Separation of individual operation signals from mixed sensor measurements

Abstract

Sensor system measurements are generally mixed signals measured from multiple independent/dependent operations embedded in a complex system. In this article, a novel method is developed to separate the source signals of individual operations from the mixed sensor measurements by integrating the independent component analysis method and the Sparse Component Analysis (SCA) method. The proposed method can efficiently estimate the source signals that include both independent signals and dependent signals that have some dominant components in the time or some linear transform domains (e.g., frequency domain, time/frequency domain, or wavelet domain). In addition, an SCA method is also developed in this article that can automatically identify the dominant components in multiple linear transform domains. A case study of a forging process is conducted to demonstrate the effectiveness of the proposed methods.