An increasing diversity of variants, reductions in the effective production life and a growing level of complexity in products all represent market factors. Companies must react to these factors in the field of manufacturing engineering. Flexible manufacturing systems(FMS) and computer integrated manufacturing(CIM) systems have been development to respond to the demands. FMS may be defined as a system dealing with high level distribution data processing and automated material flow using computer controlled machines, assembly cells, and so on together with computer integrated material handling and storage systems. Tooling management is one of the most essential aspects of an effectiveFMS.Implementation of an effective TMS will reduce tool cost and inventory, decrease machine downtime, and improve the quality of the parts being produced.
A proper tooling management strategy must contain monitoring and control of the tooling used in the FMS. A successful tool wear/fracture sensor has been a long standing goal of the manufacturing community. Some research workers have focused their studies on the tool life with significant amounts of experimentation and theoretical analyses and developed that correlate tool life with parameters. On the other hand, even more methods have been developed for tool wear sensing, none of which has achieved significant use in industry.
In this paper, one of the sensing methods of cutting tool condition on the basis of image processing is developed. This paper presents quantitative measuring systems of flank wear and crater wear, and a qualitative discrimination method of tool conditions using neural network is described. In the flank wear measurement, a mathematical model of tool cutting edge is established with describing minor/major cutting edge and nose as two straight lines and a circle. Flank wear can be calculated from the distant between the worn-out tool cutting edge and the fresh tool cutting edge created by the model. In the crater wear measurement, to obtain the crater contour, the contour detection algorithm which can adapt in a noisy image using the image consolidation and dilation process is suggested. Also to measure the crater depth, the automatic focusing method is applied with 1-D search algorithm for finding best focus. In the pattern recognition, a multilayered perceptron with back-propagation algorithm has been used. Some characteristic features of tool shape are extracted and used as input of neural network. The neural network system can recognize 15 kinds of tool abnormalities.
The experimental results performed in the several worn-out tools have proved the effectiveness of the method proposed.