Experimental Study On Magnetic Abrasive Finishing Based On Parameter Optimization

316L stainless steel slender tube has good fatigue resistance,corrosion resistance and high surface finish,which is commonly used in mechanical equipment transmission system.When conveying high pressure fluid,the quality of the inner wall of the slender tube directly affects the normal operation of the mechanical parts.Due to the restriction of narrow tube diameter,it is difficult to effectively finishing the inner wall by ordinary finishing.Magnetic abrasive finishing has good profiling and flexibility.The magnetic pole is used to generate a magnetic field,absorb the magnetic abrasive inside the tube fitting,form a flexible abrasive brush,and finishing the inner wall of the tube fitting,so as to remove the inner wall defects and improve the quality of the inner surface of the tube fitting.In the existing magnetic abrasive finishing,the finishing process parameters are often set based on experience,which makes it difficult to determine the best process parameters,and also can not effectively predict the workpiece surface roughness after the test.In order to solve the above problems,in the process of magnetic abrasive finishing316 L stainless steel slender tubes,the main process parameters that affect the surface quality and finishing efficiency are optimized and analyzed,and the influence of different process parameter combinations on the surface quality is studied.Using the orthogonal test data,in which the spindle speed,abrasive particle size,feed speed,and processing time are independent variables,and the surface roughness value is dependent variable,a multivariate nonlinear regression,support vector machine(SVM),and extreme learning machine(ELM)surface roughness prediction model is established.Through mathematical model calculation,the extreme learning machine model has the highest prediction accuracy.Aiming at the randomization characteristic of the parameters of the extreme learning machine,the parameters of the extreme learning machine are optimized by using particle swarm optimization(PSO)algorithm,and the PSO-ELM prediction model is established.The machine learning performance evaluation indexes(goodness of fit,mean absolute error,root mean square error)are used to verify the PSO-ELM mathematical model,and the prediction accuracy meets the requirements after calculation.The particle swarm optimization algorithm is applied to search for the best process parameters of magnetic abrasive finishing 316 L stainless steel slender tube,and the best combination of process parameters is used for finishing.The internal surface quality of tube fittings before and after finishing is compared.The test results show that the predicted results of PSO-ELM mathematical model are consistent with the actual results,which has reference and guidance significance for magnetic abrasive finishing.