清华大学学报（自然科学版）2026，Vol.66Issue(2)：309-323,15.DOI:10.16511/j.cnki.qhdxxb.2026.27.002
基于定向抽样多目标进化算法的塔阀联合调控优化
Regulation of tower-valve joint optimization model based on directed sampling multi-objective evolutionary algorithm
摘要
Abstract
[Objective]Water supply systems with branch pipes,relying exclusively on valve control,frequently demonstrate deficiencies in effectively mitigating water hammer during valve closure.This inadequacy often results in transient system pressures surpassing safety thresholds,thereby compromising operational safety.Consequently,there is an imperative for research on the combined protection of bidirectional surge tanks and valves in such systems.However,the coordinated control of tanks and valves involves numerous variables and strongly coupled nonlinear relationships,in which parameters interact,rendering it a complex optimization problem.In light of the challenges associated with meeting multiple safety requirements concurrently,it is essential to conduct a comprehensive investigation into the synergistic mechanisms between tanks and valves and to formulate a corresponding optimization and decision-making method that will facilitate the attainment of globally optimal control strategies.[Methods]The present study developed an integrated optimization model for the coordinated control of tanks and valves.The model treats the control parameters of both the valves and the bidirectional surge tank as decision variables.The objective functions were defined as minimizing the extreme pressure values in the system and shortening the valve closure duration.A novel optimization and decision-making method was proposed,which utilizes a large-scale multi-objective evolutionary algorithm with directed sampling combined with the technique for order preference by similarity to ideal solution(LMOEA-DS-TOPSIS).To solve the optimization model,LMOEA-DS was employed,and TOPSIS was applied to select the optimal solution from the generated Pareto front.[Results]An investigation into a particular water supply project in Shandong,China,revealed that the combined control of the main pipeline's end valve and the branch pipe's end valve was inadequate in ensuring safety compliance.This finding necessitated further optimization of the subsequent tank-valve joint control strategy.To address this challenge,the LMOEA-DS algorithm was employed to solve the established multi-objective optimization model for tank-valve coordination.The parameters were set as follows:50 individuals,30 guiding solutions,and a total population size of 8 000.The hypervolume(HV)and spacing(SP)metrics were 32 062.6 and 0.527 8,respectively,indicating adequate convergence performance.Moreover,the resulting Pareto front demonstrates considerable spans of 14.62 m,4.36 m,and 240 s across the three objective dimensions,with solutions exhibiting uniform distribution and adequate accessibility.This outcome suggests that Pareto solutions,when distributed widely,can be obtained in a relatively brief computational time.The optimization effect is particularly noteworthy in the context of the positive pressure protection objective.The collective analysis of these results substantiates the effectiveness of the LMOEA-DS algorithm in addressing the proposed model and validates the rationality of the parameter settings.Subsequently,employing a TOPSIS analysis to select the optimal tank-valve joint scheme from the Pareto front resulted in reductions of 21.25％and 47.72％,respectively,in the amplitude of the maximum and minimum pressure envelope lines.Furthermore,a 5.43％shortening of the total valve closure time was demonstrated.This outcome significantly enhanced the operational stability of the system and confirmed the superiority of the LMOEA-DS-TOPSIS optimization and decision-making method.[Conclusions]The results verify that the synergistic strategy employing a bidirectional surge tank and valves exhibits significantly superior effectiveness in safeguarding against transient processes in comparison to conventional single-method approaches.Moreover,the outcomes corroborate the effectiveness of the proposed LMOEA-DS-TOPSIS optimization and decision-making method for coordinated tank-valve control.The research findings not only successfully achieve effective transient protection for water supply systems with branch pipes but also provide an innovative solution for the design and optimization of water hammer protection systems in the field of engineering.关键词
水锤防护/塔阀联合调控/多目标优化/城市供水Key words
water hammer protection/tower-valve joint regulation/multi-objective optimization/urban water supply分类
建筑与水利引用本文复制引用
翟宇,刘小莲,王雪妮,张雷克,伊北,田雨..基于定向抽样多目标进化算法的塔阀联合调控优化[J].清华大学学报（自然科学版）,2026,66(2):309-323,15.基金项目
国家自然科学基金面上项目(52379091) （52379091）
山西省水利技术研究推广补助项目(2024GM21) （2024GM21）
山西省基础研究计划项目(202203021222112) （202203021222112）
