Research And Application Of Optimization Method Based On Surrogate Model

In complex engineering problems,it is often faced with the difficulty of obtaining the performance evaluation function and the huge amount of calculation.Using intelligent algorithms to directly optimize the performance function or high-precision model requires a large number of samples,which is extremely time consuming.The surrogate based optimization method provides a solution to solve such problems.In addition,there is uncertainty in the surrogate model.Robust optimization design aims to reduce the uncertainty in the optimization design.Therefore,when the surrogate model is used to simplify the calculation in the robust optimization design,the quantification of the surrogate model uncertainty exists for the robust optimization design will affect the robust of the final solution.This paper studies the method and theory of the surrogate model from the following perspectives.The main contents include:(1)Research on the reduction of the number of samples in the sequence iterative optimization based on the adaptive surrogate model.The research focus on the relationship between the surrogate model sampling method and the optimization process.The complex method-based adaptive surrogate model optimization method is proposed.The method combines the information of samples and the optimization process.Through the classic benchmark functions,the proposed method is compared with the Expected Improvement(EI),the Minimum of the Respond Surface(SURF),and the Candidate point approach(CAND).The results show that the proposed method can obtain the multiple global and local optimal solutions and higher accuracy,thus effectively improving the calculation efficiency.Taking an electric vehicle battery pack as an optimization example,the results show that the proposed method can greatly reduce the number of calls of the simulation model and effectively improve the optimization efficiency.(2)In order to improve the application efficiency of surrogate model,the relationship between swarm intelligence algorithm and surrogate model samples.The swarm intelligence algorithm needs a large number of calls to the original function during the solution process.The application of surrogate model can effectively reduce the number of such calls.However,the surrogate model and swarm intelligence algorithm are only two independent tools to solve the problem in the existing research.This paper combines the surrogate model with the crow search algorithm,and proposes an improved crow search algorithm based on the surrogate model.Through test functions,the proposed method is compared with classic algorithms.The results show that the algorithm proposed in this paper can find the global optimum with fewer samples,and is beneficial to improve the calculation accuracy and calculation efficiency.(3)The robust optimization design of the drive arm seat of RB-10-001-axis robotic is carried out in consideration of the uncertainties in surrogate model and design variables.Existing researches rarely quantify the uncertainties of surrogate models and design variables in robust design at the same time.This paper takes RB-10-001 axis robot drive arm as the research object,considering these two kinds of uncertainties at the same time,and carries out multi-objective robust optimization design of robot drive arm.By comparing with the traditional robust design,the results show that considering the uncertainties of the surrogate model and design variables can improve reliability and robustness of the drive arm seat.