Neural network based signal processing scheme for automatic tool wear recognition

Monitoring of components in the manufacturing plants involves the automatic detection and identification of failure events. One of the important machine monitoring problems is the monitoring of tool wear in automatic metal drilling systems. The purpose of tool wear detection systems is to actually track down the wearing process of the machining tool, allowing the estimation of the quality of parts being machined by tool and prediction of the useful life of tools. Conventional methods of detecting the tool wear from processing the sensor measured signals have led to tool wear detection systems which perform well for a given set of machining parameters, but are not capable of meeting performance requirements in real manufacturing operations, where the machining parameters are more varied. This paper reports a automatic tool wear recognition scheme based on neural network technology. This technology provides an improved tool wear recognition alternative because of potential of neural networks to operate in real time mode and to handle data that may be distorted and noisy.