AN INTEGRATED EXPERT SYSTEM FOR THE SELECTION AND OPTIMIZATION OF GRINDING PARAMETERS

Abstract

Grinding is one of the operations widely used in industry to obtain finished products. In our modern society, almost everything that we use has either been machined by grinding at some stages of its production, or has been processed by machines which owe their precision to abrasive operations.

The present research work centres around the development of an intelligent grinding system. A methodology that can shorten, automate and integrate the parameter selection and optimization procedure for the surface grinding process is developed in the thesis. The research focuses on planning an effective production rate, low production cost and fine surface quality process. The main work is the integration of selection and optimization of grinding parameters. An integrated technique is employed. The parameter selection problem is formulated in a hierarchical framework to assist in developing a strategy to solve the problem. The optimization problem has taken into account practical situations using a quadric successive programming method, and an optimization strategy is developed simultaneously.

The research is performed in three modules. The first module, a knowledge-based expert sub-system, is the development of a methodology by which a computer system is able to select an optimal grinding wheel based on work piece specifications and some known operating conditions. The second, a computation-based optimization sub-system, develops a process optimization strategy for the surface grinding process. The third integrates above two subsystems into an integrated expert system which can perform the selection and optimization simultaneously for the grinding process. In order to illustrate the integrated expert system, two case studies, one for rough grinding and another for finish grinding, are presented in the thesis. The results show the integrated expert system proposed in this thesis can be used to select grinding wheels and optimize grinding conditions effectively. An appropriate wheel can be obtained to satisfy a special application in about 4 minutes and optimum grinding conditions can be obtained in Jess than 6 minutes. Therefore, the entire run-time of the system is less than 10 minutes on a PC-386 machine. The case studies also show that the logical steps taken in wheel selection is correct and the mathematical optimization model is superior to the previous models.