Roughness And Wear Prediction Of Honing Cylinder Liner And Reliability Evaluation

The internal combustion engines(i.e.,=CEs)have the characteristics of high efficient energy conversion and strong power output,more attention has been paid to ICEs by transportation companies,the defense and military departments as well as the energy power departme(ts in various countries.This thesis takes the cylinder liner of high performance internal combustion engines as the object,and focuses on prediction of the Rk roughness set for honed cylinder liner surface,optimization of honing processiog parameters,the wear prediction of the cylinder liner and the reliability evaluation of the cylinder liner in service.Aiming to provide some references for improving the precision machining of key components and service performance of the whole engine of the high performance internal combustion engines.The main contents are listed as follows:(1)The back propagation neural network(i.e.,BPNN)and generalized regression neural network(i.e.,GRNN)prediction models are established for honed surface roughness Rk,Rpk and Rvk of cylinder liner based on artificial neural network(i.e.,ANN).A three-factor,three-level full factorial experiment is designed for training and validating the models,and the smoothness factor of the GRNN prediction model is optimized.The results show that the mean of the determination coefficient of the GRNN prediction model is 0.959,and the mean of the determination coefficient of the BPNN prediction model is 0.829,the GRNN prediction model has high prediction precision for roughness.The feasibility and effectiveness of GRNN roughness prediction model are validated.(2)Response regression models are respectively established for roughness Rk,Rpk and Rvk,based on response surface methodology(i.e.,RSM).Taking three roughness values to be relative minimum simultaneously as the optimization objective,the honing processing parameters are optimized by elitist non-dominated sorting genetic algorithm(i.e.,NSGA-?)multi-objective optimization algorithm,and Pareto fronts of roughness are obtained.The results show that the roughness multiple regression models have good fitting effects based on RSM.The order of significance of the influence of honing parameters on roughness is as follows:for Rk:Vro>P>Vre,for Rpk:P>Vro>Vre,for Rvk:P>Vro>Vre(rotational speed of honing head:Vro,reciprocating speed of honing head:Vre,honing pressure:P).(3)Based on the grey system theory,a grey prediction model GM(1,1)for cylinder liner wear is established,and the weight is optimized by data-driven method.The results show that the average relative error between the predicted values and the measured values is 4.8%,the maximum relative error is 7.1%,the model has a high prediction accuracy.The GM(1,1)with optimal weight has less prediction error and the prediction accuracy is increased by 44.2%.(4)The service reliability evaluation models of cylinder liner considering different factors are established based on the Wiener process,and the dynamic evolution of service reliability is analyzed by different models.The results show that the fuzzy reliability curve based on the Wiener process is steeper than the reliability curve of the Wiener degradation process considering the randomness of the drift coefficient,and much smoother than the reliability curve of the Wiener degradation process without considering the randomness of the drift coefficient.