At present, most of the three-dimensional image processing field utilizes the nonlinear mapping characteristics of neural networks to transform three-dimensional images into two-dimensional images for processing. The bus body is mainly composed of a front wall, a rear wall, and a side wall. Regardless of the manufacturing process or the visual experience, the three parts are relatively independent, and are approximately orthogonal in geometry. Therefore, the quasi three-dimensional model of the bus body composed of the three-dimensional two-dimensional graphics is practical. In the two-dimensional modeling evaluation, the plane wireframe view of the car model is used as the input sample of the neural network, and the expert's modeling evaluation of the sample is used as the ideal output of the neural network to train the neural network. The key technology of bus body shape evaluation The development of the entire modeling evaluation expert system includes six stages: collecting data, preprocessing graphics, building a knowledge base, designing a neural network model, training neural networks, and developing a software interface. The edge is relatively stable information in the image, and is relatively less affected by external conditions. Therefore, the scanning pattern of the sample data needs to be preprocessed, the main feature lines are extracted, and the wireframe view is formed as the input data of the neural network. In this paper, the method of edge extraction based on the principle of side inhibition competition is used to extract the edge of the image. It not only can effectively extract the edge of the image, but also does not shift the extracted edge. It overcomes the disadvantages of some commonly used methods for extracting image edge features, and the algorithm is simple and easy to extract in real time. Construction of Knowledge Base The neural network knowledge base acquires empirical knowledge of the evaluation of automotive body shapes from specific sample evaluations by domain experts and stores these knowledge in neural networks. The basis of the bus body modelling evaluation neural network activity is to have a good bus body modelling and expert rating sample set. Specifically, how to evaluate a certain vehicle model and adopt a general scoring method, for example, scoring a representative number of training samples according to an evaluation team composed of experienced experts, and taking a weighted average of the scores of each sample as a neural network. The ideal output. Through a large number of researches and references to domestic and foreign passenger car modelling literature, 20 representative passenger cars worldwide were selected as the sample database for this development. At the same time, after consultation with the coach modeling experts, an inquiry score sheet for the body shape development sample database was specifically designed. The inquiry form provides 3 independent 2D views of the front, rear, and side of each sample in the sample library and the 3D projection view of the vehicle as the basis for expert scoring. Under the supervision of the neural network, the relationship between the 3 independent 2D views of the front, rear, and side and the vehicle modeling score was studied. Afterwards, a trained quasi three-dimensional modeling evaluation neural network can be used to predict and evaluate the complete vehicle model of any vehicle type. Based on the knowledge base training neural network in order to ensure that the design of the network is not sensitive to noise, in the sample processing process, artificially add certain interference, so that the samples that are superimposed on different interference sources have become new training samples. Through this method, not only can improve the robustness of the software, but also increase the number of training samples, which can ensure that when the neural network learns to discern vehicle data with noisy signals, it also correctly identifies the original body shape. Application of Bus Shape Evaluation Software When a large passenger vehicle is undergoing a retrofit design, corporate designers focus on improving the grille shape of the engine vent. The related developers applied the constructed neural network-based styling evaluation expert system to evaluate the shape of the two bus bodies before and after the modification, and the cross-sectional view of the scoring interface for the expert system. Finally, through comprehensive comparison and analysis, the company confirmed the improved new model and put it into actual production. After the production, the market responded well. In this paper, the ANN technology is applied to the evaluation of bus body shape modeling, and an intelligent bus body shape evaluation expert system is researched and developed. And it is applied to the actual vehicle development task of the production enterprise and achieved satisfactory results. This shows that the ANN-based bus body shape evaluation expert system is feasible. Rain Boots Injection Machine,Rain Boots Injection Machine,Rain Boots Injection Molding Machine Fenghua Jade Motor Co., Ltd. , http://www.sole-injection-molding-machine.com