An adaptive chatter signal enhancement approach for early fault diagnosis in machining process

Abstract

Chatter is a fast-changing machining fault that needs to be timely diagnosed to avoid the deterioration. However, it is a big challenge to recognize the weak chatter component out of the machining process signal that includes strong disturbances caused by the measurement noise, the machining uncertainty in the early stage. In this paper, an adaptive chatter signal enhancement approach is proposed to improve the signal-to-noise ratio (SNR) based on the principle of statistical resonance. A Fokker-Planck model with smoothly quadratic double well potential is proposed and analytically solved without using the adiabatic approximation. Compared with the recorded double-well models, the superiority of this one mainly lies in its better resolvability and enhancement capacity. The cutting force signals are utilized to prove the efficiency of the approach to enhance and highlight the chatter signal out of strong disturbances for early chatter detection.