Gradient-Free Optimization For Quality Control In Batch Production Of Medium Voltage Insulation Parts

Medium voltage insulation parts are an important part of electrical equipment,its quality determines the life and insulation performance of electrical equipment.Therefore,the quality control of medium voltage insulation parts is the key production process of medium voltage insulation parts.In the production process of medium voltage insulation parts,many production process parameters(such as melt temperature,holding pressure,holding time,etc.)affect the final quality of medium voltage insulation parts,so in order to improve the product quality,it is necessary to determine the optimal process parameters.In the industrial field,the trial-and-error method and the experimental design method are two traditional methods to determine the optimal process parameters of products,but these two methods have high test cost and low efficiency.In order to solve the above problems and improve the quality of medium voltage insulating parts,a product quality control method based on improved gradient-free optimization is proposed in this paper.Based on the above optimization method,a systematic quality control framework for insulating parts is formed:Firstly,A gradient-free optimization method is proposed for the intermittent production of medium voltage insulating parts,which is nonlinear and not suitable for gradient optimization.Secondly,an optimization method without gradient is proposed to solve the problem of high nonlinearity and inapplicable to gradient optimization in batch production of medium voltage insulation parts.Thirdly,gradient-free optimization method and the parallel perturbation stochastic approximation optimization method,combining with the knowledge to guide mechanism,this paper using historical information such as history,gradient approximation iteration points,three different kinds of improved method is proposed,and the improved optimization method is verified by test functions compared with the traditional optimization method and less iterative times better performance characteristics.Fourthly,for gradient-free optimization method of the simplex optimization method,combining with the knowledge to guide mechanism,this paper,by using maximum points in the process of optimization,center of gravity point,such as history information,this paper proposes a new simplex method,and improved optimization method is verified by test functions into less iterative times compared with the traditional optimization method,the characteristics of better performance.Fifthly,using the wireless acquisition method,product production process parameters were collected to build insulator 3 d model,based on this model,target insulator by Moldflow simulation and the simulation data used in the model of BP neural network to generate insulator agent,to verify the improvement on the model without gradient optimization method for quality control of efficiency is better.Sixthly,based on Moldflow,the 3D model simulation of the target insulator further verified that compared with the traditional gradient-free optimization method,the improved gradient-free optimization method has lower cost and better effect on the quality control of the target insulator.At last,the prospect of improving the quality control method without gradient in the batch production of medium voltage insulation parts is put forward.