Study On The Grinding Force And Surface Roughness Of Engineering Ceramics Under Ultrasonic Grinding Based On Support Vector Machine

Due to the high brittleness and low fracture toughness of engineering ceramics,it is easy to cause damage,cracks and even brittle fracture on the surface of the workpiece during processing.Ultrasonic grinding is a kind of composite processing technology that improves the surface quality of engineering ceramics.In addition,as a machine learning method,support vector machine can solve the problems such as small sample,nonlinear,high dimensional.On the basis of the existing research results,this paper proposes an intelligent algorithm to optimize the prediction model of support vector machine,and effectively predict the grinding force and surface roughness of the processing under ultrasonic grinding under engineering ceramics.Firstly,the shake performance of the acoustic system is tested and analyzed.Built the test platform and test the performance of ultrasonic grinding with Al₂O₃ ceramic to study the effects of ultrasonic amplitude,grinding depth,grinding wheel speed and workbench speed on the grinding force and surface roughness of Al₂O₃ ceramics.Secondly,the inertia weight,learning factor and random scheme of particle swarm optimization algorithm are optimized based on the parametric simulation experiment.An adaptive particle swarm optimization algorithm is proposed,and it is used to optimize the regression support vector machine to establish the APSO-SVM model.Analyze the performance of kernel function of support vector machine and establish a linear combined hybrid kernel function between RBF kernel function and polynomial kernel function.the selection operator and mutation operator of artificial immune system is improved,and a new artificial immune system particle swarm optimization algorithm is proposed by the parallel mixing between adaptive particle swarm optimization algorithm and improved artificial immune system,and the mixing algorithm is used to optimize the regression support vector machine with mixed kernel function to establish AISPSO-SVM model.Finally,based on the experimental data,the four kinds of process parameters are comprehensively considered,and the three prediction models of tangential grinding force,normal grinding force and surface roughness are established and verified.The prediction ability and stability performance of AISPSO-SVM model is good,and it can effectively explain the relationship between tangential grinding force,normal grinding force and surface roughness and process parameters.