摘要
地下水位埋深预测是有效实施节水措施、合理控制地下水位的前提与保证。鉴于地下水位埋深与引水量、降水量、蒸发量、排水量、地下水开采量等因素之间存在复杂的非线性关系,提出了基于改进遗传算法的BP神经网络模型用于地下水位埋深预测,弥补了BP神经网络本身易陷入局部最优值的缺陷;同时在遗传算法中改进了自适应交叉、变异概率算法,提高了神经网络的逼近能力和预测精度,并采用了Matlab软件实现了BP神经网络编程。通过不同模型在河套灌区解放闸灌域2000-2013年地下水位埋深模拟与预测,结果表明改进遗传算法优化的BP神经网络模型具有较高的预测精度和收敛速度,在其他地区地下水位埋深预测中也具有较好的适应性。
The prediction of groundwater level depth is the precondition and guarantee of water-saving measures application and groundwater level control. In order to solve the complicated non-linear relations between the groundwater level depth and the amount of irrigation water, discharge, evaporation, precipitation and groundwater exploitation, a BP （ Back Propagation） neural network model based on improved GA （Genetic Algorithm） is presented to forecast the depth of groundwater level, and it avoids falling into the local optimal solution. In this paper, adaptive crossover probability and adaptive mutation probability are improved in the genetic algorithm, which improves the approximation capability and prediction accuracy of the neural network. Actual data from 2000 to 2013 of Jiefangzha irrigation area is selected to simulate and forecast the depth of groundwater level with different models in MATLAB, and the result shows that the BP neural network model based on improved GA has higher prediction precision and faster convergent rate.
出处
《节水灌溉》
北大核心
2017年第9期81-84,共4页
Water Saving Irrigation
基金
水利公益性行业专项经费项目(201401007)
关键词
地下水位埋深预测
改进遗传算法
BP神经网络
解放闸灌域
groundwater level depth forecast
improved genetic algorithm
BP neural network
Jiefangzha irrigation area
作者简介
杜云皓（1993-），男，硕士研究生，从事节水灌溉与农业水土工程优化规划。E-mail：504376636@qq.com。
通讯作者：仇锦先（1971-），男，博士，副教授，从事灌排理论及水利规划优化研究。E-mail：qiujx@yzu．edu．cn。
