Papers by Author: Keith Worden

Paper TitlePage

Authors: Andrew Spencer, Keith Worden, Gareth Pierce
Abstract: When a metal or composite structure begins to fail, for example due to high cycle fatigue, acoustic emissions caused by the propagation of cracks give rise to bursts of ultrasonic waves travelling through the structure. The health of a structure can be monitored by means of sensors which detect these waves. Acoustic emissions are often generated in experiments by breaking a pencil lead against the surface of the structure in a standardised way but the forces that this imparts are not well understood at present. A Local Interaction Simulation Approach (LISA) algorithm has been implemented to simulate the propagation of ultrasonic waves. This code has been validated against experiments in previous work and has been shown to accurately reproduce the propagation of Lamb waves (including reflections and dispersion etc.) within thin-plate like structures. This paper deals with the use of the LISA code to characterise the forces associated with standard pencil lead breaks. The displacement due to waves emanating from a break is measured and a Differential Evolution (DE) optimisation scheme is used to find the optimal profile of forcing to match the simulation with experiment.
Authors: Rhys Pullin, James J. Hensman, Karen M. Holford, Keith Worden, S.L. Evans
Abstract: Acoustic emission monitoring was completed on a painted aerospace grade steel landing gear component undergoing fatigue loading until rupture. A post-test linear location analysis of the collected signals revealed eleven groups where high activity (greater than 2000 hits) occurred within a defined location, three of which corresponded in location to the position of fracture and final rupture of the specimen. Feature data, such as amplitude, rise-time, energy etc. were used to describe the identified signals in each group. A dimension reduction through principal component analysis of the feature data of all groups was performed. Results showed that high amplitude signals associated with four groups of signals arising from noise could be separated from the fracture groups. However four groups not associated with noise or the known positions of the fracture groups were not separable from the signals attributed to fractures. The paint layer of the specimen was removed and a magnetic particle investigation was completed that showed these four groups coincided with regions of additional fracture in the component.
Authors: Rhys Pullin, Mark J. Eaton, James J. Hensman, Karen M. Holford, Keith Worden, S.L. Evans
Abstract: This work forms part of a larger investigation into fracture detection using acoustic emission (AE) during landing gear airworthiness testing. It focuses on the use of principal component analysis (PCA) to differentiate between fracture signals and high levels of background noise. An artificial acoustic emission (AE) fracture source was developed and additionally five sources were used to generate differing AE signals. Signals were recorded from all six artificial sources in a real landing gear component subject to no load. Further to this, artificial fracture signals were recorded in the same component under airworthiness test load conditions. Principal component analysis (PCA) was used to automatically differentiate between AE signals from different source types. Furthermore, successful separation of artificial fracture signals from a very high level of background noise was achieved. The presence of a load was observed to affect the ultrasonic propagation of AE signals.
Authors: M.T.H. Sultan, Alma Hodzic, W.J. Staszewski, Keith Worden
Abstract: The ultimate objective of the current programme of work is to detect and quantify low-velocity impact damage in structures made from composite materials. There are many situations in the use of composites where an impact does not result in perforation of the material but causes damage that may not be visible, yet still causes a substantial reduction in structural properties. Impacts that do not cause perforation are usually termed low-velocity. When a composite structure undergoes such impacts, it is important to know the type and level of damage and assess the residual strength. In this study, following a systematic series of experiments on the induction of impact damage in composite specimens, Scanning Electron Microscopy (SEM) was used to inspect the topographies of the specimens at high magnification. Matrix cracking, fibre fracture, fibre pullout and delamination were the types of damage observed in the composite laminates after the low-velocity impacts. The study also conducted a (very) preliminary correlation between the damage modes and the impact energy.
Authors: Ifigeneia Antoniadou, Nikolaos Dervilis, Robert J. Barthorpe, Graeme Manson, Keith Worden
Abstract: The paper summarises some advanced damage detection approaches used for Structural Health Monitoring (SHM) and Condition Monitoring (CM) of wind turbine systems. In the signal processing part, recent time-frequency analysis methods will be presented and examples of their application on condition monitoring of gearboxes will be given. In the pattern recognition part, examples of damage detection in blades will be used to introduce different algorithms for novelty detection.
Authors: Keith Worden, Graeme Manson, D.J. Allman
Authors: Nikolaos Dervilis, A.C.W. Creech, A.E. Maguire, Ifigeneia Antoniadou, R.J. Barthorpe, Keith Worden
Abstract: Reliability of offshore wind farms is one of the key areas for the successful implementation of these renewable power plants in the energy arena. Failure of the wind turbine (WT) in general could cause massive financial losses but especially for structures that are operating in offshore sites. Structural Health Monitoring (SHM) of WTs is essential in order to ensure not only structural safety but also avoidance of overdesign of components that could lead to economic and structural inefficiency. A preliminary analysis of a machine learning approach in the context of WT SHM is presented here; it is based on results from a Computational Fluid Dynamics (CFD) model of Lillgrund Wind farm. The analysis is based on neural network regression and is used to predict the measurement of each WT from the measurements of other WTs in the farm. Regression model error is used as an index of abnormal response.
Authors: Keith Worden, Graeme Manson, Cecilia Surace
Abstract: The object of this paper is to illustrate the use of novelty detection techniques in Structural Health Monitoring (SHM) by the consideration of a number of case studies of varying complexity, from a simple lumped-mass system to an FE model of an offshore structure to an experimental study of an aircraft wing.
Authors: Keith Worden, W.E. Becker, Manuela Battipede, Cecilia Surace
Abstract: This paper concerns the analysis of how uncertainty propagates through large computational models like finite element models. If a model is expensive to run, a Monte Carlo approach based on sampling over the possible model inputs will not be feasible, because the large number of model runs will be prohibitively expensive. Fortunately, an alternative to Monte Carlo is available in the form of the established Bayesian algorithm discussed here; this algorithm can provide information about uncertainty with many less model runs than Monte Carlo requires. The algorithm also provides information regarding sensitivity to the inputs i.e. the extent to which input uncertainties are responsible for output uncertainty. After describing the basic principles of the Bayesian approach, it is illustrated via two case studies: the first concerns a finite element model of a human heart valve and the second, an airship model incorporating fluid structure interaction.
Authors: Graeme Manson, Gareth Pierce, Keith Worden, Daley Chetwynd
Abstract: This paper considers the performance of radial basis function neural networks for the purpose of data classification. The methods are illustrated using a simple two class problem. Two techniques for reducing the rate of misclassifications, via the introduction of an “unable to classify” label, are presented. The first of these considers the imposition of a threshold value on the classifier outputs whilst the second considers the replacement of the crisp network weights with interval ranges. Two network training techniques are investigated and it is found that, although thresholding and uncertain weights give similar results, the level of variability of network performance is dependent upon the training approach
Showing 1 to 10 of 35 Paper Titles