prediction and analysis.The data series is divided into 85%training data and 15%test data,and the accuracy of the model is evaluated by four indicators:root-mean-square error,correlation coefficient,mean absolute percentage error and Nash coefficient.The obtained results can provide help for short-term scheduling and long-term management of the reservoirs.The results show that the LSSVR algorithm is the best among the four models,with the smallest error and the highest accuracy.The constructed model can provide a reference for the prediction of water and sediment in other regions.

关 键 词：支持向量回归算法 遗传算法-支持向量回归算法 粒子群优化-支持向量回归算法 最小二乘支持向量回归算法 径流量预测 输沙量预测 

分 类 号：TV124[水利工程—水文学及水资源]