Study On Improved Shaft Parts CAPP Expert System

Computer Aided Process Planning(CAPP) is a way to realize intelligent process design and automation, with the rapid development of the manufacture, process design of intelligence and information have become increasingly demanding. But the traditional CAPP expert system have some problems, for example: part information extraction incompletely, knowledge acquisition difficulty, matching conflict and poor reasoning accuracy, it is difficult to meet the need of modern manufacturing industry already. The paper in order to solve the above problems, we use feature-based modeling technology and artificial neural network technology to improve the traditional expert system. In the paper, the main contents include three modules: part information extraction, knowledge base design and process reasoning, we explore and research the modules in the following parts.Firstly, The work is analysis the advantages and disadvantages of traditional expert system and neural network, compare the two techniques. And build the overall frame of CAPP expert system based on neural networks. The implementation methods and related technologies are introduced and studied.Secondly, the paper researched shaft parts information extraction module, using feature and coding techniques to build the coding rules suitable for the system, and solving the problem of incomplete part information extraction in a way to build feature matrix.Then, the paper researched knowledge base design, using the IF…THEN…rules and neural network weight values to express system technology knowledge, and stored in the SQL Sever database.Again, the paper researched process reasoning, we build a hybrid reasoning mechanism based on neural network and expert system to solve the problem of matching conflicts and inference efficiency. And we detailed study expert system reasoning strategy, neural network architecture design, learning algorithm, sample selection, training issues and so on.Finally, We put the Visual Studio 2008 as the development tool, using Matlab to achieve neural network inference part. And we use VB as the development language to design work interface, combined with process reasoning instance and certified the above modules, finally proved the validity of the research method.