The fuzzy-nets based approach in predicting the cutting power of end milling operations

Chang, Chuan-Teh
Major Professor
Joseph C. Chen
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Journal Issue
Industrial Education and Technology

Process planning is a major determinant of manufacturing cost. The selection of machining parameters is an important element of process planning. The development of a utility to show the cutting power on-line would be helpful to programmers and process planners in selecting machining parameters. The relationship between the cutting power and the machining parameters is nonlinear. Presently there is no accurate or simple algorithm to calculate the required cutting power for a selected set of parameters. Although machining data handbooks, machinability data systems, and machining databases have been developed to recommend machining parameters for efficient machining, they are basically for general reference and hard to use as well;In this research, a self-organizing fuzzy-nets optimization system was developed to generate a knowledge bank that can show the required cutting power on-line for a short length of time in an NC verifier. The fuzzy-nets system (FNS) utilizes a five-step self-learning procedure. A generic FNS program consisting of fuzzification and defuzzification modules was implemented in the C++ programming language to perform the procedure. The FNS was assessed before an actual experiment was set up to collect data;The performance of the FNS was then examined for end milling operations on a Fadal VMC40 vertical machining center. The cutting force signals were measured by a three-component dynamometer mounted on the table of the Fadal CNC machine with the workpiece mounted on it. Amplified signals were collected by a personal computer on which an Omega DAS-1401 analog-to-digital (A/D) converter was installed to sample the data on-line. Data sets were collected to train and test the system. The results showed that the FNS possessed a satisfactory range of accuracy with the intended applications of the model. The values of cutting power predicted by the FNS were more accurate than the formula values. Compared to the FNS system, dynamometers and amplifiers are very expensive. Thus, most of them could be replaced with the FNS.