Miguel Angel Torres-Arredondo

Data-driven multivariate algorithms for damage detection and identification: Evaluation and comparison

This article is concerned with the experimental validation of a structural health monitoring methodology for damage detection and identification. Three different data-driven multivariate algorithms are considered here to obtain the baseline pattern. These are based on principal component analysis, independent component analysis and hierarchical non-linear principal component analysis. The contribution of this article is to examine and compare the three proposed algorithms that have been reported as reliable methods for damage detection and identification. The approach is based on a distributed piezoelectric active sensor network for the excitation and detection of structural dynamic responses. A woven multilayered composite plate and a simplified aircraft composite skin panel are used as examples to test the approaches. Data-driven baseline patterns are built when the structure is known to be healthy from wavelet coefficients of the structural dynamic responses. Damage is then simulated by adding masses at different positions of the structures. The data from the structure in different states (damaged or not) are then projected into the different models by each actuator in order to generate the input feature vectors of a self-organizing map from the computed components together with squared prediction error measures. All three methods are shown to be successful in detecting and classifying the simulated damages. At the end, a critical comparison is given in order to investigate the advantages and disadvantages of each method for the damage detection and identification tasks.

Data-driven methodology to detect and classify structural changes under temperature variations

This paper presents a methodology for the detection and classification of structural changes under different temperature scenarios using a statistical data-driven modelling approach by means of a distributed piezoelectric active sensor network at different actuation phases. An initial baseline pattern for each actuation phase for the healthy structure is built by applying multiway principal component analysis (MPCA) to wavelet approximation coefficients calculated using the discrete wavelet transform (DWT) from ultrasonic signals which are collected during several experiments. In addition, experiments are performed with the structure in different states (simulated damages), pre-processed and projected into the different baseline patterns for each actuator. Some of these projections and squared prediction errors (SPE) are used as input feature vectors to a self-organizing map (SOM), which is trained and validated in order to build a final pattern with the aim of providing an insight into the classified states. The methodology is tested using ultrasonic signals collected from an aluminium plate and a stiffened composite panel. Results show that all the simulated states are successfully classified no matter what the kind of damage or the temperature is in both structures.

Multivariate data-driven modelling and pattern recognition for damage detection and identification for acoustic emission and acousto-ultrasonics

Different methods are commonly used for non-destructive testing in structures; among others, acoustic emission and ultrasonic inspections are widely used to assess structures. The research presented in this paper is motivated by the need to improve the inspection capabilities and reliability of structural health monitoring (SHM) systems based on ultrasonic guided waves with focus on the acoustic emission and acousto-ultrasonics techniques. The use of a guided wave based approach is driven by the fact that these waves are able to propagate over relatively long distances, and interact sensitively and uniquely with different types of defect. Special attention is paid here to the development of efficient SHM methodologies. This requires robust signal processing techniques for the correct interpretation of the complex ultrasonic waves. Therefore, a variety of existing algorithms for signal processing and pattern recognition are evaluated and integrated into the different proposed methodologies. As a contribution to solve the problem, this paper presents results in damage detection and classification using a methodology based on hierarchical nonlinear principal component analysis, square prediction measurements and self-organizing maps, which are applied to data from acoustic emission tests and acousto-ultrasonic inspections. At the end, the efficiency of these methodologies is experimentally evaluated in diverse anisotropic composite structures.

Towards the Development of Predictive Models for the System Design and Modal Analysis of Acoustic Emission Based Technologies

Acoustic Emission (AE) techniques are used for the structural health monitoring (SHM) of civil, aeronautic and aerospace structures. In order to depart from the traditional reliance on parameter based analysis, AE diagnostic techniques require the analysis of wave propagation phenomena and the use of predictive modelling tools to improve the monitoring capabilities and provide reliable health monitoring. Additionally, modal based techniques offer potential for optimization of sensor networks in terms of sensor placement and number of sensors, increased source location accuracy and to get an insight into the source mechanisms. If the modes of propagation can be recognised in the received AE signals, then it would be possible to discriminate between damage types. On that account, the present paper develops two methodologies that are useful tools for the investigation and design of wave propagation based SHM systems established upon modal analysis. Firstly, a higher order plate theory for modelling disperse solutions in elastic and viscoelastic fibre-reinforced composites is proposed in order to investigate the radiation and attenuation of Lamb waves in anisotropic media. Second, spectral flat shell elements are used for the simulation of guided waves in shell structures. Numerical simulations and experiments validate the models and demonstrate that material anisotropy has a strong influence on the velocities, attenuation and acoustic energy for the different modes of propagation. It is expected that the presented methodologies may contribute to offer a higher computational efficiency and simplicity in comparison to traditional methods, and enable the design shortening time and cost of development of Lamb wave based damage detection systems for a rapid transfer from laboratory to in-service structures.

Susceptibility on the Strain Field Change as Function of the Coupling Between the Effect Produced by Damage Appearance and the Change in the Load Conditions

When strain sensors are used in order to gather valuable information about structural integrity, the main idea is to compare patterns in the strain field for the pristine conditions and possible damaged conditions. However, any change in the strain field caused by other conditions different from damage occurrence must be isolated from the analysis. In previous works the authors have demonstrated than even when the changes in the local strain field caused by a defect are very small and may go faded easily, it is possible to detect such small changes by using appropriate robust automated techniques. The authors have focused their attention in methodologies based on Principal Component Analysis (PCA) and some nonlinear extensions such as Hierarchical Nonlinear PCA (h-NLPCA) and the development of several unfolding and scaling techniques, which allows dealing with multiple load conditions. However, when the load conditions are very different and promotes big changes in the strain field, it is necessary to isolate such load conditions in order to uncoupling the effect of the damage occurrence and the effect of the severe change in load conditions. By means of automatic clustering techniques based on Self Organizing Maps (SOM), an Optimal Baseline Selection (OBS) technique was developed for damage detection based on strain measurements and strain field pattern recognition. doi: 10.12783/SHM2015/306

Structural Health Monitoring of Wind Turbine Blades Using Statistical Pattern Recognition

Wind turbine blades meet extremely complex loading cycles due to the random nature of wind conditions on their installation sites. Even though, when today available monitoring techniques have arrived at a technical level where it is possible to monitor the blades during operation, this fact still presents a series of technical and logistical challenges to be overcome when turbines are placed offshore. Additionally, offshore environment is tremendously severe and place high demands not only on the strength of the turbine blade structural design but also in the need of reliable blade monitoring systems. In this account, this paper presents an automated method for the assessment of the structural integrity of wind turbine blades. The purpose of this publication is to develop and present a methodology in order to tackle some of the challenges and problems to be solved before a full system can be proved capable of detecting critical damage in blades. The proposed system is based on an active sensor network where feature extraction, sensor data fusion and statistical baseline modelling are synergically evaluated for the purpose of damage detection within the context of pattern recognition-based structural health monitoring (SHM). Delamination fracture, one of the major damage modes in laminated composite materials, is introduced in several steps into the experimental specimen so that the proposed methodology can be evaluated. At the end, it is shown how damage and its development can be determined with the help of the proposed methodology. doi: 10.12783/SHM2015/304

Principal Component Analysis and Self-organizing Maps for Damage Detection and Classification under Temperature Variations

The use of statistical techniques for data driven has proven very useful in multivariable analysis as a pattern recognition approach. Among their multiple advantages such as data reduction, multivariable analysis and the definition of statistical models built with data from experimental trials, they provide robustness and allow avoiding the need of the development of physical models which sometimes are difficult for modelling especially when the system is complex. In this paper, a methodology based on Principal Component Analysis (PCA) is developed and used for building statistical baseline models comprising the dynamics from the monitored healthystructureunderdifferenttemperatureconditions.Inasecondstep, fortesting the proposed methodology, data from the structure at different structural states and under different temperature conditions are projected into the baseline models in order to obtain statistical measures (Scores and Q-index) which are included as feature vectors in a Self-Organizing Map for the damage detection and classification tasks. The methodology is evaluated using ultrasonic signals collected from an aluminium plate and a stiffened composite panel. Results show that all the simulated states are successfully classified no matter what the kind of damage or the temperature is present in both structures.

An Acousto-Ultrasonics Pattern Recognition Approach for Damage Detection Under Variable Temperature Conditions

Structural health monitoring (SHM) has emerged in the last few decades as an essential technology in order to improve the safety and maintainability of critical structures. Among the multiple techniques available for health monitoring, acoustoultrasonics (AU) offers the possibility of inspecting large areas of structures from a piezoelectric active sensor network with a relatively small number of sensors. Nevertheless, several points have to be taken into consideration before a robust method for damage detection can be developed. On the one hand, one of the major tasks for configuring an acousto-ultrasonics system is the selection of appropriate and robust signal processing and pattern recognition algorithms. On the other hand, the increase in complexity of the structures and variability in environmental and operational conditions makes damage detection an extremely challenging problem and points out the necessity for compensating these effects. This paper describes a health monitoring methodology combining the advantages of guided ultrasonic waves together with the compensation for temperature effects, the extraction of defect-sensitive features and sensor data-fusion for the purpose of carrying out a non-linear multivariate diagnosis of damage. Two-well known methods to compensate the temperature effects, namely Optimal Baseline Selection (OBS) and Optimal Signal Stretch (OSS), are investigated within the proposed methodology where the performance is assessed using Receiver Operating Characteristic (ROC) curves. Experimental results in a pipework demonstrate that the proposed methodology is a robust practical solution to compensate for temperature effects and improve the damage detection capabilities within the presented SHM system. doi: 10.12783/SHM2015/97