Tool Wear Modeling in Face Milling Using Acoustic Emission - Artificial Neural Network Approach

Tool condition Monitoring is an important aspect to maintain quality of machined products. A tool is considered as failed, when it is no longer able to produce parts within the required specifications. Tool Condition Monitoring Systems (TCMS) are an integral part of machine condition monitoring systems, which continuously monitor the condition of the tool either using 'direct' or 'indirect' methods. In this paper, acoustic emission signals, which are widely used in TCMS are used along with surface roughness measurements to monitor the condition of an insert in face milling operations on grey cast iron. Artificial Neural Networks (ANN), a widely used artificial intelligence technique in machining applications have been used to model tool flank wear using acoustic emission signal parameters, surface roughness parameters and cutting conditions. Three ANN techniques have been investigated and compared namely Multi-layer Perceptron Neural Network (MLPNN), Radial basis function Neural Network (RBFNN) and Summation Wavelet - Extreme Learning Machine (SW-ELM). The results indicate that SW-ELM performs better than other two techniques in tool wear modeling and achieves a prediction accuracy of more than 85 % on test data.