Abstract—This paper introduces an approach for drill wear detection. Tool failure in machining processes will result in damages to workpiece. In this approach sound signal of drilling operation is recorded and analyzed in both time and frequency domains. Trend of sound signal statistical features are extracted as the drill becomes worn. Sound signal frequency spectrum is calculated using Fast Fourier Transform (FFT) to detect the effect of drill wear on frequency components of signal. In continue Wavelet Packet Decomposition (WPD) is implemented to focus on detected frequency bands. Finally a Feedforward Backpropagation Neural Network (FBNN) is designed and trained based on sound signal features extracted from wavelet packets. The FBNN classifies the tool state into three classes of wear. This approach provides a tool wear detection strategy with capability of online implementation.

Index Terms—Tool condition monitoring, sound, frequency spectrum, wavelet packets.

Hamed Rafezi is with School of Science and Engineering, Sharif University of Technology, International Campus, Kish, Iran(e-mail:h.rafezi@gmail.com)
Mehdi Behzad and Javad Akbari are with School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran (e-mail:akbari@sharif.edu, m_behzad@sharif.edu)

Cite: Hamed Rafezi, Mehdi Behzad, and Javad Akbari, "Time Domain and Frequency Spectrum Analysis of Sound Signal for Drill Wear Detection," International Journal of Computer and Electrical Engineering vol. 4, no. 5, pp. 722-725, 2012.