论文部分内容阅读
用BP神经网络方法对山坡平均山坡的解法进行分析,以29个小流域样本的水文数据为基础,通过应用人工神经网络反向传播BP(Back Propagation)算法,引入与山坡平均坡度密切相关的流域影响因子,并且通过调整网络结构中的权因子和阈值,建立了山坡平均坡度与流域影响因子之间的BP网络模型.计算结果表明,用拓扑结构为5-12-1的BP网络,经过学习150000次后,随机测试小流域样本的山坡平均坡度其计算结果和测试结果的相对误差不超过5%;证明该ANN模型的拟合能力强,从而为小流域山坡平均坡度的计算提供了一条新途径.
Based on the hydrological data of 29 small watersheds, BP artificial neural network (BP) backpropagation (BP) algorithm is applied to the watershed which is closely related to the average slopes of slopes And by adjusting the weights and thresholds in the network structure, a BP network model is established between the average gradient of the slope and the impact factor of the river basin.The calculation results show that using the BP network with topological structure of 5-12-1, after learning After 150000 times, the relative error between the calculated results and the test results is less than 5% for the random slopes of small watershed samples. The ANN model is proved to have good fitting ability and thus provides a new calculation method for the average slopes of small watershed slopes way.