Research On Digital Optimization Design Of Automobile Based On User Image

In order to achieve the strategic goal of a strong manufacturing country,China is taking the "Made in China 2025" action program as the strategic support and guarantee.And The "three steps" is made to achieve the strategic goal of a strong manufacturing country,to transformat and upgrade the manufacturing industry.The automobile industry is the benchmark of the national manufacturing industry.With the increasing demand for energy conservation and environmental protection,new energy vehicles are developing rapidly.In order to better meet the psychological needs of users,the research on automobile design methods has become a hot spot in the industry.Taking the design of new energy SUV model as an example,this paper extracts vehicle modeling elements through various digital modeling methods,based on user image,and uses multi-evaluation method,multi-objective evolution algorithm and computeraided design software to optimize vehicle digital modeling.Firstly,to collect the initial sample of the new energy SUV model and disassemble the model by the combination of “characteristic surface” and “characteristic line” by morphological analysis method,and decompose the modeling part from four.Dimensions detailed statistical sample characteristics,and cluster analysis using SPSS software to obtain representative modeling samples.After the investigation and screening the initial image vocabulary of the model,they are paired into mutually exclusive adjective pairs by semantic difference method.The factor analysis method is used to extract the principal component factors to form representative image adjective pairs.Secondly,based on the quantitative I-class theory,experts are invited to evaluate the semantic matching degree of representative image semantic adjective pairs and typical SUV samples.The statistical results are processed by triangular fuzzy number optimization.The imaged adjective scores of the typical samples and the codes of the feature modeling elements decided by switch method are unified into a comprehensive perceptual evaluation matrix.In the use of MATLAB and SPSS software,the partial correlation coefficient and linear model of the vehicle modeling parts and representative adjective pairs have been obtained.The scoring relationship between modeling elements and image semantics have also been concluded.Thirdly,through genetic algorithm with non-dominated sorting of elite strategy genetic algorithm(NSGA-II),the morphological optimization design of modeling elements and multiple image semantics is carried out.Based on the "feature points",the key feature points are combined.The mapping relationship between the modeling feature points and the image vocabulary is obtained based on the BP neural network.The vehicle body type combination with the largest contribution value obtained in Chapter 4 is taken as the initial sample,and the multi-target image is genetically encoded and decoded according to the NSGA-II algorithm.The result is an optimal evolutionary styling solution for the front,body and rear.Finally,the evolutionary optimal schemes are combined to evaluate the TOPSIS vehicle modeling scheme.After the expert evaluation,the evaluation matrix is dimensionless,normalized and deblurred.The optimal scheme is digitally constructed model display.The evaluation results prove the effectiveness of the digital design optimization effect based on the user image.The styling optimization method studied in this paper is applicable to the innovative design of various products.It will improve the original methods in process design,user feedback and optimization design,and further enhance product competitiveness and market share.