| In practical applications,time-consuming computational optimization problems are common,such as computational fluid dynamics simulation,chemical distillation unit design,etc.For this reason,time-consuming evolutionary optimization has become a hot topic.The use of surrogate models to replace process simulation in complex industrial processes or chemical product design has attracted extensive attention from researchers and related companies.As more and more researches are carried out,researchers find that the surrogate model is suitable for solving nonlinear and complex optimization problems with high dimension of decision variables and small amount of sample data.The surrogate model is used to replace the real target model,and the fitness evaluation step of the evolutionary algorithm is optimized,which saves the calculation time and achieves the goal of improving the efficiency of the optimal design.However,it is still a great challenge to solve the complex optimization design problem with the surrogate-based evolutionary algorithm.For example,how to determine the number of surrogate models so that the algorithm can achieve a balance between prediction accuracy and calculation time,how to select appropriate surrogate models for different types of optimization problems,and what model methods are adopted to build agents,etc.Moreover,most researchers focus on the complex optimization of the objective function,without considering the complex situation of the constraint function.At the same time,the sampling accuracy of the surrogate model and the selection of the meta-surrogate model will also affect the optimization efficiency.These are topics that require further study.In this regard,based on the previous research work,the first part of this paper proposes a differential evolution constrained optimization algorithm based on the surrogate model for the time consuming computational objective optimization problem with constraints.The algorithm mainly deals with the complex single objective optimization problem with multiple constraints,establishes the static surrogate model for the complex objective function,and combines the information of the objective function with the feasibility law to accelerate the convergence speed of the algorithm.In the second part,aiming at the problem that the static surrogate model is not easy to balance the prediction accuracy and calculation time,a management strategy of the surrogate model is introduced,which greatly improves the optimization efficiency of the surrogate model.Finally,this algorithm is applied to the optimization design of chemical distillation unit.The experimental results of the test function show that surrogate-assisted constrained evolutionary optimization algorithm proposed in this paper can reduce the number of calls of the original model and so as to improve the evolutionary optimization efficiency of the timeconsuming process system.Finally,the approach is applied to the operation optimization design of the distillation unit,and the results show that the approach can meet the production conditions of the crude oil distillation unit. |
