Research On Robotic Spraying Process And Film Thick Prediction Of High-speed Rail Body

High-speed railway coating is an important technology in the production of high-speed railway.The quality of surface coating determines the life and performance of high-speed railway car body.Due to the large volume of the car body itself and high requirements for spraying quality,the current domestic body in white spraying workshop is mainly completed by manual,in the spraying process and spraying quality control also mostly depends on the experience and proficiency of workers.Because of its low cost and high flexibility,robot spraying is gradually favored by major manufacturers.However,the immature research on the spraying process leads to the difficulty in controlling the quality of spraying,which is also the reason that hinders its wide application.Therefore,in order to improve the spraying efficiency and achieve accurate control of the spraying process,based on the traditional thickness modeling method,this paper considers the influence of process parameters,establishes a multivariable coating thickness accumulation model and studies the variation rule of each parameter and the film thickness.Furthermore,the thickness prediction model was established by machine learning,which realized the accurate prediction of the film thickness during the spraying process.The following is the research work of this paper:A multi-variable coating cumulative thickness model based on elliptic double Gaussian sum model is established.Based on the traditional static thickness distribution model without parameters,the cumulative coating thickness models with single and multi-channel trajectories are studied.By analyzing the common spraying process methods and influencing factors of spraying quality,the multi-parameter spraying process modeling was carried out,and the corresponding laws of influencing factors and film thickness were fitted according to the experimental data.The experimental results display that the proposed model can well reflect the distribution of film thickness and has a certain universality.A method for predicting the thickness of robot spraying film based on machine learning algorithm was proposed.K-fold cross validation is used to optimize the hyperparameter of the model,and the optimized model is used for training.In order to verify the accuracy of film thickness prediction,the prediction effects of GBDT,XGBoost,LightGBM,SVM and other algorithms were compared and analyzed.The results show that the predicted value of XGBR model is basically consistent with the real value,and its explained variance,mean absolute percentage error and the r2?score are 0.99,0.02 and 0.98,respectively.It can meet the prediction requirement of film thickness.Set up the robot spraying technology experiment platform,the spraying process test scheme design,and the spray process test scheme was designed,and the parameters in the proposed elliptical double Gaussian sum models and the parameters in the multi-parameter cumulative coating thickness distribution model were solved,and then by dynamic spraying and static spraying experiments verify the accuracy of the request parameters,finally,the spraying distance has been optimized,to ensure the uniformity and consistency of spraying coating requirements.