论文部分内容阅读
鉴于基流过程对降雨不敏感,致使传统的BP神经网络日径流预测性能受到制约的问题,结合LyneHollick(LH)数字滤波算法和BP算法的优点,建立了基于LH分割基流与BP神经网络日径流预测的松散耦合模型(LH-BP)。先采用LH数字滤波算法分割出基流,再利用BP神经网络预测锦江流域四个水文站的直接径流和基流。结果表明,LH-BP耦合模型较传统的BP模型性能更优,弥补了传统的BP模型对日径流模拟与预测的不足。
In view of the fact that the base course is not sensitive to rainfall and the traditional BP neural network is constrained by the daily runoff forecasting performance, and based on the advantages of the LyneHollick (LH) digital filtering algorithm and the BP algorithm, a method based on the LH segmented base flow and the BP neural network Runoff Prediction of Loose Coupling Model (LH-BP). Firstly, LH digital filtering algorithm was used to segment the base flow and then BP neural network was used to predict the direct runoff and base flow of the four hydrological stations in the Jinjiang River Basin. The results show that the LH-BP coupling model is better than the traditional BP model, which makes up for the shortcomings of the traditional BP model on the simulation and forecast of the daily runoff.