Research On Milling Test And Parameter Optimization Of Variable Section Scroll

In order to meet the requirements of efficient,energy-saving and reliable use of scroll compressors,the design of the scroll line is often a combination of several curves,which makes the variable-section scroll become a thick-walled deep cavity part,and its milling Processing is more difficult than an equal-section scroll.Under the conditions determined by the processing equipment,whether the selection of milling parameters is reasonable is the key factor determining the quality of the scroll processing.In this study,the three-segment base involute is selected as the profile of the variable-section scroll,and under the condition of orthogonal milling test of variable-section scroll,the multi-objective parameter optimization based on milling force,tooth surface roughness and milling time was studied.(1)Data processing of three-way milling force of F_x,F_y and F_z.In order to deal with hundreds of thousands of data measured and recorded by the three-way piezoelectric dynamometer during variable-section scroll milling,this paper proposes a"fuzzy judgment method"—using the analysis of variance in SPSS software to screen out the variance within 6The data is calculated for the average milling force under orthogonal milling parameters.(2)Based on the empirical formula,an improved BP neural network prediction model of milling force was established.The influence of milling parameters on F_x,F_y and F_z was obtained by single factor response prediction.Based on the processed milling force data,a multivariate nonlinear regression prediction model for milling force and an improved BP neural network prediction model for milling force are established.It is proved that the improved milling force BP neural network model has higher prediction accuracy.Based on the prediction value of the milling force regression model,the improved milling force BP neural network model is used to predict the single factor response of each milling factor,which intuitively reflects the influence of each milling parameter on F_x,F_y and F_z.(3)A double prediction model of the tooth surface roughness of the variable-section scroll is established,and the influence degree of each milling factor on the tooth surface roughness is accurately obtained.Firstly,the multi-linear nonlinear regression model of the tooth surface roughness and the improved tooth surface roughness BP neural network prediction model are established by using the milling parameters and the corresponding roughness values under the orthogonal milling test conditions of the variable-section scroll.Combining the advantages of the two models,a bi-predictive model of the tooth surface roughness of the variable-section scroll is established.Then,it is proved by experiments that the double prediction model of roughness is more accurate than the single prediction model,and the discrete error of the single prediction model can be avoided.Finally,the single-factor response prediction was performed using the double-predictive model of the tooth surface roughness,and the mapping relationship between the tooth surface roughness and the milling parameters was visually expressed.(4)With the range of each milling parameter as the constraint,the multi-objective optimization model is established with the minimum milling force,the tooth surface roughness not more than 0.8μm and the shortest milling time.The genetic algorithm is used to optimize the multi-objective constrained space and the optimal combination of specific milling parameters is obtained.when a_p=1.65mm,f_z=0.18mm,n=3757r/min,a_e=0.74mm.the optimum milling effect is F_H=186.27N,R_a=0.78μm,t_m=17.37s.