A machine learning–based approach for adaptive sparse time–frequency analysis used in structural health monitoring

2020 ◽  
Vol 19 (6) ◽  
pp. 1963-1975 ◽  
Author(s):  
Yuequan Bao ◽  
Yibing Guo ◽  
Hui Li
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Structural Health Monitoring ◽  
Frequency Analysis ◽  
Health Monitoring ◽  
Instantaneous Frequency ◽  
Time Frequency Analysis ◽  
Time Frequency ◽  
Structural Health ◽  
Reconstructed Signal

Time–frequency analysis is an essential subject in nonlinear and non-stationary signal processing in structural health monitoring, which can give a clear illustration of the variation trend of time-varying parameters. Thus, it plays a significant role in structural health monitoring, such as data analysis, and nonlinear damage detection. Adaptive sparse time–frequency analysis is a recently developed method used to estimate an instantaneous frequency, which can achieve high-resolution adaptivity by looking for the sparsest time–frequency representation of the signal within the largest possible time–frequency dictionary. However, in adaptive sparse time–frequency analysis, non-convex least-square optimization is the most important and difficult part of the algorithm; therefore, in this research the powerful optimization capabilities of machine learning were employed to solve the non-convex least-square optimization and achieve the accurate estimation of the instantaneous frequency. First, the adaptive sparse time–frequency analysis was formalized into a machine-learning task. Then, a four-layer neural network was designed, the first layer of which was used for training the coefficients of the envelope of each basic functions in a linear space. The next two merge layers were used to solve the complex calculation in a neural network. Finally, the real and imaginary parts of the reconstructed signal were the outputs of the output layer. The optimal weights in this designed neural network were trained and optimized by comparing the output reconstructed signal with the target signal, and a stochastic gradient descent optimizer was used to update the weights of the network. Finally, the numerical examples and experimental examples of a cable model were employed to illustrate the ability of the proposed method. The results show that the proposed method which is called neural network–adaptive sparse time–frequency analysis can give accurate identification of the instantaneous frequency, and it has a better robustness to initial values when compared with adaptive sparse time–frequency analysis.

Enhancing Vibration-Based Structural Health Monitoring via Edge Computing: A Tiny Machine Learning Perspective

2021 ◽  
Author(s):  
Federica Zonzini ◽  
Francesca Romano ◽  
Antonio Carbone ◽  
Matteo Zauli ◽  
Luca De Marchi
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Structural Health Monitoring ◽  
Health Monitoring ◽  
Network Architecture ◽  
Binary Classification ◽  
Low Cost ◽  
Classification Problems ◽  
Computationally Efficient ◽  
Structural Health

Abstract Despite the outstanding improvements achieved by artificial intelligence in the Structural Health Monitoring (SHM) field, some challenges need to be coped with. Among them, the necessity to reduce the complexity of the models and the data-to-user latency time which are still affecting state-of-the-art solutions. This is due to the continuous forwarding of a huge amount of data to centralized servers, where the inference process is usually executed in a bulky manner. Conversely, the emerging field of Tiny Machine Learning (TinyML), promoted by the recent advancements by the electronic and information engineering community, made sensor-near data inference a tangible, low-cost and computationally efficient alternative. In line with this observation, this work explored the embodiment of the One Class Classifier Neural Network, i.e., a neural network architecture solving binary classification problems for vibration-based SHM scenarios, into a resource-constrained device. To this end, OCCNN has been ported on the Arduino Nano 33 BLE Sense platform and validated with experimental data from the Z24 bridge use case, reaching an average accuracy and precision of 95% and 94%, respectively.

PCA and ICA Based Prognostic Health Monitoring of Electronic Assemblies Subjected to Simultaneous Temperature-Vibration Loads

2017 ◽  
Author(s):  
Pradeep Lall ◽  
Tony Thomas
Keyword(s):  
Frequency Analysis ◽  
Health Monitoring ◽  
Instantaneous Frequency ◽  
Principal Component ◽  
Component Analysis ◽  
Time Frequency Analysis ◽  
Time Frequency ◽  
Strain Components ◽  
Frequency Components ◽  
Electronic Assemblies

This paper focusses on health monitoring of electronic assemblies under vibration load of 14 G until failure at an ambient temperature of 55 degree Celsius. Strain measurements of the electronic assemblies were measured using the voltage outputs from the strain gauges which are fixed at different locations on the Printed Circuit Board (PCB). Various analysis was conducted on the strain signals include Time-frequency analysis (TFA), Joint Time-Frequency analysis (JTFA) and Statistical techniques like Principal component analysis (PCA), Independent component analysis (ICA) to monitor the health of the packages during the experiment. Frequency analysis techniques were used to get a detailed understanding of the different frequency components before and after the failure of the electronic assemblies. Different filtering algorithms and frequency quantization techniques gave insight about the change in the frequency components with the time of vibration and the energy content of the strain signals was also studied using the joint time-frequency analysis. It is seen that as the vibration time increases the occurrence of new high-frequency components increases and further the amplitude of the high-frequency components also has increased compared to the before failure condition. Statistical techniques such as PCA and ICA were primarily used to reduce the dimensions of the larger data sets and provide a pattern without losing the different characteristics of the strain signals during the course of vibration of electronic assemblies till failure. This helps to represent the complete behavior of the electronic assemblies and to understand the change in the behavior of the strain components till failure. The principal components which were calculated using PCA discretely separated the before failure and after failure strain components and this behavior were also seen by the independent components which were calculated using the Independent Component Analysis (ICA). To quantify the prognostics and hence the health of the electronic assemblies, different statistical prediction algorithms were applied to the coefficients of principal and independent components calculated from PCA and ICA analysis. The instantaneous frequency of the strain signals was calculated and PCA and ICA analysis on the instantaneous frequency matrix was done. The variance of the principal components of instantaneous frequency showed an increasing trend during the initial hours of vibration and after attaining a maximum value it then has a decreasing trend till before failure. During the failure of components, the variance of the principal component decreased further and attained a lowest value. This behavior of the instantaneous frequency over the period of vibration is used as a health monitoring feature.

scholarly journals Wavlet Decomposition based Diagnostic for Structural Health Monitoring on Metallic Aircrafts: Case of Crack Triangulation and Corrosion Detection

2020 ◽  
Vol 4 (1) ◽  
Author(s):  
Hamza Boukabache ◽  
Christophe Escriba ◽  
Sabeha Zedek ◽  
Jean-Yves Fourniols
Keyword(s):  
Structural Health Monitoring ◽  
Frequency Analysis ◽  
Health Monitoring ◽  
Control Method ◽  
Temporal Frequency ◽  
Time Frequency ◽  
Structural Health ◽  
Acoustic Signatures ◽  
Integrated Modular Avionics ◽  
Pulse Echo

This work focus on the structural health monitoring of aircrafts parts specimen structures made of 2024 Aluminum alloys using a reliable Joint Time Frequency Analysis calculation (Joint Temporal Frequency Analysis). In this paper we demonstrate the feasibility of a new non destructive control method capable to probe very large structures within a short time. The method we developed is based through a wide piezoelectric sensors network on a smart comparison between two acoustic signatures: the healthy structure response captured before the commissioning of the plane and “an after flight” response. The sensors network exploits the capability of piezoelectric patches to generate/measure specific Lamb wave’s modes. The system is therefore dynamically configured to localize mechanicals flaws using a triangulation algorithm that operates using different techniques like pitch-catch and pulse-echo. The aim of this paper is to highlight a methodology that is currently being integrated into reconfigurable qualified and certified hardware architecture. The idea behind is to interface the airplane's structure to an integrated modular avionics calculator (IMA).An analytic study is performed and tests to prove the proposed method feasibility on corroded and damaged structures specimens are provided at the end of this paper.

scholarly journals Structural Health Monitoring Using Machine Learning and Cumulative Absolute Velocity Features

2021 ◽  
Vol 11 (12) ◽  
pp. 5727
Author(s):  
Sifat Muin ◽  
Khalid M. Mosalam
Keyword(s):  
Machine Learning ◽  
Logistic Regression ◽  
Structural Health Monitoring ◽  
Health Monitoring ◽  
Degree Of Freedom ◽  
Absolute Velocity ◽  
Support Vector ◽  
Damage State ◽  
Structural Health ◽  
Cumulative Absolute Velocity

Machine learning (ML)-aided structural health monitoring (SHM) can rapidly evaluate the safety and integrity of the aging infrastructure following an earthquake. The conventional damage features used in ML-based SHM methodologies face the curse of dimensionality. This paper introduces low dimensional, namely, cumulative absolute velocity (CAV)-based features, to enable the use of ML for rapid damage assessment. A computer experiment is performed to identify the appropriate features and the ML algorithm using data from a simulated single-degree-of-freedom system. A comparative analysis of five ML models (logistic regression (LR), ordinal logistic regression (OLR), artificial neural networks with 10 and 100 neurons (ANN10 and ANN100), and support vector machines (SVM)) is performed. Two test sets were used where Set-1 originated from the same distribution as the training set and Set-2 came from a different distribution. The results showed that the combination of the CAV and the relative CAV with respect to the linear response, i.e., RCAV, performed the best among the different feature combinations. Among the ML models, OLR showed good generalization capabilities when compared to SVM and ANN models. Subsequently, OLR is successfully applied to assess the damage of two numerical multi-degree of freedom (MDOF) models and an instrumented building with CAV and RCAV as features. For the MDOF models, the damage state was identified with accuracy ranging from 84% to 97% and the damage location was identified with accuracy ranging from 93% to 97.5%. The features and the OLR models successfully captured the damage information for the instrumented structure as well. The proposed methodology is capable of ensuring rapid decision-making and improving community resiliency.

Time–frequency and time–scale analyses for structural health monitoring

2006 ◽  
Vol 365 (1851) ◽  
pp. 449-477 ◽  
Author(s):  
Wiesław J Staszewski ◽  
Amy N Robertson
Keyword(s):  
Signal Processing ◽  
Structural Health Monitoring ◽  
Damage Detection ◽  
Health Monitoring ◽  
Time Scale ◽  
Time Frequency ◽  
Structural Health ◽  
Variant Analysis

Signal processing is one of the most important elements of structural health monitoring. This paper documents applications of time-variant analysis for damage detection. Two main approaches, the time–frequency and the time–scale analyses are discussed. The discussion is illustrated by application examples relevant to damage detection.

scholarly journals Application of Machine Learning with Impedance Based Techniques for Structural Health Monitoring of Civil Infrastructure

2019 ◽  
Vol 8 (6S4) ◽  
pp. 1139-1148
Keyword(s):  
Machine Learning ◽  
Signal Processing ◽  
Structural Health Monitoring ◽  
Damage Detection ◽  
Soft Computing ◽  
Health Monitoring ◽  
Multiple Cracks ◽  
Civil Infrastructure ◽  
Structural Health ◽  
Soft Computing Techniques

Increased attentiveness on the environmental and effects of aging, deterioration and extreme events on civil infrastructure has created the need for more advanced damage detection tools and structural health monitoring (SHM). Today, these tasks are performed by signal processing, visual inspection techniques along with traditional well known impedance based health monitoring EMI technique. New research areas have been explored that improves damage detection at incipient stage and when the damage is substantial. Addressing these issues at early age prevents catastrophe situation for the safety of human lives. To improve the existing damage detection newly developed techniques in conjugation with EMI innovative new sensors, signal processing and soft computing techniques are discussed in details this paper. The advanced techniques (soft computing, signal processing, visual based, embedded IOT) are employed as a global method in prediction, to identify, locate, optimize, the damage area and deterioration. The amount and severity, multiple cracks on civil infrastructure like concrete and RC structures (beams and bridges) using above techniques along with EMI technique and use of PZT transducer. In addition to survey advanced innovative signal processing, machine learning techniques civil infrastructure connected to IOT that can make infrastructure smart and increases its efficiency that is aimed at socioeconomic, environmental and sustainable development.

Sign in / Sign up

Export Citation Format

Share Document