摘要
人工神经网络能够充分挖掘已知样本中的规律,从而对未观测数据进行预测,可应用于降雨量空间插值计算中。在BP神经网络进行降雨空间插值的基础上,引入遗传、粒子群和蚁群3种仿生算法对BP神经网络初始权值和阈值进行优化,将优化后的BP神经网络应用于三峡区间流域年、月和日3个时间尺度的降雨空间插值中。结果表明:仿生算法对BP神经网络初始权值和阈值优化求解后,降低了BP神经网络陷入局部最小以及过拟合的风险,在插值过程中表现出较好的稳定性,取得了理想的插值结果。
The artificial neural network can dig out the regulation of the observed data fully to have a prediction to the unobserved data, and can be applied in the spatial interpolation of rainfall. Based on the traditional BP neural network in spatial interpolation of rainfall, this paper introduces the genetic algorithm, particle swarm optimization and ant colony algorithm to optimize the initial weights and thresholds of BP neural network, and are applied in the spatial interpolation of rainfall of year, month and day such three temporal scales in the Three Gorges region. The results showed that after the optimization of initial weights and thresholds of BP neural network by bionic algorithms, it can reduce risk of BP trapping in the partial smallest and the over fitting. There is a better stability in the interpolation process to get the ideal interpolation.
作者
王汉涛
张潇潇
WANG Hantao;ZHANG Xiaoxiao(China Three Gorges Corporation, Yichang 443002, China;Three Gorges Cascade Dispatching & Communication Center, Chengdu 610042, China;College of Water Resource and Hydropower,Sichuan University, Chengdu 610065, China)
出处
《水资源与水工程学报》
CSCD
2019年第3期106-112,共7页
Journal of Water Resources and Water Engineering
基金
国家“十二五”水专项(2014ZX07104-005)
国家重点研发计划项目(2016YFC0402210)
关键词
遗传算法
粒子群算法
蚁群算法
BP神经网络
降雨空间插值
三峡区间流域
genetic algorithm
particle swarm optimization algorithm
ant colony algorithm
BP neural network
rainfall spatial interpolation
Three Gorges region basin
