论文部分内容阅读
通过对比相关系数法、逐步回归法及综合方法3个预报因子筛选方案的模拟结果,确定优选预报因子的最佳方法,采用BP人工神经网络模型对大通、屏山和汉口3个站点进行枯水期(当年11月-次年4月)径流预报研究.结果显示,采用相关系数法初选及逐步回归法优选所筛选出的预报因子集合可以得到更好的预报效果;该模型在枯水期月尺度径流预报中,检验期的平均合格率为56.44%,达不到实际预报的需求.而采用旬尺度模拟计算月径流的预报效果要远远高于月尺度径流模拟,检验期平均相对误差与合格率分别为12.27%和71.63%,有较好的预报精度.可以为长江流域水文预报工作提供一定的参考.
By comparing the simulation results of the three forecasting factor screening schemes by the correlation coefficient method, the stepwise regression method and the comprehensive method, the best method of selecting the best forecasting factor is determined. The BP artificial neural network model is used to simulate the dry season in the three sites of Datong, Pingshan and Hankou November-April of the following year.) The runoff forecast study showed that the forecasting results could be obtained better by using the correlation coefficient method and the stepwise regression method to optimize the selected set of predictors. In the monthly runoff forecast of dry season , The average passing rate of the inspection period was 56.44%, which could not meet the actual forecast requirements.While forecasting the monthly runoff by using the ten-yard scale simulation is much higher than the monthly runoff simulation, the average relative error and passing rate in the inspection period were 12.27% and 71.63%, respectively, which has better prediction accuracy and can provide some reference for hydrological forecasting in the Yangtze River Basin.