Graeme Manson

The fundamental axioms of structural health monitoring

Based on the extensive literature that has developed on structural health monitoring over the last 20 years, it can be argued that this field has matured to the point where several fundamental axioms, or gen eral principles, have emerged. The intention of this paper is to explicitly state and justify these axioms. In so doing, it is hoped that two subsequent goals are facilitated. First, the statement of such axioms will give new researchers in the field a starting point that alleviates the need to review the vast amounts of literature in this field. Second, the authors hope to stimulate discussion and thought within the community regarding these axioms.

The application of machine learning to structural health monitoring

In broad terms, there are two approaches to damage identification. Model-driven methods establish a high-fidelity physical model of the structure, usually by finite element analysis, and then establish a comparison metric between the model and the measured data from the real structure. If the model is for a system or structure in normal (i.e. undamaged) condition, any departures indicate that the structure has deviated from normal condition and damage is inferred. Data-driven approaches also establish a model, but this is usually a statistical representation of the system, e.g. a probability density function of the normal condition. Departures from normality are then signalled by measured data appearing in regions of very low density. The algorithms that have been developed over the years for data-driven approaches are mainly drawn from the discipline of pattern recognition, or more broadly, machine learning. The object of this paper is to illustrate the utility of the data-driven approach to damage identification by means of a number of case studies.

Uncertainty in structural dynamics

The effects of uncertainty are of growing concern in the design of engineering structures. The fact that the properties of the structure are uncertain implies that there is consequent uncertainty in the dynamic response. Similarly, there is inevitable manufacturing variability: mass-produced items are never identical. Indeed the properties of an individual system will change with time due to environmental conditions, loads, wear, etc. Uncertainty and variability raise issues concerning safety, reliability, quality of performance, worst-case behaviour and so on, and in turn these issues lead to demands for modelling methods which specifically include uncertainties in the properties of the structure. In the past, factors of safety might be introduced. However, the desire for greater efficiency, improved performance and reduced costs has led to a demand for improved computational methods, especially for high-cost structures. The goal is to apply such methods at the design stage to produce structures which are safe, reliable and have acceptable noise and vibration performance under all environmental and operating conditions which they are expected to encounter, and to produce designs which are robust with respect to manufacturing variability.

Identification of pre-sliding friction dynamics

The hysteretic nonlinear dependence of pre-sliding friction force on displacement is modeled using different physics-based and black-box approaches including various Maxwell-slip models, NARX models, neural networks, nonparametric (local) models and dynamical networks. The efficiency and accuracy of these identification methods is compared for an experimental time series where the observed friction force is predicted from the measured displacement. All models, although varying in their degree of accuracy, show good prediction capability of pre-sliding friction. Finally, we show that even better results can be achieved by using an ensemble of the best models for prediction.

Detection of defects in composite plates using Lamb waves and novelty detection

The problem of detecting damage in composite plates is addressed here using L amb waves and novelty detection. Damage can be inferred from the scattering and modification of the Lamb wavefield as it passes through a defect. In order to produce an automatic diagnostic tool which can operate on measured time data, the method of novelty detection is used. This depends on establishing a description of normality which then allows subsequent signals to be flagged as anomalous if they deviate from normal condition. Three methods of novelty detection are illustrated: two statistical methods and one neural. The methods are demonstrated on experimental data captured from two composite plates.

Application of ultrasonic Lamb wave techniques to the evaluation of advanced composite structures

Ultrasonic Lamb waves have been investigated extensively for damage detection in advanced composite materials. They are particularly suitable for proving thin plate structures of large area, where the Lamb wave approach offers a considerable saving in time over through-the-thickness inspection. However the potential complexity of the propagation can introduce significant difficulties to the technique. We present a review of work conducted at The University of Strathclyde in collaboration with several European partners into the feasibility of Lamb wave inspection. Specifically we will address issues of Lamb wave generation, propagation, defect interaction and detection.

Visualisation and Dimension Reduction of Acoustic Emission Data for Damage Detection

This paper is concerned with the compression or dimensional reduction of a number of acoustic emission (AE) signals generated by damage to a structure, namely, a box girder from a bridge. The object of the exercise is to visualise the data and thereby associate clusters tothe measured AE bursts. These can be used tohelp interpret the signals. Two methods are used to reduce the dimension: principal component analysis and Sammon mapping.

Environmental Effects on Lamb Wave Responses from Piezoceramic Sensors

Structural Health Monitoring using Lamb waves is based on the theory of elastic waves propagating in plates. Signal attenuation and mode conversion are used for damage detection. It is well known that the voltage amplitude from low-profile piezoceramic sensors used for inspection may change not only due to damage but also due to an inappropriate transducer coupling and temperature effects. The paper studies the effect of temperature on Lamb wave responses. The work involves a simple experiment in which two piezoelectric ceramics are bonded on an aluminium plate. The plate is exposed to various levels of temperature. A couple of simple features extracted from the Lamb wave responses are analysed to show the effect of the temperature changes. The plate was later subjected to damage and the series of tests and analysis repeated in order to give some comparison between the effects of damage and temperature change upon the responses. Several ways of how this problem can be circumvented using appropriate signal processing techniques are discussed.

Long-term stability of normal condition data for novelty detection

As a technique of diagnosing failure in structures and systems, the method of novelty detection shows considerable merit. The basis of the approach is simple: given measured data from normal condition of the structure, the diagnostic system builds an internal representation of the system normal condition in such a way that subsequent departures from this condition can be identified with confidence in a robust manner. The success or failure of the method is contingent on the accuracy of the description of normal condition. In many cases, the normal condition data may have quite a complex structure: for example, an aircraft may experience a wide range of ambient temperatures in the course of a single flight. Also, the operational loads experienced by the craft as a result of flight manoeuvres may have wide-ranging effects on the measured states. The object of the current paper is to explore the normal condition space for a simple benchmark monitoring system. The said system uses Lamb-wave inspection to diagnose damage in a composite plate. Both short-term and long-term experiments are carried out in order to examine the variations in normal condition as a result of run-in of the instrumentation and variations in ambient temperature. The exercise is not purely academic as the fiber-optic monitoring system is a serious candidate for a practical diagnostic system.

Bayesian sensitivity analysis of a model of the aortic valve.

Understanding the mechanics of the aortic valve has been a focus of attention for many years in the biomechanics literature, with the aim of improving the longevity of prosthetic replacements. Finite element models have been extensively used to investigate stresses and deformations in the valve in considerable detail. However, the effect of uncertainties in loading, material properties and model dimensions has remained uninvestigated. This paper presents a formal statistical consideration of a selected set of uncertainties on a fluid-driven finite element model of the aortic valve and examines the magnitudes of the resulting output uncertainties. Furthermore, the importance of each parameter is investigated by means of a global sensitivity analysis. To reduce computational cost, a Bayesian emulator-based approach is adopted whereby a Gaussian process is fitted to a small set of training data and then used to infer detailed sensitivity analysis information. From the set of uncertain parameters considered, it was found that output standard deviations were as high as 44% of the mean. It was also found that the material properties of the sinus and aorta were considerably more important in determining leaflet stress than the material properties of the leaflets themselves.