Acoustic emission signal processing for rolling bearing running state assessment using compressive sensing. (July 2017)

• Record Type: Journal Article
• Authors: Liu, Chang; Wu, Xing; Mao, Jianlin; Liu, Xiaoqin
• Is Part Of: Mechanical systems and signal processing. Volume 91(2017)
• Page Start: 395
• Page End: 406
• Publication Date: 2017-07
• Subjects: Acoustic emission -- Compressive sensing -- State assessment -- Compressive feature; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2016.12.010
• Languages: English
• ISSNs: 0888-3270

Abstract

Highlights: A method of extraction AE compression feature (AECF) based on compressive sensing is proposed. AECF feature extract from compressively-sensed data directly. We can use the AECF to evaluate the running state of the rolling bearing. The advantages of AECF include high efficiency, small computational cost, low sampling frequency demand. Abstract: In the signal processing domain, there has been growing interest in using acoustic emission (AE) signals for the fault diagnosis and condition assessment instead of vibration signals, which has been advocated as an effective technique for identifying fracture, crack or damage. The AE signal has high frequencies up to several MHz which can avoid some signals interference, such as the parts of bearing (i.e. rolling elements, ring and so on) and other rotating parts of machine. However, acoustic emission signal necessitates advanced signal sampling capabilities and requests ability to deal with large amounts of sampling data. In this paper, compressive sensing (CS) is introduced as a processing framework, and then a compressive features extraction method is proposed. We use it for extracting the compressive features from compressively-sensed data directly, and also prove the energy preservation properties. First, we study the AE signals under the CS framework. The sparsity of AE signal of the rolling bearing is checked. The observation and reconstruction of signal is also studied. Second, we present a method of extraction AE compressive feature (AECF) from compressively-sensed data directly. We demonstrate the energy preservation properties and the processing of the extracted AECF feature. We assess the running state of the bearing using the AECF trend. The AECF trend of the running state of rolling bearings is consistent with the trend of traditional features. Thus, the method is an effective way to evaluate the running trend of rolling bearings. The results of the experiments have verified that the signal processing and the condition assessment based on AECF is simpler, the amount of data required is smaller, and the amount of computation is greatly reduced.