Collaborative Design Conflict Resolution Research Base On CAE Data-driven Under Tensor Flow

Under the development trend of the era of cloud computing,the structure of the automobile enterprise collaborative design environment,ways of information transmission and conflict resolution methods can largely affect the product development cycle and are important factors that can't be neglected in the independent research and development ability of automobile enterprises.At present,there are various methods to resolve conflicts.Some scholars choose negative methods to avoid conflicts.However,due to the independent and interdependent relationship between designers,conflicts in the initial design scheme are often unavoidable.When conflicts are inevitable,the efficiency of arbitration is often not satisfactory,and method of optimization algorithm means a lot of calculation.With the advent of the era of artificial intelligence,the study of machine learning algorithms is becoming more and more mature.Either in the part of designing or applying,the demand of intelligence is becoming more urgent in automobile industry.At the same time,open source of various machine learning frameworks,learning related algorithms and building custom frameworks are becoming cheaper.Based on the messaging principle of collaborative design,the server has the ability to collect a large amount of modeling data,and the modeling data of the final design contains product-related constraints and CAE information.In the current background where automotive lightweight research is popular,the design of structural parts is particularly important.In this paper,considering that the subjective judgment and experience differences can make conflict CAD design scheme to be diverse and complex.The application of CAE data drive and neural network is combined,and finally the following research work is done in this paper.1.Based on the collaborative parameterization design of COMX platform,the collaborative environment,data generation and transmission methods are determined by creating custom components for the application of NX Open API.2.Through the ladder shaft design example,the different conflict parameters from designing experience are simulated.Then use CAE multi-objective optimization results as the final results to design the corresponding database which is used for machine learning model of training and conflict resolution.3.Through the open source framework TensorFlow structures,the BP neural network is used to study and prediction of data.Comparing with the predictions of test examples,the4.results show that this kind of prediction method based on CAE data-driven has reference value to the designer to resolve conflicts and illuminates the efficiency of traditional methods have greatly improved.