Abstract: A fault detection method for Cutting tools based on wavelet network, which collects multi-source feature parameters of cutting tools is proposed to realize the on-line state detection based on the non-liner model and leaning system of wavelet NN. Then aiming at the problem of “MIMO” diagnosis system—— the “dimension disaster” and the slow learning speed, the wavelet network is improved by optimization algorithm that can adjust and search for the wavelet parameter adaptively in order to find the optimum wavelet neurons. Finally, the simpler structure and quickly convergent velocity of the new algorithm is demonstrated by simulation results.

Key words: Adjusting adaptively, Fault detection on cutting tools, Optimum parameter searching, Wavelet network, Actuation sense and control integration, Bioinspired soft robotics, Soft robotic gripper, Variable stiffness