Journal of Hydroinformatics (Jun 2024)

A parallel multi-objective optimization based on adaptive surrogate model for combined operation of multiple hydraulic facilities in water diversion project

  • Xiaolian Liu,
  • Zirong Liu,
  • Xiaopeng Hou,
  • Yu Tian,
  • Xueni Wang,
  • Leike Zhang,
  • Hao Wang

DOI
https://doi.org/10.2166/hydro.2024.285
Journal volume & issue
Vol. 26, no. 6
pp. 1351 – 1369

Abstract

Read online

In a complex pressurized water diversion project (WDP), the combined optimal operation of multiple hydraulic facilities is computationally expensive owing to the requirement of massive mathematical simulation model runs. A parallel multi-objective optimization based on adaptive surrogate model (PMO-ASMO) was proposed in this study to alleviate the computational burden while maintaining its effectiveness. At the simulation level, an adaptive surrogate model was established, while a parallel non-dominated sorting genetic algorithm II (PNSGA-II) was utilized at the optimization level. Taking the successive shutdown of pumps as the operating process, the PMO-ASMO was applied to a complex pressurized diversion section of the Jiaodong WDP in China, and the results were compared with those obtained by NSGA-II and PNSGA-II. The results showed that the time consumption of PMO-ASMO was only 9.97% of that acquired by NSGA-II, which was comparable to that of PNSGA-II, in the case of 10-core parallelism. Moreover, compared with PNSGA-II, PMO-ASMO could find the optimal and stable Pareto front with the same number of simulation model runs, or even fewer runs. These results validated the effectiveness and efficiency of the PMO-ASMO. Therefore, the proposed framework based on multi-objective optimization is efficient for combined optimal operation of multiple hydraulic facilities. HIGHLIGHTS A combined optimal operation of multiple hydraulic facilities within complex pressurized water diversion projects is achieved.; The combined optimal operation model of multiple hydraulic facilities coupled with the simulation model is established.; The algorithm that combines the adaptive surrogate model and parallel computing is provided.; The effectiveness and efficiency of the proposed method are confirmed by an engineering case.;

Keywords