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

Structures usually have complex geometries, material properties, and boundary conditions, and exhibit spatially local, temporally transient, dynamic behaviors. High spatial and temporal resolution vibration measurements and modeling are thus required for high-fidelity characterization, analysis, and prediction of the structure’s dynamic phenomena. For example, high spatial density mode shapes are needed for accurate vibration-based damage localization. Also, higher order (frequency) vibration modes typically contain local structural features that are essential for high-fidelity modeling of the structure’s dynamics. In addition, while it is possible to build a highly-refined mathematical model (e.g., a finite element model) of the structure, it needs to be experimentally validated and updated with high-resolution vibration measurements. However, it is a significant challenge to obtain high-resolution vibration measurements using traditional techniques. For example, the widely-used accelerometers and straingauge sensors can only provide low spatial resolution measurements. Laser vibrometers provide high-resolution measurements, but are expensive and make sequential measurements that are time-consuming. On the other hand, photogrammetry as an alternative non-contact optical measurement method uses passive white-light imaging of digital video cameras that are relatively low-cost, agile, and provide non-contact, high spatial resolution, simultaneous, measurements where every pixel effectively becomes a measurement point on the structure. A new framework is developed for the blind extraction and realistic visualization of the full-field, high-resolution, dynamics behaviors of an operating structure from only its digital video measurements, possibly temporally-aliased (sub-Nyquist), using video motion manipulation and unsupervised machine learning techniques. This highresolution, full-field structural dynamics characterization framework addresses a variety of problems that traditionally have been challenging, including the ability to localize minute, non-visible, structural damage at a pixel resolution. Laboratory experiments on bench-scale building structures, cantilever beams, and stay cables are demonstrated.

Full-field Imaging and Modeling of Structural Dynamics with Digital Video Cameras

The process of implementing a damage identification strategy for aerospace systems is referred to as structural health monitoring (SHM). Here, damage is defined as changes to the material and/or geometric properties of these systems, including changes to the boundary conditions and system connectivity, which adversely affect the system's current or future performance. A wide variety of highly effective local non-destructive evaluation tools are available for such monitoring. However, the majority of SHM research conducted over the last 30 years has attempted to identify damage in structures on a more global basis and in a more autonomous manner. Recent research has begun to recognize that the SHM problem is fundamentally one of statistical pattern recognition, and a strategy to address this problem as it applies to aerospace structures is described herein. To conclude this discussion, technical challenges that must be addressed if SHM is to gain wider application in the aerospace industry are summarized.


Keywords:

damage detection;
data normalization;
feature extraction;
health and usage monitoring;
operational evaluation;
pattern recognition;
sensors;
dtatistical classification;
structural health monitoring;
pattern recognition

Nondestructive Evaluation of Structures

The process of implementing a damage detection strategy for aerospace, civil and mechanical engineering infrastructure is referred to as structural health monitoring (SHM). Here damage is defined as changes to the material and/or geometric properties of these systems, including changes to the boundary conditions and system connectivity, which adversely affect the system's current or future performance. Our approach is to address the SHM problem in the context of a statistical pattern recognition paradigm (Farrar, Nix and Doebling, 2001). In this paradigm, the process can be broken down into four parts: (1) Operational Evaluation, (2) Data Acquisition, (3) Feature Extraction, and (4) Statistical Model Development for Feature Discrimination. When one attempts to apply this paradigm to data from 'real-world' structures, it quickly becomes apparent that data cleansing, normalization, fusion and compression, which can be implemented with either hardware or software, are inherent in Parts 2-4 of this paradigm. The authors believe that all approaches to SHM, as well as all traditional non-destructive evaluation procedures (e.g. ultrasonic inspection, acoustic emissions, active thermography) can be cast in the context of this statistical pattern recognition paradigm. It should be noted that the statistical modeling portion of the structural health monitoring process has received themore » least attention in the technical literature. The algorithms used in statistical model development usually fall into the three categories of group classification, regression analysis or outlier detection. The ability to use a particular statistical procedure from one of these categories will depend on the availability of data from both an undamaged and damaged structure. This paper will discuss each portion of the SHM statistical pattern recognition paradigm.« less

A statistical pattern recognition paradigm for structural health monitoring

A nonlinear time series approach is presented to detect damage in systems by using a state-space reconstruction to infer the geometrical structure of a deterministic dynamical system from observed time series response at multiple locations. The unique contribution of this approach is using a Multivariate Autoregressive (MAR) model of a baseline condition to predict the state space, where the model encodes the embedding vectors rather than scalar time series. A hypothesis test is established that the MAR model will fail to predict future response if damage is present in the test condition, and this test is investigated for robustness in the context of operational and environmental variability. The applicability of this approach is demonstrated using acceleration time series from a base-excited 3-story frame structure.

Autoregressive modeling with state-space embedding vectors for damage detection under operational and environmental variability

In order to realize the wide-scale deployment of high-endurance, unattended mobile sensing technologies, it is
vital to ensure the self-preservation of the sensing assets Deployed mobile sensor nodes face a variety of physical
security threats including theft, vandalism and physical damage Unattended mobile sensor nodes must be able to
respond to these threats with control policies that facilitate escape and evasion to a low-risk state In this work the
Precision Immobilization Technique (PIT) problem has been considered The PIT maneuver is a technique that a
pursuing, car-like vehicle can use to force a fleeing vehicle to abruptly turn ninety degrees to the direction of travel The
abrupt change in direction generally causes the fleeing driver to lose control and stop The PIT maneuver was originally
developed by law enforcement to end vehicular pursuits in a manner that minimizes damage to the persons and property
involved It is easy to imagine that unattended autonomous convoys could be targets of this type of action by adversarial
agents This effort focused on developing control policies unattended mobile sensor nodes could employ to escape,
evade and recover from PIT-maneuver-like attacks The development of these control policies involved both simulation
as well as small-scale experimental testing The goal of this work is to be a step toward ensuring the physical security of
unattended sensor node assets

/pdf/escape-and-evade-control-policies-for-ensuring-the-physical-2xa7sob1ld.pdf

Charles R. Farrar

Papers

Full-field Imaging and Modeling of Structural Dynamics with Digital Video Cameras

Nondestructive Evaluation of Structures

A statistical pattern recognition paradigm for structural health monitoring

Autoregressive modeling with state-space embedding vectors for damage detection under operational and environmental variability

Escape and evade control policies for ensuring the physical security of nonholonomic, ground-based, unattended mobile sensor nodes