Wavelet-Based Online Evaluation and Classification of Wear State Using Acoustic Emission

2012 ◽  
Author(s):  
Dorra Baccar ◽  
Dirk Söffker
Keyword(s):  
Acoustic Emission ◽  
Low Cost ◽  
Time Frequency ◽  
Online Evaluation ◽  
Cost System ◽  
State Classification ◽  
Online Examination ◽  
Field Programmable ◽  
Short Time

Advanced signal processing approaches such time-frequency analysis are widely used for online evaluation, damage detection, and wear state classification. The idea of this paper is to introduce a new methodology for online examination of wear phenomena in metallic structure by means of acoustic emission (AE), Short-Time Fourier Transform (STFT) and Wavelet Transform (WT). The proposed novel low-cost system is developed for analyzing and monitoring specific signals indicating tribological effects with focus on field programmable gate array (FPGA) implementation of discrete WT (DWT). In addition, experimental results obtained from each approach are given showing the success of the introduced approach.

scholarly journals Application of the Acoustic Emission Method for Diagnosis of On-Load Tap Changer

2017 ◽  
Vol 42 (1) ◽  
pp. 29-35 ◽  
Author(s):  
Henryk Majchrzak ◽  
Andrzej Cichoń ◽  
Sebastian Borucki
Keyword(s):  
Acoustic Emission ◽  
Frequency Domain ◽  
Time Domain ◽  
Operating Conditions ◽  
Time Frequency ◽  
Measurement Results ◽  
Tap Changer ◽  
The Time Domain ◽  
Load Tap Changer ◽  
Short Time

Abstract This paper provides an example of the application of the acoustic emission (AE) method for the diagnosis of technical conditions of a three-phase on-load tap-changer (OLTC) GIII type. The measurements were performed for an amount of 10 items of OLTCs, installed in power transformers with a capacity of 250 MVA. The study was conducted in two different OLTC operating conditions during the tapping process: under load and free running conditions. The analysis of the measurement results was made in both time domain and time-frequency domain. The description of the AE signals generated by the OLTC in the time domain was performed using the analysis of waveforms and determined characteristic times. Within the time-frequency domain the measured signals were described by short-time Fourier transform spectrograms.

scholarly journals Undervoltage Identification in Three Phase Induction Motor Using Low-Cost Piezoelectric Sensors and STFT Technique

Proceedings ◽  
2019 ◽  
Vol 42 (1) ◽  
pp. 72
Author(s):  
Leonardo Carvalho ◽  
Guilherme Lucas ◽  
Marco Rocha ◽  
Claudio Fraga ◽  
Andre Andreoli
Keyword(s):  
Acoustic Emission ◽  
Induction Motor ◽  
Large Scale ◽  
New Technologies ◽  
Low Cost ◽  
Industrial Applications ◽  
Piezoelectric Sensors ◽  
Time Frequency ◽  
Three Phase ◽  
Ae Signals

Three-phase induction motors (IMs) are electrical machines used on a large scale in industrial applications because they are versatile, robust and low maintenance devices. However, IMs are significantly affected when fed by unbalanced voltages. Prolonged operation under voltage unbalance (VU) conditions degrades performance and shortens machine life by producing imbalances in stator currents that abnormally raise winding temperature. With the development of new technologies and research on non-destructive techniques (NDT) for fault diagnoses in IMs, it is relevant to obtain economically accessible, efficient and reliable sensors capable of acquiring signals that allow the identification of this type of failure. The objective of this study is to evaluate the application of low-cost piezoelectric sensors in the acquisition of acoustic emission (AE) signals and the identification of VU through the analysis of short-term Fourier transform (STFT) spectrograms. The piezoelectric sensor makes NDT feasible, as it is an affordable and inexpensive component. In addition, STFT allows time-frequency analyses of acoustic emission signals. In this NDT, two sensors were coupled on both sides of an induction motor frame. The AE signals obtained during the IM operation were processed and the resulting spectrograms were analyzed to identify the different VU levels. After comparing the AE signals for faulty conditions with the signals for the IM operating at balanced voltages, it was possible to obtain a desired identification that confirmed the successful application of low-cost piezoelectric sensors for VU condition detection in three-phase induction machines.

scholarly journals Electrodermal Activity Based Emotion Recognition using Time-Frequency Methods and Machine Learning Algorithms

2021 ◽  
Vol 7 (2) ◽  
pp. 863-866
Author(s):  
Yedukondala Rao Veeranki ◽  
Nagarajan Ganapathy ◽  
Ramakrishnan Swaminathan
Keyword(s):  
Random Forest ◽  
Electrodermal Activity ◽  
Machine Learning Algorithms ◽  
Support Vector ◽  
Emotional States ◽  
Time Frequency ◽  
Suggested Technique ◽  
Clinical Conditions ◽  
Short Time

Abstract In this work, the feasibility of time-frequency methods, namely short-time Fourier transform, Choi Williams distribution, and smoothed pseudo-Wigner-Ville distribution in the classification of happy and sad emotional states using Electrodermal activity signals have been explored. For this, the annotated happy and sad signals are obtained from an online public database and decomposed into phasic components. The time-frequency analysis has been performed on the phasic components using three different methods. Four statistical features, namely mean, variance, kurtosis, and skewness are extracted from each method. Four classifiers, namely logistic regression, Naive Bayes, random forest, and support vector machine, have been used for the classification. The combination of the smoothed pseudo-Wigner-Ville distribution and random forest yields the highest F-measure of 68.74% for classifying happy and sad emotional states. Thus, it appears that the suggested technique could be helpful in the diagnosis of clinical conditions linked to happy and sad emotional states.

scholarly journals A Time–Frequency Acoustic Emission-Based Technique to Assess Workpiece Surface Quality in Ceramic Grinding with PZT Transducer

Sensors ◽  
2019 ◽  
Vol 19 (18) ◽  
pp. 3913 ◽  
Author(s):  
Martin A. Aulestia Viera ◽  
Paulo R. Aguiar ◽  
Pedro Oliveira Junior ◽  
Felipe A. Alexandre ◽  
Wenderson N. Lopes ◽  
...  
Keyword(s):  
Acoustic Emission ◽  
Surface Quality ◽  
Low Cost ◽  
Digital Signal ◽  
Ceramic Materials ◽  
Fast Response ◽  
Machining Process ◽  
Process Conditions ◽  
Time Frequency ◽  
Ceramic Components

Innovative monitoring systems based on sensor signals have emerged in recent years in view of their potential for diagnosing machining process conditions. In this context, preliminary applications of fast-response and low-cost piezoelectric diaphragms (PZT) have recently emerged in the grinding monitoring field. However, there is a lack of application regarding the grinding of ceramic materials. Thus, this work presents an analysis of the feasibility of using the acoustic emission signals obtained through the PZT diaphragm, together with digital signal processing in the time–frequency domain, in the monitoring of the surface quality of ceramic components during the surface grinding process. For comparative purpose, an acoustic emission (AE) sensor, commonly used in industry, was used as a baseline. The results obtained by the PZT diaphragm were similar to the results obtained using the AE sensor. The time–frequency analysis allowed to identify irregularities throughout the monitored process.

scholarly journals Development of Classification Algorithms for the Detection of Postures Using Non-Marker-Based Motion Capture Systems

2020 ◽  
Vol 10 (11) ◽  
pp. 4028
Author(s):  
Tatiana Klishkovskaia ◽  
Andrey Aksenov ◽  
Aleksandr Sinitca ◽  
Anna Zamansky ◽  
Oleg A. Markelov ◽  
...  
Keyword(s):  
Low Cost ◽  
Total Error ◽  
Rapid Development ◽  
Simple Algorithm ◽  
Classification Algorithms ◽  
Health And Wellbeing ◽  
Cost System ◽  
Two Parameters ◽  
Subject Performance

The rapid development of algorithms for skeletal postural detection with relatively inexpensive contactless systems and cameras opens up the possibility of monitoring and assessing the health and wellbeing of humans. However, the evaluation and confirmation of posture classifications are still needed. The purpose of this study was therefore to develop a simple algorithm for the automatic classification of human posture detection. The most affordable solution for this project was through using a Kinect V2, enabling the identification of 25 joints, so as to record movements and postures for data analysis. A total of 10 subjects volunteered for this study. Three algorithms were developed for the classification of different postures in Matlab. These were based on a total error of vector lengths, a total error of angles, multiplication of these two parameters and the simultaneous analysis of the first and second parameters. A base of 13 exercises was then created to test the recognition of postures by the algorithm and analyze subject performance. The best results for posture classification were shown by the second algorithm, with an accuracy of 94.9%. The average degree of correctness of the exercises among the 10 participants was 94.2% (SD1.8%). It was shown that the proposed algorithms provide the same accuracy as that obtained from machine learning-based algorithms and algorithms with neural networks, but have less computational complexity and do not need resources for training. The algorithms developed and evaluated in this study have demonstrated a reasonable level of accuracy, and could potentially form the basis for developing a low-cost system for the remote monitoring of humans.

scholarly journals Can a Smartphone Diagnose Parkinson Disease? A Deep Neural Network Method and Telediagnosis System Implementation

2017 ◽  
Vol 2017 ◽  
pp. 1-11 ◽  
Author(s):  
Y. N. Zhang
Keyword(s):  
Machine Learning ◽  
Nearest Neighbor ◽  
Low Cost ◽  
Walking Test ◽  
Time Frequency ◽  
Advantages And Disadvantages ◽  
Stacked Autoencoders ◽  
Knn Classifier ◽  
Frequency Features

Parkinson’s disease (PD) is primarily diagnosed by clinical examinations, such as walking test, handwriting test, and MRI diagnostic. In this paper, we propose a machine learning based PD telediagnosis method for smartphone. Classification of PD using speech records is a challenging task owing to the fact that the classification accuracy is still lower than doctor-level. Here we demonstrate automatic classification of PD using time frequency features, stacked autoencoders (SAE), and K nearest neighbor (KNN) classifier. KNN classifier can produce promising classification results from useful representations which were learned by SAE. Empirical results show that the proposed method achieves better performance with all tested cases across classification tasks, demonstrating machine learning capable of classifying PD with a level of competence comparable to doctor. It concludes that a smartphone can therefore potentially provide low-cost PD diagnostic care. This paper also gives an implementation on browser/server system and reports the running time cost. Both advantages and disadvantages of the proposed telediagnosis system are discussed.

scholarly journals Identification of Radar Signals Based on Time-Frequency Agility using Short-Time Fourier Transform

2018 ◽  
Vol 1 (2) ◽  
pp. 1-8 ◽  
Author(s):  
Ashraf Adamu Ahmad ◽  
A. S. Saliu ◽  
Abel E. Airoboman ◽  
U. M. Mahmud ◽  
S. L. Abdullahi
Keyword(s):  
Fourier Transform ◽  
Classification Accuracy ◽  
Additive White Gaussian Noise ◽  
Noise Power ◽  
Short Time Fourier Transform ◽  
Operating Characteristics ◽  
Time Frequency ◽  
Radar Signals ◽  
Short Time

With modern advances in radar technologies and increased complexity in aerial battle, there is need for knowledge acquisition on the abilities and operating characteristics of intercepted hostile systems. The required knowledge obtained through advanced signal processing is necessary for either real time-warning or in order to determine Electronic Order of Battle (EOB) of these systems. An algorithm was therefore developed in this paper based on a joint Time-Frequency Distribution (TFD) in order to identify the time-frequency agility of radar signals based on its changing pulse characteristics. The joint TFD used in this paper was the square magnitude of the Short-Time Fourier Transform (STFT), where power and frequency obtained at instants of time from its Time-Frequency Representation (TFR) was used to estimate the time and frequency parameters of the radar signals respectively. Identification was thereafter done through classification of the signals using a rule-based classifier formed from the estimated time and frequency parameters. The signals considered in this paper were the simple pulsed, pulse repetition interval modulated, frequency hopping and the agile pulsed radar signals, which represent cases of various forms of agility associated with modern radar technologies. Classification accuracy was verified using the Monte Carlo simulation performed at various ranges of Signal-to-Noise Ratios (SNRs) in the presence of noise modelled by the Additive White Gaussian Noise (AWGN). Results obtained showed identification accuracy of 99% irrespective of the signal at a minimum SNR of 0dB where signal and noise power were the same. The obtained minimum SNR at this classification accuracy showed that the developed algorithm can be deployed practically in the electronic warfare field for accurate agility classification of airborne radar signals.

Wear state evaluation of inner-diameter saw blade based on vibration and noise signals during processing

2020 ◽  
pp. 095440542094910
Author(s):  
Yong-Chen Pei ◽  
Qin-Jian Liu ◽  
Ru-Shi Zhao ◽  
Hang Zhang
Keyword(s):  
Real Time ◽  
Low Cost ◽  
Good Method ◽  
Short Time Fourier Transform ◽  
Time Frequency ◽  
State Evaluation ◽  
Inner Diameter ◽  
Saw Blade ◽  
Short Time

Slicing and cutting processing by inner-diameter saw blade has advantages of low cost and convenient adjustment of machine tool. In small size and batch processing, inner-diameter saw blade slicing and cutting is a common-used processing method. However, the wear of inner-diameter saw blade will seriously affect surface quality of workpiece cross section. It is necessary to monitor and evaluate the wear in real time. Based on the short-time Fourier transform, this article introduces a new method for assessing the wear of inner-diameter saw blade. By measuring and analyzing time–frequency characteristics of vibration and machining noise signals, wear is monitored in real time. The results show that it is a good method to monitor the wear of inner-diameter saw blade based on vibration of machine and machining noise signals. In addition, this method has no interference and influence on cutting work of inner-diameter saw blade, which provides a new idea for other forms of tool wear detection in engineering.

Deep Learning of EEG Time–Frequency Representations for Identifying Eye States

2018 ◽  
Vol 10 (02) ◽  
pp. 1840006 ◽  
Author(s):  
Dharmendra Gurve ◽  
Sridhar Krishnan
Keyword(s):  
Biomedical Signals ◽  
Time Frequency ◽  
Bipolar Mood Disorder ◽  
Detection Analysis ◽  
State Identification ◽  
State Classification ◽  
Eye Fatigue ◽  
Eeg Data ◽  
Fully Connected

A new Convolutional Neural Network (CNN) architecture to classify nonstationary biomedical signals using their time–frequency representations is proposed. The present method uses the spectrogram of the biomedical signals as an input to CNN, in addition Non-negative matrix factorization (NMF) dictionary elements are used as an additional feature to improve the performance of the CNN model. Considering a number of applications involving eye state classification, such as in Parkinson’s disease detection, analysis of eye fatigue in 3D TVs, driver’s drowsiness detection, infant sleep-waking state identification, and classification of bipolar mood disorder and attention deficit hyperactivity, the proposed method was applied to Electroencephalography (EEG) data for classification of eye state. First, the spectrogram of EEG signal is obtained and used as an image input to CNN, simultaneously, the NMF feature is also fed to CNN. Further, both features are combined in fully connected layer of CNN architecture. The proposed method is compared with other existing methods for eye state detection and shows good classification accuracy with 96.16%. The prediction rate for the proposed method is 134 observations/second, which is suitable for brain–computer interface applications.

Sign in / Sign up

Export Citation Format

Share Document