基于PSO-LSSVM算法的阶梯式溢洪道复氧率预测

Prediction of Re-aeration Rate of Stepped Spillway Based on PSO-LSSVM Algorithm

最近,机器学习方法逐渐在水利工程中得到广泛运用。研究将采用最小二乘支持向量机(LSSVM)方法,建立阶梯式溢洪道各种流态下复氧率的预测模型。采用粒子群优化算法(PSO)优化了LSSVM算法的参数(惩罚函数γ和核函数常数σ^2),新的PSO-LSSVM模型预测精度相对于常用的BP模型明显提高。误差分析表明,在测试集上PSO-LSSVM模型的平均绝对百分比误差MAPE、均方差RMSE和平方相关系数R^2分别为1.100 0×10^-3, 4.899 6×10^-4和9.998 6×10^-1。最后,采用平均影响值法评价了输入参数对复氧率的影响程度。

Recently, machine learning methods have been widely used in hydraulic engineering. In this study, the least squares support vector machine(LSSVM) algorithm is used to establish a prediction model for the re-aeration rate of stepped spillway under various flow regimes. Particle swarm optimization(PSO) is used to optimize the parameters(the penalty parameter γ and kernel constant σ^2) of the LSSVM algorithm. The prediction accuracy of the new PSO-LSSVM model is significantly improved compared with the commonly used BP model. Error analysis shows that the average absolute percentage error MAPE, root-mean-square error RMSE and square correlation coefficient R^2 of PSO-LSSVM model on test sets are 1.100 0×10^-3, 4.899 6×10^-4 and 9.998 6×10^-1, respectively. Finally, the effect of input parameters on the re-aeration rate is evaluated by means of the mean impact value method.