Robust Thresholds Research Articles

Distributed anomaly detection for industrial wireless sensor networks based on fuzzy data modelling

Modern infrastructure increasingly depends on large computerized systems for their reliable operation. Supervisory Control and Data Acquisition (SCADA) systems are being deployed to monitor and control large scale distributed infrastructures (e.g. power plants, water distribution systems). A recent trend is to incorporate Wireless Sensor Networks (WSNs) to sense and gather data. However, due to the broadcast nature of the network and inherent limitations in the sensor nodes themselves, they are vulnerable to different types of security attacks. Given the critical aspects of the underlying infrastructure it is an extremely important research challenge to provide effective methods to detect malicious activities on these networks. This paper proposes a robust and scalable mechanism that aims to detect malicious anomalies accurately and efficiently using distributed in-network processing in a hierarchical framework. Unsupervised data partitioning is performed distributively adapting fuzzy c-means clustering in an incremental model. Non-parametric and non-probabilistic anomaly detection is performed through fuzzy membership evaluations and thresholds on observed inter-cluster distances. Robust thresholds are determined adaptively using second order statistical knowledge at each evaluation stage. Extensive experiments were performed and the results demonstrate that the proposed framework achieves high detection accuracy compared to existing data clustering approaches with more than 96% less communication overheads opposed to a centralized approach.

Debris-flow susceptibility of upland catchments

Over the last three decades, many regional studies in mountain ranges under temperate climate revealed that it is possible to discriminate debris-flow and fluvial fans from morphometric indicators measured at the scale of the catchment and the fan itself. The most commonly used indicators are the Melton index (R), a normalized index of the gravitational energy of the catchment, and the fan slope (S). A wide range of thresholds have been proposed for discriminating purpose, but these are generally based on a small population of catchments and may be highly influenced by ambiguous fans included in the data set. A database of 620 upland catchments from several mountain ranges under temperate climate was compiled from the literature to propose robust discriminant morphometric thresholds for debris-flow versus fluvial responses. Linear discriminant analysis (LDA) and logistic regression (LR) were performed using the whole data set, and a leave-one-out cross-validation was used to evaluate performances of the models. Sensitivity and specificity scores obtained for LDA and LR were 0.96 and 0.73, and 0.95 and 0.75, respectively. It is also shown that the channel slope above which debris-flow is observed decreases with the gravitational energy of the catchment. Limitations of the morphometric discrimination are discussed.

New algorithm for CBR-RBR fusion with robust thresholds

The case-based reasoning(CBR) and rule-based reasoning(RBR) fusion systems include a diverse range of fusion methods and their tasks are characterized by interleaving combination of the reasoning procedures. Existing approaches cannot clarify the complex relationships between data from the knowledge sources nor uniformly represent the heterogeneous case and rule knowledge in one fusion space. As a result, existing approaches fail to solve system fragility due to knowledge uncertainty and reasoning unreliability. For the purpose of addressing the difficulties, a novel algorithm for CBR-RBR fusion with robust thresholds(CRFRT) is proposed. Heterogeneous case and rule knowledge are uniformly represented in one defined fusion unitary space. The robust thresholds have been achieved to distinguish the complex relationships between meta-knowledge in the fusion space and to enhance system capacity of knowledge identification. Furthermore, fusion reasoning strategies are constructed for CRFRT and its procedure based on which robust solution of the fusion reasoning problem is obtained. Finally, CRFRT is validated by benchmark problems in machine learning. Compared with other CBR and RBR approaches, the reasoning efficiency and accuracy are increased by 5% and 2.2% respectively. The variations of system accuracy are decreased by 2% to 3.8%. The above results show that the CRFRT algorithm boosts the system’s effectiveness and robustness. The proposed CRFRT can solve the fragility of complex intelligence decision system and give quality performance for fault diagnosis.

Robust Smoothing of Quantitative Genomic Data Using Second-Generation Wavelets and Bivariate Shrinkage

Recent high-throughput nucleotide sequencing technologies provide large amounts of quantitative genomic data, and thus, biologists currently need to process vast quantities of the data on a regular basis. The first step of the process is almost always smoothing of the data because biomedical data generally tend to contain a lot of noise. In this first step, classical wavelet transforms are widely used; however, the second-generation wavelet transform has not been used in biomedical studies. Smoothing based on the second-generation wavelets is more effective than classical wavelets-based methods because it employs data-dependent wavelet functions and does not require predefined explicit base functions. Since biomedical data usually lack regularity, it is more useful in biomedical research to use the second-generation wavelets than to use the classical wavelets. Therefore, we propose a novel smoothing method based on the second-generation wavelets and bivariate shrinkage, which enables to determine robust thresholds for wavelet-based smoothing, and apply it to synthetic and real genomic data. Experimental results demonstrate the effectiveness of the proposed method.

Robust Adaptive Thresholds under Additive and Multiplicative Disturbances

This paper deals with a set-membership approach for the fast computation of robust adaptive thresholds. Dynamic model based fault detection under bounded additive and multiplicative disturbances is considered. Based on recent results related to the design of stable interval observers for linear systems with additive time-varying zonotopic input bounds, an extension guaranteeing the robustness with respect to structured and bounded multiplicative disturbances is proposed. A sufficient condition not only ensuring the stability of the initial uncertain model but also the non divergence of the computed bounds (i.e. adaptive thresholds) is also given. Then, a numerical example illustrates the proposed scheme.

Bond Graph Model Based and Fuzzy Logic For Robust FDI of Mechatronic Systems

Fault diagnosis is crucial for ensuring the safe operation of complex engineering systems and avoiding to execute an unsafe behaviour. This paper deals with robust decision making (RDM) for fault detection of an electromechanical system by combining the advantages of Bond Graph (BG) modelling and Fuzzy logic reasoning. The proposed fault diagnosis method is implemented in two stages. In the first stage, the residuals are deduced from the BG model allowing to build a Fault Signature Matrix (FSM) according to the sensitivity of residuals to different parameters. In the second stage, the result of FSM and the robust residual thresholds are used by the fuzzy reasoning mechanism in order to evaluate a degree of detectability for each set of components. Finally, in order to make robust decision according to the detected fault component, an analysis is done between the output variables of the fuzzy system and components having the same signature in the FSM. The performance of the proposed fault diagnosis methodology is demonstrated through experimental data of an omnidirectional robot.

Threshold computation for fault detection in a class of discrete‐time nonlinear systems

AbstractIn this paper, we address the problem of designing robust thresholds for fault detection in discrete‐time nonlinear uncertain systems in the presence of process disturbances. Both constant and dynamic thresholds are proposed. For the computation of constant thresholds, a generalized framework based on signal norms is developed. Different kinds of constant thresholds are studied in the framework proposed. Using linear matrix inequalities (LMI) techniques, algorithms are derived for the computation of these thresholds. Similarly, the dynamic threshold is designed by deriving an inequality on the upper bound of the modulus of the residual signal. This inequality is based on the solution of discrete‐time nonlinear uncertain systems. The simulation examples illustrate that false alarms are successfully eliminated using the proposed thresholds. Copyright © 2010 John Wiley & Sons, Ltd.

Estimating Optimal Parameters for MRF Stereo from a Single Image Pair

This paper presents a novel approach for estimating the parameters for MRF-based stereo algorithms. This approach is based on a new formulation of stereo as a maximum a posterior (MAP) problem in which both a disparity map and MRF parameters are estimated from the stereo pair itself. We present an iterative algorithm for the MAP estimation that alternates between estimating the parameters while fixing the disparity map and estimating the disparity map while fixing the parameters. The estimated parameters include robust truncation thresholds for both data and neighborhood terms, as well as a regularization weight. The regularization weight can be either a constant for the whole image or spatially-varying, depending on local intensity gradients. In the latter case, the weights for intensity gradients are also estimated. Our approach works as a wrapper for existing stereo algorithms based on graph cuts or belief propagation, automatically tuning their parameters to improve performance without requiring the stereo code to be modified. Experiments demonstrate that our approach moves a baseline belief propagation stereo algorithm up six slots in the Middlebury rankings.

Dynamics of allosteric action in multisite protein modification

Protein functions in cells may be activated or modified by the attachment of several kinds of chemical groups. While protein phosphorylation, i.e., the attachment of a phosphoryl ( PO 3 - ) group, is the most studied form of protein modification, and is known to regulate the functions of many proteins, protein behavior can also be modified by nitrosylation, acetylation, methylation, etc. A protein can have multiple modification sites, and displays some form of transition only when enough sites are modified. In a previous paper we have modeled the generic equilibrium properties of multisite protein modification [R. Chignola, C. Dalla Pellegrina, A. Del Fabbro, E. Milotti, Physica A 371 (2006) 463] and we have shown that it can account both for sharp, robust thresholds and for information transfer between processes with widely separated timescales. Here we use the same concepts to expand that analysis starting from a dynamical description of multisite modification: we give analytical results for the basic dynamics and numerical results in an example where the modification chain is cascaded with a Michaelis–Menten step. We modify the dynamics and analyze an example with realistic phosphorylation/dephosphorylation steps, and give numerical evidence of the independence of the allosteric effect from the details of the attachment–detachment processes. We conclude that multisite protein modification is dynamically equivalent to the classic allosteric effect.

Dynamic threshold generators for robust fault detection in linear systems with parameter uncertainty

The problem of developing robust thresholds for fault detection is addressed. An inequality for the solution of a linear system with uncertain parameters is provided and is shown to be a valuable tool for developing dynamic threshold generators for fault detection. Such threshold generators are desirable for achieving robustness against model uncertainty in combination with sensitivity to small faults. The usefulness of the inequality is illustrated by developing an algorithm for detection of sensor faults in a turbofan engine. The proposed algorithm consists of a state observer with integral action. A dynamic threshold generator is derived under the assumption of parametric uncertainty in the process model. Successful simulations with measurement data show that the algorithm is capable of detecting faults without generating false alarms.

Ten-year prediction of osteoporosis from baseline bone mineral density: development of prognostic thresholds in healthy postmenopausal women. The Danish Osteoporosis Prevention Study

Osteopenia is common in healthy women examined in the first year or two following menopause. Short-term fracture risk is low, but we lack algorithms to assess long-term risk of osteoporosis. Because bone loss proceeds at only a few percent per year, we speculated that baseline bone mineral density (BMD) would predict a large proportion of 10-year BMD and be useful for deriving predictive thresholds. We aimed to identify prognostic thresholds associated with less than 10% risk of osteoporosis by 10 years in the individual participant, in order to allow rational osteodensitometry and intervention. We analyzed dual energy X-ray absorptometry (DXA) of the lumbar spine (LS) and femoral neck (FN) from 872 women, who participated in the non-HRT arms of the Danish Osteoporosis Prevention Study and had remained on no HRT, bisphosphonates or raloxifene since inclusion 10 years ago. We defined development of a T -score below -2.5 at the LS and/or FN or incident fracture as end-point, and we derived prognostic thresholds for baseline BMD, defining 90% NPV (negative predictive value) and 90% sensitivity, respectively. Seventy-six percent of the variation in BMD of the LS at 10 years was predicted by baseline BMD. In an individual participant, a baseline BMD T -score above -1.4 (FN or LS, whichever was lower) was associated with a 10-year risk of less than 10% of developing osteoporotic BMD or fracture. This covered 69% of the population. By contrast, participants with T -scores below -1.4 had a 56% risk of fracture or low BMD within 10 years. At the population level, baseline T -score cutoffs below 0 at the LS (68% of the population), 0 at the FN (72%) or -0.6 (62%) at the lower of the two sites capture 90% of the population that developed osteoporosis during the following 10 years. A BMD measurement, performed in the first two years following menopause, is a strong long-term predictor of BMD in healthy women. The association is strong enough to provide robust prognostic thresholds, which can be used to divide the population into two prognostic classes at menopause.

DYNAMIC THRESHOLD GENERATORS FOR FAULT DETECTION IN UNCERTAIN SYSTEMS

The problem of developing robust thresholds for fault detection is addressed. An inequality for the solution of a linear system with uncertain parameters is provided and is shown to be a valuable tool for developing dynamic threshold generators for fault detection. Such threshold generators are desirable for achieving robustness against model uncertainty in combination with sensitivity to small faults.The usefulness of the inequality is illustrated by developing an algorithm for detection of clogging in the valves of a flotation process. Simulations with measurement data show that the algorithm detects faults without generating false alarms.

Robust fault diagnosis for large-scale analog circuits with measurement noises

A hierarchical frequency-domain robust component failure detection (RCFD) scheme is proposed for the diagnosis of large-scale analog circuits with measurement noises based on the design of robust detection filters and robust residual thresholds. The robust residuals generated by the robust detection filters are employed to realize both fault isolation (FI) and component failure detection (CFD). A procedure for the design of robust detection filters is proposed and the associated rule for the selection of measurement variables is suggested. Furthermore, the effect of the measurement noises to the diagnostic result can be reduced by the proper design of residual thresholds, and the robustness of the diagnosis is enhanced. Finally, an example is illustrated to show the efficacy of the proposed method.