基于SWMM和改进NSGA-Ⅱ算法的雨水管网多目标优化研究

Multi⁃objective Optimization of Rainwater Pipe Network Based on SWMM and Improved NSGA-ⅡAlgorithm

为解决城市内涝问题和极端降雨影响,保证雨水管网系统正常运行。首先设置不同重现期的雨型,利用SWMM(storm water management model)模型对管线进行模拟输出不同重现期下的管网溢流程度。以管网投资最小和排水性能最大为目标函数,管网的水压、流速、充满度等为约束条件,建立多目标管网优化模型,以云南石屏县为例进行分析,将惩罚函数引入NSGA-Ⅱ(non-dominated sorting genetic algorithms-Ⅱ)算法对模型求解,采用精英保留策略反向约束选择最优个体。经优化后,排水效率得到改善,溢流量降至100 m^(3)以下。从收敛速度和最优解来看,其优化结果与GA(genetic algorithm)算法和普通NSGA-Ⅱ算法对比寻优效果更好;为克服特殊个体早熟或随机发散,缩小目标解的范围,将F2取值范围设为[0,300],依次采用不同算法评估。通过统计分析与最优解偏差可知,改进NSGA-Ⅱ算法的计算精度更高,可达到0.96,评估次数为3656次,用时最短,其寻优效率更高。利用改进NSGA-Ⅱ算法对模型求解,其收敛速度更快,评估次数更少。该方法能为实际工程的管网优化问题提供新的思路。

In order to solve the problem of urban water logging and the influence of extreme rainfall,ensure the normal operation of rainwater pipe network system.Rain patterns in different return periods were set,and SWMM(storm water management model)model was used to simulate the pipeline network overflow degree in different return periods.Taking the minimum investment and maximum drainage performance as objective functions,and the water pressure,flow rate and filling degree of the pipe network as constraints,a multi⁃objective pipe network optimization model was established.Taking Shiping County,Yunnan Province as an example,the penalty function was introduced into the NSGA-Ⅱ(non⁃dominated sorting genetic algorithms⁃Ⅱ)algorithm to solve the model,and the elite retention strategy was used to select the optimal individual with reverse constraints.After optimization,the drainage efficiency is improved.The overflow flow falls below 100 m^(3).In terms of convergence rate and optimal solution,the optimization results are better than those of GA(genetic algorithm)algorithm and ordinary NSGA-Ⅱalgorithm.In order to overcome prematurity or random divergence of special individuals and narrow the range of target solutions,the value range of F2 was set as[0,300],and different algorithms were used successively for evaluation.According to the deviation between the statistical analysis and the optimal solution,the calculation accuracy of the modified NSGA-Ⅱalgorithm is higher,up to 0.96,the evaluation times are 3656,the shortest time,and the optimization efficiency is higher.Using the improved NSGA-Ⅱalgorithm to solve the model,the convergence speed is faster and the evaluation times are less.This method can provide a new idea for the optimization of pipe network in practical engineering.