Acoustic Emission Signal Processing Research Articles

A Review on Fault Diagnosis and Condition Monitoring of Gearboxes by Using AE Technique

Gearboxes have a significant role in the operation of rotating machinery. Several techniques exist for the diagnosis of fault in gearboxes. This work presents the techniques for the condition monitoring and fault diagnosis of the gearboxes based on Acoustic Emission (AE). The review describes the application of AE to detect the various types of fault in the gearboxes for different types of gear, the AE parameters used for the gearbox fault diagnosis, AE signal processing techniques, feature extraction techniques for the fault identification of gearbox, modelling of AE generated in gearbox and fault location detection using AE technique. After detailed description of the various techniques, some potential research issues are also discussed which is associated with condition monitoring and fault diagnosis of gearboxes using AE.

Smart machining: Monitoring of CFRP milling using AE and IR

In aerospace industry, polymers reinforced with carbon fibre offer several advantages, such as weight/strength ratio and corrosion resistance. However, the manufacturing process is complex, especially the machining of composite materials, due to their specific properties and characteristics. The detection and prediction of surface finishing and occurrence of defects during manufacturing using non-destructive techniques is industrially useful, in terms of fomenting automated manufacturing systems to increase productivity and quality control. Acoustic emission signal processing is being successful in mechanical tests for determining material defects, more recently used as a method for manufacturing evaluation of machined parts. This paper focused in the machining process of CFRP plate, in terms of surface finishing evaluation with comprehension of process phenomenology using analyses of temperature and acoustic emission signal features. The aim of this paper is to show the possibility of online monitoring of the cutting process and to understand mechanism behind variables of control, with comparisons of different milling parameters and focusing in their impacts in acoustic emission signals and surface finishing, with suggestions of evaluation and prediction methods of machined surface quality using mainly acoustic emission signals.

Study on Acoustic Emission Characteristics and Failure Prediction of Post-High-Temperature Granite

Abstract In an attempt to study the acoustic emission (AE) characteristics of granite specimens (25°C~600°C) during high-temperature deformation, the uniaxial compression and general triaxial compression tests were carried out. The AE signals were collected simultaneously, following which the AE signals were processed to predict rock failure. The study makes some interesting conclusions. In the uniaxial compression test, AE activity is found to be more active during the postpeak stress phase. In the triaxial compression test, AE activity is found to be more active during the postpeak stress phase. The AE energy and the cumulative energy under uniaxial compression are much larger than the triaxial compression at the failure moment of the granite specimen after the same heat treatment. As the heat-treatment temperature increases, the peak AE energy decreases, but the cumulative AE energy decreases first and then increases. The failure of granite was predicted by AE signal processing, and good prediction results were obtained for different experimental conditions and the different preset failure time tf. The time at which the forecast point appears is usually close to the time at which the rock specimen enters the plastic deformation phase. The value of the parameter tf impacts the curve shape and prediction duration without affecting the occurrence of the forecast point. In addition, the shape of curve changes stepwise as the tf changes.

An improved variational mode decomposition method based on particle swarm optimization for leak detection of liquid pipelines

Leak detection is critical for the safety management of pipelines since leakages may cause serious accidents. The present paper aims to develop an efficient method able to detect the presence and importance of leaks in pipelines.This method relies on adequate signal processing of acoustic emission (AE) signals, and improves the variational mode decomposition (VMD) for signal de-noising. In order to optimize the governing parameters, i.e. the penalty term and the mode number of VMD, the particle swarm optimization (PSO) algorithm is coupled to a fitness function based on maximum entropy (ME). After the signal reconstruction, based on the energy ratio of each VMD sub-mode, the waveform feature vectors for leak detection are extracted. Finally, the support vector machine (SVM) is employed for leak pattern recognition. For calibration purposes, artificial input signal is simulated. The results show that the proposed PSO-VMD method is capable of de-noising background noise.For validation purposes, experiments have been conducted on metal pipelines, with water flow. For sensitivity analysis, a set of five different leak apertures are adopted: aperture diameters as {10; 12; 15; 20; 27} mm, whereas the pipeline diameter is 108 mm. A database of AE signals is collected for each leak aperture, and different time sequences. The proposed method appears to be efficient since the classification accuracy of the SVM method reaches up to 100% in identifying the size of the leak on the basis of the AE signals collected in the database for the same leak size, and 89.3% on the basis of the whole database.

Acoustic Emission Recognition Based On A Two-Streams Convolutional Neural Network

The Convolutional Neural Network (CNN) is a widely used deep neural network. Compared with the shallow neural network, the CNN network has better performance and faster computing in some image recognition tasks. It can effectively avoid the problem that network training falls into local extremes. At present, CNN has been applied in many different fields, including fault diagnosis, and it has improved the level and efficiency of fault diagnosis. In this paper, a two-streams convolutional neural network (TCNN) model is proposed. Based on the short-time Fourier transform (STFT) spectral and Mel Frequency Cepstrum Coefficient (MFCC) input characteristics of two-streams acoustic emission (AE) signals, an AE signal processing and classification system is constructed and compared with the traditional recognition methods of AE signals and traditional CNN networks. The experimental results illustrate the effectiveness of the proposed model. Compared with single-stream convolutional neural network and a simple Long Short-Term Memory (LSTM) network, the performance of TCNN which combines spatial and temporal features is greatly improved, and the accuracy rate can reach 100% on the current database, which is 12% higher than that of single-stream neural network.

Experimental investigation on location of debris impact source based on acoustic emission

In view of the objective reality that it takes a long time for debris to locate the impact point when debris impacts the spacecraft bulkhead, timely and accurate determination of the location of the impact source is the premise and basis for damage repair, which is of great significance to the safety of astronauts. In this paper, the experiments of 2A12 aluminium plate impacted by 2A12 aluminium projectile under different impact parameters have been performed by using the self-built acoustic emission testing system and two-stage light gas gun loading system. The calculated coordinates are compared with the experimental coordinates by using triangular positioning method, quadrilateral positioning method and acoustic emission (AE) signal processing method based on Wavelet transform. The experimental results show that the positioning accuracy of quadrilateral positioning method is better than that of triangular positioning method, and the acoustic emission signal processing method based on wavelet transform is obviously better than that of quadrilateral positioning method.

Damage characterization of carbon/epoxy composites using acoustic emission signals wavelet analysis

ABSTRACTThis paper will investigate wavelet packet transform of acoustic emission (AE) signal processing in order to identify failure mechanisms of unidirectional carbon/epoxy composites. For this purpose, carbon/epoxy composites were tested in tensile test by two different loading conditions. For the first specimen, the load was applied through fiber direction (T0) and for the second specimen, the load was applied perpendicular to the fiber direction (T90), AE signals were collected during experimental test procedure and the signal’s waveforms were decomposed into various wavelet levels that each includes details and approximations so-called. Each data is related to a specific frequency range. The results show that type and percentage of failure mechanisms in various loadings are different. In T0, four micro-failure mechanisms of matrix cracking, fiber/matrix debonding, fiber pull-out and fiber breakage were observed and Fiber pull-out was determined as main inter-phase failure mechanism. Debonding were dominant failure micro mechanisms of T90. The results were verified with scanning electron microscopic observations.

Acoustic emission for diagnosing cable way steel support towers

The paper reports results of the study on the possibility of using the acoustic emission method in diagnosing fatigue and corrosion damage in steel elements of the cable way support towers. The assessment of the sensitivity of the structure to the recorded destructive processes is based on the structural damage classification method using the patterns created as a result of statistical and mathematical processing of acoustic emission signals through image analysis and grouping methods.

Research and Application of Acoustic Emission Signal Processing Technology

Acoustic emission signal processing has always been a key problem in damage assessment of components and materials, and parameter analysis and waveform analysis are two important means of acoustic emission signal processing. In this paper, the parameter analysis method in signal processing and the meaning and characteristics of each parameter are introduced. Then, the waveform analysis method is introduced and the basic theory of wavelet transform and its spectrum analysis in waveform analysis method are mainly analyzed. The comparison between wavelet transform and Fourier transform is made: traditional Fourier transform signal processing methods can only be used to process transient stationary signals, and the transient non-stationary signals cannot be analyzed locally. The wavelet transform can take into account the characteristics of both the time and frequency domains, and it can also reflect the local signal well. Through the analysis of the signals in the process of concrete damage, it is found that the two signal processing methods mentioned above can well characterize the damage evolution of materials and damage precursors.

Fractal signal processing method of acoustic emission monitoring for seismic damage of concrete columns

In this paper, a new fractal theory-based acoustic emission (AE) signal processing method is proposed. It is found that both the curve lengths and fractal dimensions (FDs) of AE signal are related ...

Analysis of Surface Defects in Composites Using Digital Image Correlation and Acoustic Emission

This work employs the acoustic emission (AE) method for material state monitoring. AE presents a non‑destructive evaluation technique, which could be used for detection of microstructural changes in composite material. Work describes the process of acquisition of AE in tensile loading of carbon composite materials. In course of tensile stress, the composite was monitored with optical method, applying principles of digital image correction (DIC). Optical stereovision method enables calculation of field shift and field of proportional deformation at composite surface. The objective is analysis of damage in carbon composite materials and employ the methodology of AE signal processing for facilitation of early damage diagnostics and prediction of structural failure. For this purpose, the experimental setup was designed to obtain results from 50 nominally identical composite samples in tensile loading test. Force load applied on samples was synchronically recorded along with AE and image data. Experimental data were subsequently analysed in a way enabling the description of typical phenomena in course of every measurement. Results show that observation of AE sources could be employed in facilitation of early damage diagnostics and establishment of failure prognosis. It is about internal changes in composite material.

SOME ASPECTS OF ACOUSTIC EMISSION AT MACHINING COMPOSITE MATERIALS

The processing of outcomes acoustic emission signals simulation is conducted at ascending composite machining speed and cutting depth, change of treated composite physical-mechanical characteristics with definition the ratio of acoustic emission signal amplitude average level to acoustic emission signal amplitude average level standard deviation. Are determined the regularity of dimensionless coefficient change, which characterizes the ratio of acoustic emission signal amplitude average level to its standard deviation at change of the influential factors. The processing of experimental acoustic emission signals is conducted at composite machining speed ascending and ascending cutting depth with definition the regularity of dimensionless coefficient change, which characterizes the ratio of acoustic emission signal amplitude average level to its standard. It is shown that the theoretical and experimental regularity of investigated coefficient value change at ascending machining speed and cutting depths will be agreed one another.

Acoustic emission signal processing framework to identify fracture in aluminum alloys

Acoustic emission (AE) is a common nondestructive evaluation tool that has been used to monitor fracture in materials and structures. The direct connection between AE events and their source, however, is difficult because of material, geometry and sensor contributions to the recorded signals. Moreover, the recorded AE activity is affected by several noise sources which further complicate the identification process. This article uses a combination of in situ experiments inside the scanning electron microscope to observe fracture in an aluminum alloy at the time and scale it occurs and a novel AE signal processing framework to identify characteristics that correlate with fracture events. Specifically, a signal processing method is designed to cluster AE activity based on the selection of a subset of features objectively identified by examining their correlation and variance. The identified clusters are then compared to both mechanical and in situ observed microstructural damage. Results from a set of nanoindentation tests as well as a carefully designed computational model are also presented to validate the conclusions drawn from signal processing.

Determination of the distortions introduced by the compression algorithm in the results of the measurement of the local impact control device “Prognoz-L”.

The article discusses the application of the u-law algorithm from the G.711 standard, which performs lossy compression. A theoretical error introduced by this algorithm into discrete samples of digitized acoustic emission signals is determined. The effect of compression on signal parameters was also determined. During the analysis, the data from the local "Prognoz-L" device were used. "Prognoz-L" performs rockburst hazard assessment. The data were processed in an automatic mode to isolate pulsed signals containing acoustic emission. The processing of the obtained sample of pulse signals made it possible to determine the basic statistical characteristics of the errors introduced by the compression algorithm into the values of the signal parameters. The possibility of using this algorithm for preliminary processing of acoustic emission signals is established with the aim of keeping rock pressure monitoring in low-power acoustic-emission devices.

Dimensionality Reduction of Sensorial Features by Principal Component Analysis for ANN Machine Learning in Tool Condition Monitoring of CFRP Drilling

With the aim to perform sensor monitoring of tool conditions in drilling of stacks made of two carbon fiber reinforced plastic (CFRP) laminates, a machine learning procedure based on the acquisition and processing of thrust force, torque, acoustic emission and vibration sensor signals during drilling is developed. From the acquired sensor signals, multiple sensorial features are extracted to feed artificial neural network-based machine learning paradigms, and an advanced feature extraction methodology based on Principal Component Analysis (PCA) is implemented to decrease the dimensionality of sensorial features via linear projection of the original features into a new space. By feeding artificial neural networks with the PCA features, the diagnosis of tool flank wear is accurately carried out.

Tool Condition Monitoring of Single-point Dressing Operation by Digital Signal Processing of AE and AI

This work aims at determining the right moment to stop single-point dressing the grinding wheel in order to optimize the grinding process as a whole. Acoustic emission signals and signal processing tools are used as primary approach. An acoustic emission (AE) sensor was connected to a signal processing module. The AE sensor was attached to the dresser holder, which was specifically built to perform dressing tests. In this work there were three types of test where the edit parameters of each dressing test are: the passes number, the dressing speed, the width of action of the dresser, the dressing time and the sharpness. Artificial Neural Networks (ANNs) technique is employed to classify and predict the best moment for stopping the dressing operation. During the ANNs use, the results from Supervised Neural Networks and Unsupervised Neural Networks are compared.

Detecting transition to chatter mode in peakless tool turning by monitoring vibration and acoustic emission signals

Stability of a peakless tool turning on slender shafts was studied under conditions of low- and high-magnitude vibrations by registering and short-time Fourier transformation (STFT) processing of acoustic emission (AE) and vibration acceleration (VA) signals. Both VA and AE signals have been registered in three positions of the cutting tool on the workpiece and for different shaft diameters. Both amplitude- and frequency-dependent AE and VA characteristics were obtained and analyzed for overall process signal length as well as for single frames. It was shown that power spectrum characteristic could be used for monitoring the fast-occurring changes in the cutting process stability. A criterion of the cutting process stability based on the power spectrum has been offered.

Digital signal processing of acoustic emission signals using power spectral density and counts statistic applied to single‐point dressing operation

Dressing is an important operation for the grinding process. Its goal is to recondition the wheel tool to re-establish its cutting characteristics, owing to the wear produced after successive passes. Monitoring systems that use acoustic emission (AE) have been studied to correlate the signals with several tool conditions. This study brings a new approach of processing AE signals with the purpose of identifying the correct moment to stop the dressing, which is essential in an automatic control system. From the AE signals collected in dressing tests with aluminium oxide grinding wheel and single-point dresser, spectral analysis was made through power spectral density, selecting frequencies bands that best characterise the process. The statistical parameter ‘counts’ was applied to the raw signal unfiltered and filtered in the selected bands in order to identify the tool condition and, in turn, towards a monitoring system implementation. Results showed an expressive relation between tool cutting conditions and processed signals in the selected bands. There was a great disparity of the filtered signals in the selected bands and signals unfiltered, reflecting that the filtered ones were more efficient in terms of process automation.

Dry Turning of Ti6Al4V: Tool Wear Curve Reconstruction Based on Cognitive Sensor Monitoring

Dry turning of Ti6Al4V alloy is a hard process due to the low thermal conductivity and flexibility of this difficult-to-machine material, generating very high temperatures in both workpiece and tool cutting edge, rapid tool wear and high vibrations during machining. The use of coolants offers the advantage to reduce the high process temperatures, but is not suitable in a green technology perspective due to its high environmental impact. With the aim to allow for dry turning of Ti6Al4V alloy, monitoring of tool wear during the process is required. To achieve this goal, a cognitive sensor monitoring procedure based on the acquisition and processing of force, acoustic emission and vibration sensor signals is implemented, allowing for an accurate tool wear curve reconstruction.

AE Signal Processing for Propagating Crack of Wind Turbine Blade Based on Optimized Morlet Wavelet

. Because the AE (Acoustic Emission, AE) characteristics of crack propagating of wind turbine blade is sudden and impactive, Shannon entropy is adopted to optimize the Morlet wavelet basis function by bandwidth parameters. Experimental results show that wavelet scale has relatively concentrated time-frequency characteristics when optimal bandwidth spectral parameter is 5.1. Compared with situation without optimization, it improves the noise interference phenomenon and be able to clearly show the impact characteristics of spectrum, which indicates that optimized wavelet scale is suitable for the extraction of the AE signal characteristic in wind turbine blade crack propagation.