A NEW TOOL LIFE CRITERION FOR TOOL CONDITION MONITORING USING A NEURAL-NETWORK

On-line tool condition monitoring is important to prevent workpieces a nd tools from damage, and to increase the effective machining time of a machine tool. It is necessary to define tool-life criteria clearly, for indirect methods of on-line tool condition monitoring. There are m any tool life criteria that depend on wear manner, economic considerat ions, workpiece dimensional tolerance and surface roughness. However, the signal measured by a sensor (e.g. cutting force) usually represent s the tool wear condition contributed from a different wear zone. This implies that it is difficult to extract a single wear criterion from a convoluted sensor signal. When multiple signal features are used, th e response of the features to the tool life cannot be clearly seen, an d the tool life prediction may not be reliable. This paper presents an investigation into tool life criteria in raw turning. A new tool-life criterion depending on a pattern-recognition technique is proposed. T he neural network and wavelet techniques are used to realize the new c riterion. The experimental results show that this criterion is applica ble to tool condition monitoring in a wide range of cutting conditions .