A novel machine-learning approach for structural state identification using ultrasonic guided waves

Roy, Surajit; Chang, Fu‐Kuo; Lee, S; Pollock, Patrick; Janapati, Vishnuvardhan

doi:10.1201/b16387-48

Cited by 14 publications

(10 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Current SHM and DP research efforts have focused on utilizing either, ground-based (pre-flight) or in-flight modal testing [6,7], ultrasonic guided waves [8,9], or acoustic emission monitoring [10,11], using sparse sensor arrays that rely on large detailed correlated computational models. These methods have shortfalls that make it unrealistic for actual flight vehicles having large, complex internal structures assembled using skins, stringers, spars and frames, operating in complex, varying environments.…”

Section: Review Of the Literaturementioning

confidence: 99%

Detecting Damage in a UAV Composite Wing Spar Using Distributed Fiber Optic Strain Sensors

Martins

Kosmatka

2015

56th AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference

View full text Add to dashboard Cite

The primary structure of an unmanned air vehicle (UAV) wing is commonly composed of outer wing skins attached (bonded or fastened) to a main wing spar both composed of advanced composite materials. The main wing spar is designed to take the primary bending and transverse shear loads, whereas the wing skin is designed to take the wing torsion loads. Since these aircraft are developed for long range continuous (surveillance) flight with minimal time for ground inspection, it is desirable to have an in-flight structural health monitoring (SHM) system for insuring the structural integrity of the wing structural components and attachments. In the current study an SHM computational and experimental testbed is developed for analyzing the structural integrity of a composite wing spar, where fixturing is included to produce realistic wing bending and torsion natural frequencies and corresponding coupled bending-torsion vibration modes. The wing spar is instrumented using Bragg graded fiber optic distributed strain sensors along the spar length on three sides to determine the axial strain and bending modal curvature distributions and in a wavy configuration on the fourth side in an attempt to monitor spar twisting behavior. Initially, finite element studies of a damaged composite wing spar are performed to determine the modal curvature distributions for evaluation of numerous published SHM approaches. Finally, an experimental study is performed where the free vibration response of the undamaged and damaged wing spar is monitored based upon either an impact excitation or an initial displacement. While these results clearly show the damage using the computational models, there are difficulties locating damage using the distributed fiber optic strain sensors.

show abstract

Section: Review Of the Literaturementioning

confidence: 99%

Detecting Damage in a UAV Composite Wing Spar Using Distributed Fiber Optic Strain Sensors

Martins

Kosmatka

2015

56th AIAA/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference

View full text Add to dashboard Cite

show abstract

“…), variation in boundary conditions, sensor and structural aging, and measurement noise among others, as well as the sensing network layout itself. [16][17][18][19][20][21][22][23][24][25] Usually, the uncertainties due to environmental effects have the largest impact on the system performance. An SHM system needs to be robust to uncertainties, but sensitive enough to detect the required minimum damage even when the sensor data are ''corrupted'' by these uncertainties.…”

Section: Introductionmentioning

confidence: 99%

“…Toward this end, statistical tools and machine learning algorithms are used to effectively tackle the detection and estimation problems. 16,23,[26][27][28][29][30][31][32] SHM technologies, oftentimes in combination with appropriate NDE and/or health and usage monitoring systems (HUMS), along with sophisticated data management systems and life-prediction models constitute the required steps for the transition to condition-based maintenance (CBM). Typically, SHM methods involve the detection, monitoring, evaluation, and assessment of an adverse event that may affect the structural health state.…”

Section: Introductionmentioning

confidence: 99%

Damage detection sensitivity characterization of acousto-ultrasound-based structural health monitoring techniques

Janapati

Kopsaftopoulos

et al. 2016

Structural Health Monitoring

106

117

View full text Add to dashboard Cite

Reliability quantification is a critical and necessary process for the evaluation and assessment of any inspection technology that may be classified either as a nondestructive evaluation or structural health monitoring technique. Based on the sensitivity characterization of nondestructive evaluation techniques, appropriate processes have been developed and established for the reliability quantification of their performance with respect to damage/flaw detection in materials or structures. However, in the case of structural health monitoring methods, no such well-defined and general applicable approaches have been established for neither active nor passive sensing techniques that allow for their accurate reliability quantification. The objective of this study is to characterize the sensitivity of active-sensing acousto-ultrasound-based structural health monitoring techniques with respect to damage detection, as well as to identify the parameters that influence their sensitivity. With such an understanding, it is believed that adequate quantitative methods could then be established to enable the practical use of acousto-ultrasound structural health monitoring methods in the aerospace and mechanical engineering communities. In order to evaluate the sensitivity of a pre-selected active-sensing acousto-ultrasound structural health monitoring system, both numerical simulations and experiments were performed on 30 aluminum coupons each outfitted with a pair of lead zirconate titanate piezoelectric sensors/actuators. A damage index versus damage size relationship was investigated numerically and experimentally to assess the applicability of the traditional nondestructive evaluation linear regression framework for probability of detection for an active-sensing structural health monitoring system. The results of the study show that the position of each sensor–actuator pair with respect to a known damage location and the damage growth pattern are the two most critical parameters influencing the reliability of the same structural health monitoring system applied to identical structural components under the same environmental conditions.

show abstract

“…The experimental results showed that a satisfactory damage detection rate was achieved. Roy et al 7 presented a structural damage detection model based on an unsupervised learning technique. In this novelty detection approach, a neural-network-based sparse auto-encoder algorithm was integrated into a statistical outlier analysis method.…”

Section: Introductionmentioning

confidence: 99%

Unsupervised novelty detection–based structural damage localization using a density peaks-based fast clustering algorithm

Cha

Wang

2017

Structural Health Monitoring

121

View full text Add to dashboard Cite

Within machine learning, several structural damage detection and localization methods based on clustering and novelty detection methods have been proposed in the recent years in order to monitor mechanical and civil structures. In order to train a machine learning model, an unsupervised mode is preferred because it only requires sufficient normal data from the intact states of a structure for training, and the testing abnormal data from various damage states are generally quite rare. With an unsupervised training mode, the capability of detecting structural damage mainly depends on the identification of abnormal data from the testing data. This identification process is termed unsupervised novelty detection. The premise of unsupervised novelty detection is that a large volume of a normal data set is available first to train a normal model that is established by machine learning algorithms. Then, the trained normal model can be used to identify abnormal data from future testing data. In this article, a new structural damage detection and localization method is proposed using a density peaks-based fast clustering algorithm. In order to realize damage detection, the original density peaks-based fast clustering algorithm is modified to an unsupervised machine learning method by adding training and testing processes. Furthermore, to improve the performance of the proposed method, the Gaussian kernel function of radius is introduced to calculate the local density of data points, and a new damage-sensitive feature using a continuous wavelet transform is also proposed. Damage-sensitive features are extracted from the measured data through sensors installed on a laboratory-scale steel structure. Extensive experimental studies are carried out under various structural damage scenarios in order to validate the performance of the proposed method. The proposed density peaks-based fast clustering method shows satisfactory performance with regard to damage localization under various damage scenarios as compared to a traditional approach.

show abstract

A novel machine-learning approach for structural state identification using ultrasonic guided waves

Cited by 14 publications

References 2 publications

Detecting Damage in a UAV Composite Wing Spar Using Distributed Fiber Optic Strain Sensors

Detecting Damage in a UAV Composite Wing Spar Using Distributed Fiber Optic Strain Sensors

Damage detection sensitivity characterization of acousto-ultrasound-based structural health monitoring techniques

Unsupervised novelty detection–based structural damage localization using a density peaks-based fast clustering algorithm

Contact Info

Product

Resources

About