Damage‐sensitive feature extraction with stacked autoencoders for unsupervised damage detection

2021 ◽  
Vol 28 (5) ◽  
Author(s):  
Moisés Felipe Silva ◽  
Adam Santos ◽  
Reginaldo Santos ◽  
Eloi Figueiredo ◽  
João C.W.A. Costa
Keyword(s):  
Feature Extraction ◽  
Damage Detection ◽  
Stacked Autoencoders

scholarly journals Modal-based vibrothermography using feature extraction with application to composite materials

2019 ◽  
Vol 19 (4) ◽  
pp. 967-986 ◽  
Author(s):  
Xintian Chi ◽  
Dario Di Maio ◽  
Nicholas AJ Lieven
Keyword(s):  
Feature Extraction ◽  
Composite Materials ◽  
Damage Detection ◽  
Infrared Camera ◽  
Experimental Tests ◽  
Detection Algorithm ◽  
Modal Testing ◽  
Theoretical Development ◽  
Operational Environment ◽  
Practical Applications

This research focuses on the development of a damage detection algorithm based on modal testing, vibrothermography, and feature extraction. The theoretical development of mathematical models is presented to illustrate the principles supporting the associated algorithms, through which the importance of the three components contributing to this approach is demonstrated. Experimental tests and analytical simulations have been performed in laboratory conditions to show that the proposed damage detection algorithm is able to detect, locate, and extract the features generated due to the presence of sub-surface damage in aerospace grade composite materials captured by an infrared camera. Through tests and analyses, the reliability and repeatability of this damage detection algorithm are verified. In the concluding observations of this article, suggestions are proposed for this algorithm’s practical applications in an operational environment.

scholarly journals Machine Learning-Enriched Lamb Wave Approaches for Automated Damage Detection

Sensors ◽  
2020 ◽  
Vol 20 (6) ◽  
pp. 1790
Author(s):  
Zi Zhang ◽  
Hong Pan ◽  
Xingyu Wang ◽  
Zhibin Lin
Keyword(s):  
Machine Learning ◽  
Feature Extraction ◽  
Feature Selection ◽  
Damage Detection ◽  
Lamb Wave ◽  
Lamb Waves ◽  
Feature Selection Method ◽  
Support Vector ◽  
Time Frequency ◽  
Damage States

Lamb wave approaches have been accepted as efficiently non-destructive evaluations in structural health monitoring for identifying damage in different states. Despite significant efforts in signal process of Lamb waves, physics-based prediction is still a big challenge due to complexity nature of the Lamb wave when it propagates, scatters and disperses. Machine learning in recent years has created transformative opportunities for accelerating knowledge discovery and accurately disseminating information where conventional Lamb wave approaches cannot work. Therefore, the learning framework was proposed with a workflow from dataset generation, to sensitive feature extraction, to prediction model for lamb-wave-based damage detection. A total of 17 damage states in terms of different damage type, sizes and orientations were designed to train the feature extraction and sensitive feature selection. A machine learning method, support vector machine (SVM), was employed for the learning model. A grid searching (GS) technique was adopted to optimize the parameters of the SVM model. The results show that the machine learning-enriched Lamb wave-based damage detection method is an efficient and accuracy wave to identify the damage severity and orientation. Results demonstrated that different features generated from different domains had certain levels of sensitivity to damage, while the feature selection method revealed that time-frequency features and wavelet coefficients exhibited the highest damage-sensitivity. These features were also much more robust to noise. With increase of noise, the accuracy of the classification dramatically dropped.

Feature Extraction in Damage Detection used Neural Network

2001 ◽  
Vol 2001.10 (0) ◽  
pp. 260-261
Author(s):  
Xiang Su ◽  
Ichiro Hagiwara ◽  
Qinzhong Shi
Keyword(s):  
Neural Network ◽  
Feature Extraction ◽  
Damage Detection

scholarly journals Bird Swarm Optimization-based Stacked Autoencoder Deep Learning for Umpire Detection and Classification

2020 ◽  
Vol 21 (2) ◽  
pp. 173-188
Author(s):  
Suvarna Nandyal ◽  
Suvarna Laxmikant Kattimani
Keyword(s):  
Feature Extraction ◽  
Deep Learning ◽  
Swarm Optimization ◽  
Stacked Autoencoder ◽  
Video Frames ◽  
Utilization Of Technology ◽  
Stacked Autoencoders ◽  
Maximal Sensitivity ◽  
Sensitivity Specificity ◽  
Local Gradient

One of the most watched and a played sport is cricket, especially in South Asian countries. In cricket, umpire has the power for making significant decisions about events in the field. With the growing increase of the utilization of technology in sports, this paper presents the umpire detection and classification by proposing an optimization algorithm. The overall procedure of the proposed approach involves three steps, like segmentation, feature extraction, and the classification. At first, the video frames are extracted from input cricket video, and the segmentation is performed based on Viola-Jones algorithm. Once the segmentation is done, the feature extraction is carried out using Histogram of Oriented Gradients (HOG), and Fuzzy Local Gradient Patterns (Fuzzy LGP). Finally, the extracted features are given to the classification step. Here, the classification is done using the proposed Bird Swarm Optimization-based stacked auto encoder deep learning classifier (BSO-Stacked Autoencoders), that categories into umpire or others. The performance of the umpire detection and classification based on BSO-Stacked Autoencoders is evaluated based on sensitivity, specificity, and accuracy. The proposed BSO-Stacked Autoencoder method achieves the maximal accuracy of 96.562%, the maximal sensitivity of 91.884%, and the maximal specificity of 99%, that indicates its superiority.

Damage detection of a cable-stayed bridge using feature extraction and selection methods

2019 ◽  
Vol 15 (9) ◽  
pp. 1165-1177 ◽  
Author(s):  
Hossein Babajanian Bisheh ◽  
Gholamreza Ghodrati Amiri ◽  
Masoud Nekooei ◽  
Ehsan Darvishan
Keyword(s):  
Feature Extraction ◽  
Damage Detection ◽  
Selection Methods ◽  
Cable Stayed Bridge ◽  
Feature Extraction And Selection

Damage detection of bridges based on combining efficient cepstral coefficients

2020 ◽  
pp. 107754632095834
Author(s):  
Hossein Babajanian Bisheh ◽  
Gholamreza Ghodrati Amiri ◽  
Masoud Nekooei ◽  
Ehsan Darvishan
Keyword(s):  
Feature Extraction ◽  
Feature Selection ◽  
Damage Detection ◽  
Structural Damage ◽  
Damage Index ◽  
Support Vector ◽  
Selection Methods ◽  
Significant Information ◽  
Effective Coefficients ◽  
Cepstral Coefficients

In this article, a novel vibration-based damage detection approach is proposed based on selecting effective cepstral coefficients, consisting of three main stages: (1) signal processing and feature extraction, (2) damage detection by combining effective cepstral coefficients through feature selection methods, and (3) performance evaluation. First, two feature extraction techniques are used in damage identification systems, including linear prediction cepstral coefficients and mel frequency cepstral coefficients. Second, to improve the performance of damage detection, the combination of the effective cepstral coefficients is proposed as a damage index. By applying several feature selection methods, the most effective coefficients are found and then combined to create a subset that carries the most significant information about the structural damage. Finally, the support vector machine classifier is performed to evaluate the proposed approach in detecting the structural damage. The proposed technique is verified using a suite of numerical and full-scale studies. Results confirm that the proposed method achieves a significant performance with great accuracy and reduces false alarms.

Sign in / Sign up

Export Citation Format

Share Document