论文部分内容阅读
将人工神经网络技术应用于河川径流实时预报，建立起河川径流实时预报的BP网络模型，并针对经典BP算法所存在的缺陷，采用共轭梯度优化和误差反向传播训练算法，使得所建立的BP网络模型的收敛性大为改善，消除和避免了实际应用中可能出现的局部优化问题．利用西大洋水库1975～1995年的入库径流系列资料，对所建立的BP网络模型进行训练和检验，同时探讨了网络结构对网络模型预报结果的影响．通过大量的实际应用和对比分析，表明BP网络模型比HG分析模型和相关图法更优越、更具有实际推广和应用价值．
The artificial neural network technology is applied to the real-time forecasting of river runoff, and the BP network model of real-time forecasting of river runoff is established. According to the defects of the classical BP algorithm, the conjugate gradient optimization and the error back propagation training algorithm are used to make the established The convergence of the BP network model is greatly improved, eliminating and avoiding the local optimization problems that may arise in practical applications. The data of runoff in the reservoir from 1975 to 1995 in the Western Ocean Reservoir are used to train and test the BP network model. At the same time, the influence of the network structure on the forecast results of the network model is also discussed. Through a large number of practical applications and comparative analysis, it shows that the BP network model is superior to the HG analysis model and the related graph method, and has more practical promotion and application value.