articles+ search results

Artificial intelligence has bounced into industrial applications contributing several advantages to the field and have led to the possibility to open new ways to solve many actual problems. In this paper, a data-driven performance evaluation methodology is presented and applied to an industrial refrigeration system. The strategy takes advantage of the Multivariate Kernel Density Estimation technique and Self-Organizing Maps to develop a robust method, which is able to determine a near-optimal performance map, taking into account the system uncertainties and the multiple signals involved in the process. A normality model is used to detect and filter non-representative operating samples to subsequently develop a reliable performance map. The performance map allows comparing the plant assessment under the same operating conditions and permits to identify the potential system improvement capabilities. To ensure that the resulting evaluation is trustworthy, a robustness strategy is developed to identify either possible new operation conditions or abnormal situations in order to avoid uncertain assessments. Furthermore, the proposed approach is tested with real industrial plant data to validate the suitability of the method.
Peer Reviewed
Postprint (published version)

Differential Interferometric Synthetic Aperture Radar (DInSAR) has proved its unprecedented advantages of monitoring ground deformation on large scale with centimeter to millimeter precision in the last two decades. However, the reliability and accuracy are often contaminated with atmospheric artefacts caused by spatial and temporal variations of the atmosphere. Recent studies revealed atmospheric artefacts can be compensated with empirical models, GPS zenith path delay and numerical weather prediction models. In this paper, an improved methodology is proposed based on atmospheric reanalysis data to estimate atmospheric artefacts. With our approach, the realistic line of sight (LOS) path along satellite location and monitored points is considered, rather than the zenith path delay. The effectiveness of our method is validated over Tenerife island, Spain by using Sentinel1 datasets.
Peer Reviewed
Postprint (published version)

© 2015 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Self-adaptive systems are capable of dealing with uncertainty at runtime handling complex issues as resource variability, changing user needs, and system intrusions or faults. If the requirements depend on context, runtime uncertainty will affect the execution of these contextual requirements. This work presents SACRE, a proof-of-concept implementation of an existing approach, ACon, developed by researchers of the Univ. of Victoria (Canada) in collaboration with the UPC (Spain). ACon uses a feedback loop to detect contextual requirements affected by uncertainty and data mining techniques to determine the best operationalization of contexts on top of sensed data. The implementation is placed in the domain of smart vehicles and the contextual requirements provide functionality for drowsy drivers.
Peer Reviewed
Postprint (author's final draft)

This work evaluates by simulation the performance of the Unfalsified Adaptive Control (UAC) for Multiple Degree of Freedom (MDoF) serial manipulators. The UAC is a data-driven technique that addresses stability issues of model-based controllers for robot arms with inertial uncertainties. The unfalsified controller selects the most suitable controller from a set, based on performance, to decide whether the controller in the closed loop should be changed, using only system inputs and outputs, i.e., torques and joint variables of the robotic arm, respectively. In this work, performance and robustness is evaluated by simulation on a 5-DoF manipulator showing the ability of the UAC to accomplish tracking tasks in the presence of inertial parameters disturbances.
Peer Reviewed
Postprint (author's final draft)

Memristors are considered one of the most favorable emerging device alternatives for future memory technologies. They are attracting great attention recently, due to their high scalability and compatibility with CMOS fabrication process. Alongside their benefits, they also face reliability concerns (e.g. manufacturing variability). In this sense our work analyzes key sources of uncertainties in the operation of the memristive memory and we present an analytic approach to predict the expected lifetime distribution of a memristive crossbar.
Postprint (published version)

In this paper, the problem of fault diagnosis of a wind farm is addressed using interval nonlinear parameter varying (NLPV) parity equations. Fault detection is based on the use of parity equations assuming unknown but bounded description of the noise and modeling errors. The fault detection test is based on checking the consistency between the measurements and the model by finding if the formers are inside the interval prediction bounds. The fault isolation algorithm is based on analyzing the observed fault signatures on-line, and matching them with the theoretical ones obtained using structural analysis. Finally, the proposed approach is tested using the wind farm benchmark proposed in the context of the wind farm FDI/FTC competition.
Peer Reviewed
Postprint (published version)

© 2014 IEEE
In this paper, the robust fault diagnosis pro- blem for non-linear systems considering both bounded parametric modelling errors and noises is addressed using parity equation based Analytical Redundancy Relations and Interval Constraint Satisfaction techniques. Fault detection, isolation and estimation tasks are considered. Moreover, the paper addresses the problem of determining the uncertainty in the parameters of the used uncertain ARRs. To illustrate the usefulness of the proposed ap- proach, a case study based on the well known wind turbine benchmark is used.
Peer Reviewed
Postprint (author's final draft)

This paper focuses on the impacts of considering batteries state-of-charge (SOC) at the departure time on thedemand modeling of plug-in electric vehicles (PEVs). Almost all of the previous researches assumed that PEVs batteries at the departure time are fully charged; however, this assumption is highly questionable because it is probable for a PEV to not be charged every day. The probability density function of a vehicle owners’ willingness to fulfill the daily charging is extracted according to the initial SOC of a PEV and the estimated distance of its next trip. Afterwards, with the aim of considering the uncertainties with the associated random variables as well as properly adjusting vehicles SOC at the departure time, a Monte Carlo based multi loop (MCML) algorithm is developed which is composed of two loops, namely the inner loop and the outer loop. In order to implement the proposed stochastic method, a case study has been conducted employing the gathered datasets related to the ICE vehicles in Tehran. Appropriate Student’s t copula functions have been fitted to the datasets in order to take into account the correlation structure among them as well as to generate the required random samples.

A hyper-heuristic is a high level methodology which performs search over the space of heuristics each operating on the space of solutions to solve hard computational problems. This search process is based on either generation or selection of low level heuristics. The latter approach is used in selection hyper-heuristics. A generic selection hyper-heuristic has two main components which operate successively: heuristic selection and move acceptance methods. An initially generated solution is improved iteratively using these methods. At a given step, the most appropriate heuristic is selected from a fixed set of low level heuristics and applied to a candidate solution producing a new one. Then, a decision is made whether to accept or reject the new solution. This process is repeated until the termination criterion is satisfied. There is strong empirical evidence that the choice of selection hyper-heuristic influences its overall performance. This is one of the first studies to the best of our knowledge that suggests and explores the use of group decision making methods for move acceptance in selection hyper-heuristics. The acceptance decision for a move is performed by multiple methods instead of a single one. The performance of four such group decision making move acceptance methods are analysed within different hyper-heuristics over a set of benchmark functions. The experimental results show that the group decision making strategies have potential to improve the overall performance of selection hyper-heuristics.
Copyright © 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
13th UK Workshop on Computational Intelligence (UKCI), Guildford, UK, 9-11 September 2013

Many safety related and critical systems warn of potentially dangerous events; for example, the short term conflict alert (STCA) system warns of airspace infractions between aircraft. Although installed with current technology, such critical systems may become out of date due to changes in the circumstances in which they function, operational procedures, and the regulatory environment. Current practice is to "tune," by hand, the many parameters governing the system in order to optimize the operating point in terms of the true positive and false positive rates, which are frequently associated with highly imbalanced costs. We cast the tuning of critical systems as a multiobjective optimization problem. We show how a region of the optimal receiver operating characteristic (ROC) curve may be obtained, permitting the system operators to select the operating point. We apply this methodology to the STCA system, using a multiobjective (1+1) evolution strategy, showing that we can improve upon the current hand-tuned operating point, as well as providing the salient ROC curve describing the true positive versus false positive tradeoff. We also provide results for three-objective optimization of the alert response time in addition to the true and false positive rates. Additionally, we illustrate the use of bootstrapping for representing evaluation uncertainty on estimated Pareto fronts, where the evaluation of a system is based upon a finite set of representative data.
Copyright © 2006 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
Notes: In this paper multi-objective optimisation is used for the first time to adjust the 1500 parameters of Short-Term Conflict Alert systems to optimise the Receiver Operating Characteristic (ROC) by simultaneously reducing the false positive rate and increasing the true positive alert rate, something that previous work by other researchers had not succeeded in doing. Importantly for such safety-critical systems, the method also yields an assessment of the confidence that may be placed in the optimised ROC curves. The paper results from a collaboration with NATS and a current KTP project, also with NATS, is deploying the methods in air-traffic control centres nationwide.

We describe a Bayesian framework for active learning for non-separable data, which incorporates a query density to explicitly model how new data is to be sampled. The model makes no assumption of independence between queried data-points; rather it updates model parameters on the basis of both observations and how those observations were sampled. A `hypothetical' look-ahead is employed to evaluate expected cost in the next time-step. We show the efficacy of this algorithm on the probabilistic high-low game which is a non-separable generalisation of the separable high-low game introduced by Seung et al. Our results indicate that the active Bayes algorithm performs significantly better than passive learning even when the overlap region is wide, covering over 30% of the feature space.
Copyright © 2010 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Dynamically substructured systems (DSS) play an important role in modern testing methods. DSS enables full-size critical components of a complete system to be tested physically in realtime, while the remaining parts of the system run in parallel as a real-time simulation. The performance of DSS testing is influenced by the synchronization of the physical and numerical substructures, which necessitates the design of a DSS controller. Since the testing signal is known and can be assumed to be a perfectly measured disturbance, the DSS control can be viewed as a regulation control problem with measured disturbance attenuation. A potential problem with DSS control arises from actuator saturation, which can be encountered in DSS transfer systems and can significantly influence the testing accuracy. This paper demonstrates the application of a novel robust disturbance rejection antiwindup (AW) technique, to cope with the actuator saturation problem in DSS. Implementation results from a hydraulically-actuated DSS test rig confirm the advantage of using this novel approach over some other existing AW approaches. Furthermore, some specific practical issues are discussed for the AW compensator design, such as the tuning of parameters.
Copyright © 2013 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

Many safety related and critical systems warn of potentially dangerous events; for example, the short term conflict alert (STCA) system warns of airspace infractions between aircraft. Although installed with current technology, such critical systems may become out of date due to changes in the circumstances in which they function, operational procedures, and the regulatory environment. Current practice is to "tune," by hand, the many parameters governing the system in order to optimize the operating point in terms of the true positive and false positive rates, which are frequently associated with highly imbalanced costs. We cast the tuning of critical systems as a multiobjective optimization problem. We show how a region of the optimal receiver operating characteristic (ROC) curve may be obtained, permitting the system operators to select the operating point. We apply this methodology to the STCA system, using a multiobjective (1+1) evolution strategy, showing that we can improve upon the current hand-tuned operating point, as well as providing the salient ROC curve describing the true positive versus false positive tradeoff. We also provide results for three-objective optimization of the alert response time in addition to the true and false positive rates. Additionally, we illustrate the use of bootstrapping for representing evaluation uncertainty on estimated Pareto fronts, where the evaluation of a system is based upon a finite set of representative data.
Copyright © 2006 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
Notes: In this paper multi-objective optimisation is used for the first time to adjust the 1500 parameters of Short-Term Conflict Alert systems to optimise the Receiver Operating Characteristic (ROC) by simultaneously reducing the false positive rate and increasing the true positive alert rate, something that previous work by other researchers had not succeeded in doing. Importantly for such safety-critical systems, the method also yields an assessment of the confidence that may be placed in the optimised ROC curves. The paper results from a collaboration with NATS and a current KTP project, also with NATS, is deploying the methods in air-traffic control centres nationwide.

When optimising receiver operating characteristic (ROC) curves there is an inherent degree of uncertainty associated with the operating point evaluation of a model parameterisation x. This is due to the finite amount of training data used to evaluate the true and false positive rates of x. The uncertainty associated with any particular x can be reduced, but only at the computation cost of evaluating more data. Here we explicitly represent this uncertainty through the use of probabilistically non-dominated archives, and show how expensive ROC optimisation problems may be tackled by only evaluating a small subset of the available data at each generation of an optimisation algorithm. Illustrative results are given on data sets from the well known UCI machine learning repository.
Copyright © 2008 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
IEEE Congress on Evolutionary Computation 2008 (CEC 2008). (IEEE World Congress on Computational Intelligence), Hong Kong, 1-6 June 2008

There has been only limited discussion on the effect of uncertainty and noise in multi-objective optimisation problems and how to deal with it. Here we address this problem by assessing the probability of dominance and maintaining an archive of solutions which are, with some known probability, mutually non-dominating.We examine methods for estimating the probability of dominance. These depend crucially on estimating the effective noise variance and we introduce a novel method of learning the variance during optimisation.Probabilistic domination contours are presented as a method for conveying the confidence that may be placed in objectives that are optimised in the presence of uncertainty.
Copyright © 2005 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works.
2005 IEEE Congress on Evolutionary Computation, Edinburgh, Scotland, 2-5 September 2005

We describe a Bayesian framework for active learning for non-separable data, which incorporates a query density to explicitly model how new data is to be sampled. The model makes no assumption of independence between queried data-points; rather it updates model parameters on the basis of both observations and how those observations were sampled. A `hypothetical' look-ahead is employed to evaluate expected cost in the next time-step. We show the efficacy of this algorithm on the probabilistic high-low game which is a non-separable generalisation of the separable high-low game introduced by Seung et al. Our results indicate that the active Bayes algorithm performs significantly better than passive learning even when the overlap region is wide, covering over 30% of the feature space.
Copyright © 2010 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

[EN] High frequency current transformers (HFCT) are widely used in partial discharge field and laboratory tests. Using these devices as sensors in partial discharge cable tests brings some advantages, namely: high degree of insulation, relatively good sensibility and the possibility of broadband measures. This paper analyzes the fundamentals of HFCTs measurements from the point of view of charge evaluation. Three different procedures for charge evaluation and the associated uncertainty are presented for partial discharge measurements in cables by means of HFCTs. Experimental and simulation data are presented and the associated uncertainty is calculated.

[EN] High frequency current transformers (HFCT) are widely used in partial discharge field and laboratory tests. Using these devices as sensors in partial discharge cable tests brings some advantages, namely: high degree of insulation, relatively good sensibility and the possibility of broadband measures. This paper analyzes the fundamentals of HFCTs measurements from the point of view of charge evaluation. Three different procedures for charge evaluation and the associated uncertainty are presented for partial discharge measurements in cables by means of HFCTs. Experimental and simulation data are presented and the associated uncertainty is calculated.

High frequency current transformers (HFCT) are widely used in partial discharge field and laboratory tests. Using these devices as sensors in partial discharge cable tests brings some advantages, namely: high degree of insulation, relatively good sensibility and the possibility of broadband measures. This paper analyzes the fundamentals of HFCTs measurements from the point of view of charge evaluation. Three different procedures for charge evaluation and the associated uncertainty are presented for partial discharge measurements in cables by means of HFCTs. Experimental and simulation data are presented and the associated uncertainty is calculated.

[EN] High frequency current transformers (HFCT) are now widely used for partial discharges detection and measurement, specially for partial discharge location and source identification. However, many difficulties arise due to HFCT in the determination of the apparent charge in picocoulombs (pC) because of large variations according to sensors or pulses shapes. Some authors are even using milivolts (mV) as representative magnitude. As a result, measurements carried out using HFCT are not comparable in terms of apparent charge determination. In this paper, a new method of charge evaluation by means of HFCT is described solving the calibration procedure according to the sensor bandwidth and providing a criterion for sensors evaluation.