摘要
收集柳江流域的24个气象站点降雨数据和重要水文站流量数据,采用Pearson相关系数法和Spearman相关系数法对流域不同时间尺度的降雨量和径流深进行相关性分析,并基于遗传算法和粒子群算法,利用支持向量机模型开展多时间尺度的流域降雨径流模拟研究。结果表明:(1)柳江流域年降雨量和年径流深、汛期降雨量和汛期径流深、主汛期降雨量和主汛期径流深表示的相关性基本一致,都有明显的相关性,通过了99%的显著性水平检验。降雨量和径流深的相关性在主汛期最高,汛期次之,枯水期最差,这主要与枯水期径流多依靠水利工程的补水调节作用有关。(2)支持向量机模型在柳江流域的年径流量、汛期径流量和主汛期径流量的模拟中,通过遗传算法和粒子群算法优选出的参数的模拟效果都是相对准确的。在柳江流域年径流量的模拟中,粒子群算法优选的参数得到的支持向量机模型泛化能力更好,而在汛期径流和主汛期径流的模拟中,遗传算法优选的参数得到的支持向量机模型更加平稳。模拟结果可以为柳江流域防灾减灾提供科学依据。
Based on the rainfall data at 24 meteorological stations and the discharge data of important hydrological station in the Liujiang River Basin, we analyzed the correlation between rainfall and runoff at multi time scales by Pearson correlation and Spearman correlation analysis. Then we used Support Vector Machine （SVM） to study the simulation of rainfall-runoff at multiple time scales based on particle swarm optimizer （PSO） and genetic algorithm （GA）. The results show that： （1） The rainfall has a significant positive correlation with yearly runoff flood season and main flood season, with the confidence level being 99%. The correlation of rainfall and runoff in the main flood season is the highest, and it is also high in the flood season. While the correlation is the lowest in the dry season, because the runoff relies on the water conservancy project. （2） In the simulation of SVM in annual flood season and main flood season, the simulation results of the parameters optimized by genetic algorithm and particle swarm optimization are relatively accurate. The simulation results of the parameters optimized by particle swarm optimization is better in the annual period, while the simulation results of the parameters optimized by genetic algorithm is better in the flood season and the main flood season. The optimal parameters by the two algorithms are applied well to runoff simulation in the Liujiang River Basin. Simulation results serve as a reference for disaster prevention and reduction in the Liujiang River Basin.
出处
《中国农村水利水电》
北大核心
2017年第12期64-69,共6页
China Rural Water and Hydropower
基金
"十二五"科技支撑计划项目(2015BAK11B02)
国家自然科学基金项目(50479033)
西江流域水文气象耦合洪水预报技术研究(201301070)
广东省科技计划项目(2013B020200007)
