基于基因表达式编程算法的数据驱动型平板闸门流量计算方法  

Research on Flow Calculation of Flat Gate Based on Numerical Simulation and Gene Expression Programming Algorithm

作　　者：冯敏 章少辉[1,2] 白美健 张宝忠[1,2] FENG Min;ZHANG Shao-hui;BAI Mei-jian;ZHANG Bao-zhong(Key Laboratory of Watershed Water Cycle,Beijing 100048,China;China Institute of Water Resources and Hydropower Research,Beijing 100048,China)

机构地区：[1]流域水循环模拟与调控国家重点实验室,北京100048 [2]中国水利水电科学研究院,北京100048

出　　处：《节水灌溉》2024年第7期88-94,共7页Water Saving Irrigation

基　　金：国家重点研发项目(2022YFD1900804);中国水利水电科学研究院创新团队项目(ID0145B022021);中国水利水电科学研究院专项项目(ID0145B042021)。

摘　　要：灌区是国家水网建设的重要内容,精准量测和控制过闸流量过程是灌区用水管理的关键。为实时获取高精度的平板闸门过流量,以物理试验为基础,首先采用仿真模型分析了不同工况下闸前闸后流线与流速分布、水深-流量变化关系,其次,以仿真模型的模拟结果作为训练集,采用基因表达式编程算法(Gene Expression Programming,GEP)构建了融合闸前/后水深与开度的闸门过流量算法,并以更多的仿真模型模拟结果作为测试集,对比分析了该算法的性能。结果表明:①fluent数值模拟能在不同闸门开度和流量组合下重现复杂流线和流速分布过程,且模拟的流量过程与实测结果拟合良好,但效率低下,无法满足实际闸控的实时性需求;②GEP算法能够保持仿真模型的精度,且能达到实时计算;③与存在多个率定参数的经典闸孔出流分段公式相比,无任何率定参数的GEP算法的精度更高;④以仿真模型的模拟结果为基准,GEP算法比BP神经网络的精度更高、泛化性更好。因此,GEP算法更适宜平板闸门过流量的计算,可为灌区用水管理提供技术支撑。Irrigation areas are crucial for the national water network construction,and precise measurement and control of the flow process through gates are key to water management in irrigation areas.This study aims to achieve real-time high-precision flow rate estimation for flat gates through physical experiments.Firstly,a simulation model was used to analyze the flow line,velocity distribution and the relationship between water depth and flow rate under different working conditions.Secondly,the results of the simulation model were used as the training set,and the Gene Expression Programming(GEP)algorithm was utilized to construct a gate flow rate algorithm that integrates water depth and opening before/after the gate.More simulation results were used as the test set to compare and analyze the performance of the algorithm.The results show that:①Fluent numerical simulation can reproduce complex streamline and velocity distribution processes under different gate opening and flow rate combinations,and the simulated flow rate process fits well with the measured results.However,the efficiency is low and cannot meet the real-time requirements of actual gate control.②The GEP algorithm can maintain the accuracy of the simulation model and achieve real-time computation.③Compared with the classical segmented formula for gate outlet flow with multiple calibration parameters,the GEP algorithm without any calibration parameters has higher accuracy.④Based on the simulation results of the model,the GEP algorithm has higher accuracy and better generalization than the BP neural network.Therefore,the GEP algorithm is more suitable for calculating the flow rate of flat gate and can provide technical support for water management in irrigation areas.

关 键 词：数值模拟 平板闸门 流量 基因表达式编程 

分 类 号：TV663.1[水利工程—水利水电工程]