Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Anomaly detection is an important problem that has been researched within diverse research areas and application domains. Many anomaly detection techniques have been specifically developed for certain application domains, while others are more generic. This survey tries to provide a structured and comprehensive overview of the research on anomaly detection. We have grouped existing techniques into different categories based on the underlying approach adopted by each technique. For each category we have identified key assumptions, which are used by the techniques to differentiate between normal and anomalous behavior. When applying a given technique to a particular domain, these assumptions can be used as guidelines to assess the effectiveness of the technique in that domain. For each category, we provide a basic anomaly detection technique, and then show how the different existing techniques in that category are variants of the basic technique. This template provides an easier and more succinct understanding of the techniques belonging to each category. Further, for each category, we identify the advantages and disadvantages of the techniques in that category. We also provide a discussion on the computational complexity of the techniques since it is an important issue in real application domains. We hope that this survey will provide a better understanding of the different directions in which research has been done on this topic, and how techniques developed in one area can be applied in domains for which they were not intended to begin with.

/pdf/anomaly-detection-a-survey-1x33g9m0a3.pdf

Anomaly detection: A survey

物件導向軟體之架構(Object-Oriented Software Construction)探討

A review on machinery diagnostics and prognostics implementing condition-based maintenance

This report contains a review of the technical literature concerning the detection, location, and characterization of structural damage via techniques that examine changes in measured structural vibration response. The report first categorizes the methods according to required measured data and analysis technique. The analysis categories include changes in modal frequencies, changes in measured mode shapes (and their derivatives), and changes in measured flexibility coefficients. Methods that use property (stiffness, mass, damping) matrix updating, detection of nonlinear response, and damage detection via neural networks are also summarized. The applications of the various methods to different types of engineering problems are categorized by type of structure and are summarized. The types of structures include beams, trusses, plates, shells, bridges, offshore platforms, other large civil structures, aerospace structures, and composite structures. The report describes the development of the damage-identification methods and applications and summarizes the current state-of-the-art of the technology. The critical issues for future research in the area of damage identification are also discussed.

/pdf/damage-identification-and-health-monitoring-of-structural-3s8adkph2w.pdf

Damage identification and health monitoring of structural and mechanical systems from changes in their vibration characteristics: A literature review

Real world structures, such as bridges and skyscrapers, are often subjected to dynamic loading and changing environments. It seems prudent to measure high resolution vibration data, in order to perform accurate damage detection and to validate and update the models and knowledge about the operating structure (aka finite element models). Many existing vibration measurement methods could be either low resolution (e.g., accelerometers or strain gauges), and time and labor consuming to deploy in field (e.g., laser interferometry). Previous work by Yang et al. has shown that low-cost regular digital video cameras enhanced by advanced computer vision and machine learning algorithms can extract very high resolution dynamic information about the structure and perform damage detection at novel scales in an relatively efficient and unsupervised manner. More interestingly this work used a machine learning pipeline that made minimal assumptions about lighting conditions or the nature of the structure in order to perform modal decomposition. The technique is currently limited to two dimensions if only one digital video camera is used. This paper uses light field imagers - a new camera system that captures the direction light entered the camera - to make depth measurements of scenes and extend the modal analysis technique proposed in Yang et al. to three dimensions. The new method is verified experimentally on vibrating cantilever beams with out of plane vibration, whose full-field modal parameters are extracted from the light field measurements. The experimental results are discussed and some limitations are pointed out for future work.

Light Field Imaging of Three-Dimensional Structural Dynamics

The Info-Gap Decision Theory (IGDT) is here adopted to assess the robust- ness of a technique aimed at identifying the optimal excitation signal within a structural health monitoring (SHM) procedure. Given limited system response measurements and ever-present physical limits on the level of excitation, the ultimate goal of the mentioned technique is to improve the detectability of the damage increasing the difference between measurable outputs of the undamaged and damaged system. In particular, a 2 DOF mass- spring-damper system characterized by the presence of a nonlinear stiffness is considered. Uncertainty is introduced within the system under the form of deviations of its parameters (mass, stiffness, damping ratio...) from their nominal values. Variations in the performance of the mentioned technique are then evaluated both in terms of changes in the estimated difference between the responses of the damaged and undamaged system and in terms of deviations of the identified optimal input signal from its nominal estimation. Finally, plots of the performances of the analyzed algorithm for different levels of uncertainty are ob- tained, showing which parameters are more sensitive to the presence of uncertainty and thus enabling a clear evaluation of its robustness.

/pdf/on-assessing-the-robustness-of-an-input-signal-optimization-52zplj9un9.pdf

On assessing the robustness of an input signal optimization algorithm for damage detection: the Info-Gap Decision Theory approach

This paper introduces a new method for Structural Health Monitoring using error functions computed from guided waves reflected from damage. The approach is experimentally tested on anisotropic specimens such as composite plates. The baseline and test signals of each sensing path (between two PZT transducers) are measured and the energy of the scattered signal for each path is calculated in a given frequency range. Assuming that there is damage in the evaluated position, the wave will reflect at this point and travel to the next transducer. According to the distance between the first transducer to the evaluated point plus the distance between same point to the second transducer (pitch-catch configuration) the time-of-flight is calculated for each grid point on the structure. The wave speeds in anisotropic specimens are propagation direction dependent. The wave speed for different angles were experimentally computed and incorporated in the algorithm in order to calculate the proper time-offlight. The energy of the scattered signal is computed in a time range based on the time of flight of each analyzed position. Finally, a resultant error function for an estimation of the damage location in the monitoring area is applied. As the error function is based on the interference of the damage in the propagation of guided waves, the higher value of the error implies the less likelihood of damage in that position. An image is generated with an error value for each mesh position in the plate. This error function compares the energy in the given ranges for each pair of transducers. The experiment was performed in a 500x500x2mm carbon/epoxy composite formed by 10 plainweave layers with 9 PZT transducers in the surface. The resultant error function at each driving frequency is calculated as a sum of all error functions. In addition, several frequencies were tested and the results for each one were combined in order to get a better result. doi: 10.12783/SHM2015/223

Adaptive Reverberation Suppression Techniques for SHM in Composite Materials

This paper presents a guided wave structural health monitoring (SHM) technique, based on the use of piezoelectric active-sensors, used to determine the structural integrity of stiffened aluminum plates. For damage detection, the transmitted power between piezoelectric transducers used for Lamb wave propagation is utilized to analyze the extent of damage in the structure. Damage initiation and propagation were successfully monitored with the guided wave technique. Overall, these methods yielded sufficient damage detection capability to warrant further investigation into field deployment. This paper summarizes considerations needed to design such SHM systems, experimental procedures and results, and recommendations that can be used as guidelines for future investigations.

Charles R. Farrar

Papers

Light Field Imaging of Three-Dimensional Structural Dynamics

On assessing the robustness of an input signal optimization algorithm for damage detection: the Info-Gap Decision Theory approach

Adaptive Reverberation Suppression Techniques for SHM in Composite Materials

Structural Damage Identification in Stiffened Plate Fatigue Specimens Using Piezoelectric Active Sensing

Integrated wireless powering and data interrogation for civil infrastructure monitoring