Study on tool wear monitoring based on multi-source information fusion

Abstract

The paper developed a system of tool wear monitoring in advanced manufacture systems. In conventional wear-monitoring method, it cannot exhibit unique behavior found in regular modern machining systems. Because a single monitoring signal, such as the signal of force, temperature, ultrasound or AE, cannot exactly describe the state of tool work for monitoring in advanced manufacture. This paper, therefore, mainly researched on real-time cutter state monitoring using neural network, neural network integration and multi-sensor information integrating technology. Picture pattern-recognition and feature extracting were adopted and combined with other information of the cutter dynamically. The characteristic information was gathered using an appropriate model of cutter wear or damage. Neural network were used to imitate the complicated nonlinear mapping relationship and to fuse multi-kind sensors that collect wearing and damage information and make decision and judgment rapidly. © 2008 Springer-Verlag Berlin Heidelberg.