The Application Of The Tooth-like Parts Information Input Method Based On Neural Network In CAPP

With the development of the scanning and recognition technologies, it is very common to adopt these technologies. The methods about the description and input of mechanical parts in the Computer Aided Process Planning (CAPP) system are discussed in the paper. So the method for describing mechanical parts which is based on the feature unit is proposed in order that the description for mechanical parts is convenient and completely.Not only the proper method for the preprocessing of image is researched in the paper, but also the algorithms of extraction and calculation about the image features of mechanical parts are proposed. The essence of the neural network utilization for pattern recognition is the changing of the pattern space. Therefore, the constitutions of feature vector are the basis of the using of neural network classifier. The paper considers the whole profile and the internal structure feature, and will be the splendid basis for the intelligent integrated systems.During the process of recognition, a improved algorithm for neural network in Back梡ropagation learning algorithm is provided, and the algorithm is applied to mechanical part images recognition .The result shows that the testing is very good. With the characteristics of the neural network imitating human thinking in images, the functions of studied and memories, the detail classification and location, and then using the method of feature description, the file of technologies can be obtained searchingly. According to the result of parts classification, the system of CAPP can search the characteristic technologies of the part group, and then finish the comparison between the parts and the process technologies.