关键词: 最小二乘支持向量机, 粒子群算法, 水文预测, 参数优化, 回归

Abstract: Support Vector Machine (SVM) algorithm provides a new way for the study of mid-and-long term hydrological forecasting that needs a learning of finite samples. Concerning the time-consumption and unsatisfactory performance in the conventional parameter choosing method, a Least Square Support Vector Machine (LS-SVM) model based on Particle Swarm Optimization (PSO) was given in this paper. The model was built by using the regression principle of least square support vector machine, the key parameters in this model were optimized by PSO algorithm with random seeking strategy. Monthly runoff forecasting in Yele Hydropower Station on Nanya river indicates that the algorithm is able to promote efficiency and accuracy.

Key words: Least Square-Support Vector Machines (LS-SVM), Particle Swarm Optimization (PSO), hydrological forecasting, parameter optimization, regression