Optimization of water hammer protection against sudden pump stoppage using machine learning models and intelligent algorithms

作者：Hongyan Li, Yu Zhang, Jianguo Cui, Feng Zhang, Xiaoyu Yang, Wentao Shi, Libo Mao · 发表于：Journal of Hydraulic Research · 年份：2025 · DOI：10.1080/00221686.2024.2446578 · 被引用次数：10 · 研究领域：Water Systems and Optimization、Oil and Gas Production Techniques、Water Quality Monitoring Technologies

Pump-stoppage-induced water hammer is very harmful, and there is usually a requirement for protective equipment to be installed. In order to guarantee the safety of pipelines while reducing engineering cost, a method was proposed to simulate hydraulic transients by using back propagation neural network (BPNN), random forest (RF) and support vector machine (SVM), while optimizing with the multi-objective particle swarm optimization (MOPSO) algorithm and making decisions with entropy weight-technique for order preference by similarity to an ideal solution (EW-TOPSIS). The machine learning model was used to fit the mapping associations between protective equipment parameters and optimization objectives. Taking the maximum and minimum water hammer pressure and costs as the objective functions, a multi-objective optimization model was constructed to determine the effective linkage between the hydraulic transient and the intelligent optimization algorithm, and a scheme with comprehensive benefits was obtained through optimization and decision-making. The case study showed that the proposed scheme can greatly improve the protection effect and reduce costs.