Researchers believe that the acoustic emission (AE) signal contains potentially valuable information for tool wear prediction and monitoring. However, it is not easy to obtain an effective result by this raw AE signal because AE signal generated from interaction of tool bit and workpiece is easily distorted by the transmission path and the measurement systems. Hence, in this research, an effective adaptive signal decomposing technique, empirical mode decomposition (EMD), is used to decompose the AE signal. The instantaneous frequencies and their amplitudes of the decomposed AE signal were extracted using Hilbert Huang transform. These features were used to train a competitive neural network (CNN) to classify the state of the tool. From the HHT analysis, it is found that the increase in tool flank wear resulted in an increase of AE signal amplitude and decreasein instantaneous frequency. This correlation enabled the competitive neural network to perform tool wear classification with acceptable accuracy. Hence, the method of classifying the condition of the tool from generated AE signal with EMD for decomposition and CNN for classification is proposed for tool condition monitoring.

tool wear, AE signal, empirical mode decomposition, competitive neural network.