Multiple-signal Synchronous Change-point Detection by Piecewise Linear Regression with Group Lasso Constraints

Abstract

This paper studies the synchronous multiple change-point detection problem involving multiple signals. The original signals are fitted by a piecewise linear regression with Group Lasso constraints to ensure the synchronism of the change points. Then, first-order difference is used to determine the candidate set of change points for the resultant fitted curve, and then the Bayesian information criterion (BIC) is utilized to determine change points from the candidate set. Monte Carlo simulation-based experiments are used to compare the new method with three commonly-used multi-signal synchronous change-point detection methods. The results show that the proposed method is superior in detecting both the number and the position of change points. The performance in real multiple vibration signals of cutting tools data further verifies the effectiveness of the method.