Structural Optimization Of Double-row Angular Contact Bearings Based On Ranking Differential Evolution Algorithm

Rolling bearings are widely used in various industrial fields such as aviation,automobiles,railways,machinery and instrumentation.Due to the wide range of applications of bearings,improving bearing capacity is of great significance to bearing optimization.The design of rolling bearings must meet the constraints of geometry,kinematics and strength to have excellent performance,long life and high reliability.This requires an optimal design method to achieve these goals together.In order to obtain the best rated dynamic load,rated static load and oil film thickness of double row angular contact bearings.Based on this,a mathematical model of three objective functions is established for the bearing structure.Taking 17 sets of double-row angular contact bearings as the optimization object,the rated dynamic load,static load rating and oil film thickness of the double-row angular contact bearings are taken as the three objective functions.The pitch circle diameter of the double-row angular contact bearings and the ball diameter of rolling elements,The number of rolling elements and the ratio of the radius of curvature of the inner and outer ring channels to the spherical diameter are used as the decision vector.Single-objective and multi-objective evolutionary algorithms are used to optimize the bearing structure.In recent years,the differential evolution algorithm(DE)has been widely used to solve many practical problems.However,the differential evolution algorithm may encounter stagnation during the iteration process.Therefore,a reinforcement learning method is used to improve the search ability of DE,and a new algorithm is proposed,named RUSDE.The top-ranked individuals are selected and stored in an archive.Compared with individuals with updated fitness,this choice is more stringent;by simulating human social behavior,different mutation strategies are used.In this learning strategy,choose according to the update situation of the parent.In this method,this paper uses the CEC2014 benchmark test,including single-peak,basic multi-peak,extended multi-peak and mixed problems to test the performance of RUSDE,and discusses the impact of file capacity.The results show that when the file capacity is half of the total population Time,you can get better results.The performance of RUSDE is compared with DErand,j DE,Sa DE,Rankj DE,SPS-j DE,j DE-EIG,KH,LBSA,DGSTLBO,and SCA.The results and statistical analysis show that RUSDE is superior to DErand,j DE,Rank-in30-dimensional problems.j DE,j DE-EIG,KH,LBSA,DGSTLBO and SCA are slightly better than SPSj DE and Sa DE.When dealing with 50-dimensional problems,the performance of RUSDE is slightly better than SPS-j DE and better than the other nine algorithms.Theoptimization results show that the bearing capacity of the bearing is improved by more than60% ? 120% after the optimization algorithm is optimized compared with the bearing without optimization.