基于遗传算法-生成对抗神经网络模型的宁夏自流灌区水资源优化调度研究  

Optimal dispatching of water resources in artesian irrigation district of Ningxia based on genetic algorithm-generative adversarial neural network model

作　　者：董陈超 DONG Chen-chao(Business School,Hohai University,Changzhou 213022,Jiangsu,China)

机构地区：[1]河海大学商学院,江苏常州213022

出　　处：《湖北农业科学》2021年第15期174-180,187,共8页Hubei Agricultural Sciences

基　　金：国家社会科学基金项目(17BTQ055);河海大学中央高校基本科研业务费资助项目(2018B04614)。

摘　　要：针对宁夏自流灌区灌溉用水存在农作物耗水量大、用水集中、灌溉效率低等现象,面向宁夏区域用水实际,围绕渠道进水闸和出水口,以满足农田基本灌溉用水为前提,以灌溉效率最大化为目标,采用机器学习方法,构建遗传算法-生成对抗神经网络的宁夏自流灌区水资源优化调度模型,并在宁夏秦汉渠管理处农场渠所管辖的30余公里渠道及其灌区进行验证和应用。结果表明,模型在学习传统调度方案的基础上深度挖掘各取水口用水规律,实现高效的取水口联合调度,月节约灌溉用水315109-1050362 m^(3),显著提高了宁夏水资源利用效率。In view of the phenomena of large crop water consumption,concentrated water use,and low irrigation efficiency in Ningxia artesian irrigation area.According to the actual water use in Ningxia,focusing on the water inlet and outlet of the channel,the premise is to meet the basic irrigation water of farmland,and the goal is to maximize irrigation efficiency.Using machine learning methods to build genetic algorithm-generative adversarial neural network model in Ningxia artesian irrigation districts,and verify and apply them in more than 30 kilometers of channels and irrigation areas of the Qinhan Canal Management Office in Ningxia.The results show that the model deeply excavates the water usage rules of each water intake on the basis of learning traditional scheduling schemes,estab⁃lishes efficient water intake joint scheduling irrigation methods,saves 315109-1050362 m^(3) of irrigation water per month,and signifi⁃cantly improves the efficiency of water resource utilization in Ningxia.

关 键 词：灌溉水资源 优化调度 遗传算法 生成对抗神经网络 机器学习 

分 类 号：TV213.9[水利工程—水文学及水资源]