Research And Development Of Neural Net-Based Intelligent Optimization System Of Metal Turning Parameters

Tool cutting is a main processing method in machine-building industry. Improving the tool cutting efficiency and reducing the machining cost have been an important task, the study of which the experts in machining field are devoting their efforts to. The cutting data are the basic factor to measure the level of cutting technology, so using the optimized cutting data is a basic way to make the most of machine functions and to improve the cutting efficiency.In the traditional cutting process, the determination of cutting parameters is dependent mainly on the experience of the technologists or process handbooks, which is difficult to meet the requirement of machining accuracy, and the time cost by the determination of parameters is too long. In this paper, the database technology and the artificial neural network technology are introduced into the optimization of cutting parameters, and the intelligent optimization system of metal cutting parameters in built. This paper mainly completed the following studies:1. By referring to a lot of literatures, the current status and development of the cutting database at home and abroad are introduced, and its existing problems and development trends in the future are analyzed.2. The MySQL database platform-based basic information library, learning sample library and awaiting evaluation data library are set up, and the efficient information management of the cutters, materials, machine tools and data of cutting parameters is realized. Besides, the bottom layer data support is also given to the neural network module and the module of information management system.3. The design layout of neural network is provided. It is divided into two parts: rough machining and finish machining, and the rough machining network and finish machining network are constructed separately. The weighted values of networks are saved according to different types of machine tool4. The network models are set up in NeuroSolutions. Network parameters such as the number of optimal hidden-Layer nodes, the optimal step size and optimal momentum term are determined by means of the analysis functions of NeuroSolutions. And the DLL (dynamic Link Library) files for information management system are generated.5. The object-oriented programming method is used, and the Windows-based information management system is designed by use of OLE (object Linking and embedding technology) to provide a friendly, fast and easy-to-operate man-machine interface. At the end of the paper, the Operation process of the system is shown by means of a practical example.Through the above work the following conclusions can be show. Introducing the neural network technology and the database technology into the cutting parameters optimization process is very suitable way at first, and secondly, every parameter of BP network should be optimized, and the precision of optimized BP network can make a considerable progress. Thirdly, a new sample is utilized to test the network for verifying the predictive function of the model. The test results are reasonable.