Conflict Resolution Of Car Door Collaborative Design Based On Machine Learning

Nowadays,the complexity of industrial products and the scale of production are increasing.The design of a product usually requires the cooperation of different people or different teams.The same is true for automotive industry.The structure of collaborative design environment,the method of information transmission and the resolution of conflict will greatly affect the capabilities of independent research and development in automotive industry.The resolution of conflicts has always been the key in collaborative design.In recent years,many researchers have proposed different algorithms for conflict resolution,mainly based on the detection of conflicts or the avoidance of errors(locking),all of which have their shortcomings.These methods will decrease the efficiency of collaborative design.With the improvement of information technology and computer technology,AI has become ubiquitous.The study of machine learning has also become more and more mature.The cost of learning of different algorithms and building custom frameworks are decreasing.The demand for AI is becoming more and more urgent in automotive industry.A large number of manual operations are required in the traditional design process of the car door,especially in the design of the door hinge and the main section of the door.This requires designers to have a wealth of experience,which will decrease research efficiency.Therefore,it is necessary to propose a new design method to solve this problem.In view of all the above problems,this paper provides a method based on machine learning for car door collaborative design conflict resolution,and finally the following work is done in this paper:1.Based on COMX platform,a parameterized collaborative design framework is realized.Combined NX Open development and NX/Spreadsheet to realize the parameter transfer in collaborative design.2.By introducing the design process of car door,a design method based on parameter driven of door hinge and door main section is proposed to improve the efficiency of the door design.When conflict occurs in the design process,the algorithm is established based on the design principle,and parameter database is designed for machine learning training.3.The BP neural network is built in MATLAB,and the conflicting parameters in the door design process are built into the database for training.When a conflict occurs,a new parameter scheme can be quickly predicted under existing data driven.This method reduces the repeatability and exploratory,which greatly improves the quality and efficiency of the car door design.