基于RBF代理模型的调水过程优化研究

调蓄工程通过泵站调水过程中,需要通过合理的控制泵站启闭时间来调节水位,从而不断优化调水过程,但以往的优化方法效率低而且不易得到最优方案。为解决这一问题,本文以南水北调东线山东段南四湖下级湖为研究对象,基于径向基函数(Radial Basis Function,RBF)代理模型建立调水过程优化模型,得到了调水过程方案参数区间内的最优方案,并基于实际调水情况求得不同起调水位下的调水过程最优方案。首先根据调水过程方案参数区间自动选取80个调水过程方案样本,并利用一维二维耦合水动力模型算出每个方案的水位变化过程;其次采用RBF代理模型建立并验证调水过程方案与最高水位、最低水位的响应关系;最后基于RBF代理模型,以泵站工作总时间最短为目标,考虑水量平衡和水位约束建立优化模型,采用粒子群算法求解。研究结果表明,基于RBF代理模型的调水过程最优方案结果与耦合模型计算该方案结果的绝对水深误差不超过0.05 m,相对水深误差不超过0.99%,模型计算精度高。基于RBF代理模型的调水过程优化模型,求解得到调水过程参数区间内的最优方案,解决了传统方法在人为设定有限个方案内得到较优方案的局限性。

Study on optimization of water transfer process based on RBF surrogate model

In the water transfer process of the regulating and storage project through the pump station, the water level needs to be adjusted through reasonable control of the opening and closing time of the pump station, so as to constantly optimize the water transfer process. However, previous optimization methods are inefficient and difficult to obtain an optimal solution. In order to solve this problem, this paper takes the Lower Nansi Lake in Shandong reach of the eastern route of South-to-North Water Transfer Project as the research object,and establishes the optimization model of water transfer process based on RBF (Radial Ba-sis Function) surrogate model to study the optimal water transfer process of the Lower Nansi Lake. The method is used to obtain the optimal scheme within the parameter range of the water transfer process, and the optimal scheme for the water transfer process under different starting water levels is obtained based on the actual water transfer situation. Firstly,80 samples of the water transfer process are selected automatical-ly according to the parameter interval of the water transfer process scheme, and the water level of each scheme is calculated using the 1D-2D coupled hydrodynamic model. Secondly, the RBF surrogate model is adopted to establish and verify the relationship between the water transfer process scheme and the response of the highest and lowest water levels. Finally, based on the RBF surrogate model, taking the shortest working time of the pump station as the objective,the optimization model was established considering water balance and water level constraints, and particle swarm optimization was adopted to solve the problem. The research results show that the absolute depth error is no more than 0.05m and the relative depth error is no more than 0.99% compared with the calculation results of the scheme water level and the coupled mod-el in the optimized water transfer process, and the optimization model of water transfer process based on RBF surrogate model has high accuracy. Based on this model,the optimal solution within the parameter in-terval of the water transfer process is obtained, which solves the limitation of the traditional method to ob-tain the optimal solution within a limited number of artificially set schemes.