Fault Diagnosis of Tool Wear Based on Weak Feature Extraction and GA-B-spline Network

• Main Authors: Weiqing CAO, Pan FU, Genhou XU
• Format: Article
• Language: English
• Published: IFSA Publishing, S.L. 2013-05-01
• Series: Sensors & Transducers
• Subjects: Independent component analysis; Empirical mode decomposition; Genetic algorithms; Tool wear; B-spline neural networks
• Online Access: http://www.sensorsportal.com/HTML/DIGEST/may_2013/P_1192.pdf

In view of the strong background noise involved in vibration signal of tool wear and the difficulty to obtain fault frequencies, so, it is important to de-noise before the further processing. However, the traditional de-noising methods, based on Gaussian noise assumption, lose here because the noise is identified as containing a high non-Gaussian component. Independent component analysis (ICA) was recently developed to deal with the blind source separation problem and it is particularly effective in the separation of non-Gaussian signals. So, this paper proposed a signal-noise-separated method with ICA, then de-noise signals are decomposed with empirical mode decomposition (EMD) and we got the better signals characters. Finally, Tool wear by identified by GA-B-spline neural network. B-spline networks is traditionally trained by using gradient-based methods, this may fall into local minimum during the learning process. Here, it is trained using genetic algorithms to search for global optimization. The experimental results show that the diagnosis approach put forward in this paper can effectively identify tool wear fault patterns in noise background and it has great application potential in health condition monitoring of tool wear.