Research On Generative Design Method Of Vehicle Body Structure

The vehicle body structure is directly related to the low-order modal characteristics and stiffness characteristics of the vehicle body,and the two are related to the safety,comfort and NVH performance of the car.The traditional car body static performance optimization design method requires engineers and technicians to have A wealth of professional domain knowledge and experience,but also requires the use of commonly used experimental design,surrogate model,optimization algorithm and other optimization design tools and the establishment of methods,however,with the continuous improvement of the level of automotive design and manufacturing,people’s requirements for automotive quality continue to improve,Traditional optimization design methods are increasingly difficult to adapt to the ever-shrinking automotive development cycle.At the same time,current computer-aided design systems are still difficult to achieve true innovation in the product design process.Car body structure design requires new design methods to meet current production development demand.Studying the generative design method of the body structure is of great significance for improving the efficiency of the body structure design.Based on the traditional optimization design method,combined with the current rapid development of artificial neural network technology and case-based reasoning technology,particle swarm optimization algorithm,this paper studies the car body structure generative design method,aiming at the first-order torsion mode,The door frame torsion mode,the first-order bending mode,the front cabin yaw mode,the bending stiffness and the torsional stiffness,generative design was used to design the relevant body parameter that can meet the preset target performance and finally obtain the relevant car body parameters.The main research work of the thesis is as follows:(1)Aiming at the design of four low-order modal frequencies,bending stiffness and torsional stiffness of automobile white body,a generative design method of vehicle body structure parameter is proposed.This method is based on case-based reasoning technology and uses particle swarm algorithm as a case in case-based reasoning system modification method to generate a variety of solutions that meet the low-order modal frequencies and stiffness values of the preset value.(2)In view of the shortcomings of conventional particle swarm optimization algorithm in the face of high dimensional problems,this paper proposes a dynamic boundary particle swarm optimization algorithm based on human-computer interaction.The algorithm monitors the historical optimal position of particles during the algorithm iteration by human-computer interaction,dynamically modifies the search space,and faces the high-dimensional design problem with more feature items.The improved algorithm can be relatively obtain a better solution.(3)According to the researched building structure generation design method,combined with case-based reasoning technology,improved particle swarm optimization algorithm and TensorFlow neural network construction platform,develop the body structure generation design system,which can adapt to different dimensional sample data,quickly Build and train a qualified neural network model.Using this system,the 100-thickness and size parameters of an electric vehicle body-in-white are designed and generated,and the vehicle body parameters satisfying the preset body mode and stiffness error requirements are obtained.The results show that the researched body structure generative design method reduces the dependence of design experience body structure design.Improve the efficiency of the body structure design,and the system developed has practicality.