The paper presents an attempt to apply spectral analysis tools in processing acoustic emission (AE) pulses generated by partial discharge (FD). The experimental part of the paper describes spark gaps generating four types of PD and specifies parameters of measured acoustic signals and recalls the system used for measurement and analysis of the frequency spectra. Also, a spectral analysis procedure is presented, and frequency-domain descriptors characterizing AE pulses are defined. The results of the analysis are given both as time plots and amplitude and energy density spectra, related to values of the associated descriptors. The spectral analysis results cover AE pulses generated in systems modeling the following types of pD: point-plane type discharges in oil, surface discharges in oil, gas bubble discharges in oil and discharges in indeterminate-potential particles moving in oil. In conclusion, an ability to identify a specific type of FD on a basis of frequency spectra of AE pulses and their descriptors is presented. The comparative analysis enables one to select those descriptors, which can constitute unique criteria for identification of a specific PD type.

Identification of a specific type of PD from acoustic emission frequency spectra

Acoustic emission (AE) is defined as elastic waves associated with a rapid release of localized stress energy that is propagated within a material. It has been proven to be a useful tool for studying rock fractures (e.g., Lavrov 2003; Prikry et al. 2003; Tham et al. 2005; Amann et al. 2011; Shukla et al. 2013; Liu et al. 2015; Moradian et al. 2016; Zhang et al. 2016). Parametric and waveform data are two basic forms of recordable AE data. In the past few decades, parameter analysis has been widely used to observe progressive damage processes resulting from AE, as well as to evaluate the degree of damage (e.g., Butt 1999; Baddari et al. 2011; Xie et al. 2011; Zhao et al. 2013; Zhang et al. 2015). However, a single complicated AE waveform still might conceal important information, or may even present misleading information, since it is discriminated by only a few parameters (Behnia et al. 2014). Thus, it may be difficult to adequately discriminate the characteristics of AE waveforms using such parameter-based approaches. In waveform-based approaches, spectral analysis methods are often used for signal waveforms analyses. Because frequency content in an AE is determined by the wave source, studying the dominant frequency characteristics provides significant controls on investigations of microscopic fractures. Some of the pioneering work that investigated the dominant frequency characteristics of rocks was performed based on spectral analysis (e.g., Chugh et al. 1968; Armstrong 1969). Later, numerous relevant studies were conducted that evaluated the dominant frequency characteristics of AE waveforms (e.g., Stephens and Pollock 1971; Fleischmann et al. 1975; Woodward 1976; Kranz 1979; Sondergeld et al. 1980; Read et al. 1995; Xu et al. 2010). Owing to both hardware and software restrictions, it was impossible to realize continuous sampling, digitization and storage of a large quantity of AE waveforms within a short time. Previous studies did not contain statistical analyses and mainly focused on the change of dominant frequency characteristics since few waveforms could be recorded. With the development of high-performance computational processors and a new generation of sensors, the acquisition of multichannel raw waveforms has become possible (Behnia et al. 2014) and benefits the statistical analysis of the dominant frequency characteristics of AE waveforms. However, some latest studies about the dominant frequency of AE waveforms (Lu et al. 2013; Reiweger et al. 2015) still did not involve the statistical analysis. Ohnaka and Mogi (1981, 1982) expounded on the importance of statistical analysis for investigating the dominant frequency characteristics of AE waveforms. Moreover, statistical characteristics of dominant frequency of AE waveforms may be beneficial to understand the microscopic failure mechanism of rock. & L. Zheng zhenglu@scu.edu.cn

Dominant Frequency Characteristics of Acoustic Emissions in White Marble During Direct Tensile Tests

Acoustic-emission techniques are beginning to be frequently used as a means of detecting and locating deformation and fracture processes in both metallic and non-metallic engineering structures This has happened despite a relatively poor understanding of the basic physical processes involved in the generation, propagation, and detection of acoustic-emission signals, and of the influence in metals of factors such as composition and microstructure, upon the acoustic emission from different types of deformation and fracture processes The measurement of acoustic emission, potentially, could yield much information about dynamic aspects of these processes However, it is shown that the limitations of existing acoustic-emission recording instrumentation have enabled only qualitative information to be obtained to date Even this data, however, has demonstrated that metallurgical variables greatly affect the acousticemission response of metals, and simple models to account for this are discussed

Acoustic emission for physical examination of metals

Analyse spectrale et energétique d'une source ultrasonore en mouvement — application à l'Emission acoustique de l'aluminium soumis à déformation plastique

Frequency analysis has been applied to the acoustic emission signals resulting from the failure of steel wire ropes and individual wires taken from the ropes. These preliminary tests were performed on notched and unnotched specimens under rising and cyclic loads; the signals were digitized and analysed using a fast Fourier transform program. Failure of the wires generated signals with frequencies between 0 and 100 kHz; the average peak to peak amplitude values of the signals from individual wires were greater than those from the failure of notched wires within the rope. The signal duration also increased in the rope specimens.

Frequency analysis of the signals generated by the failure of constituent wires of wire rope

The authors describe a data acquisition system to determine the frequency spectrum between 500 KHz and 3 MHz of the initial acoustic emission. They measured the transfer function of the piezoelectric transducer and of the specimen for their results using a computer. They present an application of this device to the study of the emission of a low-carbon steel.

D. Rouby

Papers

A spectrum analysis of acoustic emission