Demetrios G. Eliades

A Fault Diagnosis and Security Framework for Water Systems

Water resources management is a key challenge that will become even more crucial in the years ahead. From a system-theoretic viewpoint, there is a need to develop rigorous design and analysis tools for control, fault diagnosis and security of water distribution networks. This work develops a mathematical framework suitable for fault diagnosis and security in water systems; in addition it investigates the problem of determining a suitable set of locations for sensor placement in large-scale drinking water distribution networks such that contaminant detection is optimized. This work contributes to the research by presenting a problem formulation were the state-space representation of the propagation and reaction dynamics is coupled with the impact dynamics describing the “damage” caused by a contamination of the water distribution network. We propose a solution methodology for the sensor-placement problem by considering several risk-objectives, and by utilizing various optimization and evolutionary computation techniques. To illustrate the methodology, we present results of a simplified and a real water distribution network.

ITERATIVE DEEPENING OF PARETO SOLUTIONS IN WATER SENSOR NETWORKS

This paper investigates the issue of finding the best sensor locations in a drinking water distribution network for detecting harmful substances. The problem is formulated in a multi-objective optimization framework with five performance measures: time of detection, population affected prior to detection, demand of contaminated water, detection likelihood and demand coverage. In practical application, due to the large size of water distribution networks, the space of possible solutions expands dramatically, making it difficult or impossible to determine the optimal solutions. We propose the “Iterative Deepening of Pareto Solutions” search algorithm, for locating “good enough” solutions. The algorithm solves the problem by iteratively choosing the best non-dominant solutions, and expanding them by increasing the depth of the search tree until all the sensors have been used. Simulation experiments were performed on two water distribution networks, following the formulation defined in the “Battle of the Water Sensor Networks” design challenge (Ostfeld et al. 2005). Four contamination scenarios are considered and from the sets of possible solutions, the most appropriate designs are proposed.

Indoor Localization Using Neural Networks with Location Fingerprints

Reliable localization techniques applicable to indoor environments are essential for the development of advanced location aware applications. We rely on WLAN infrastructure and exploit location related information, such as the Received Signal Strength (RSS) measurements, to estimate the unknown terminal location. We adopt Artificial Neural Networks (ANN) as a function approximation approach to map vectors of RSS samples, known as location fingerprints, to coordinates on the plane. We present an efficient algorithm based on Radial Basis Function (RBF) networks and describe a data clustering method to reduce the network size. The proposed algorithm is practical and scalable, while the experimental results indicate that it outperforms existing techniques in terms of the positioning error.

Measurement Uncertainty Considerations in Optimal Sensor Deployment for State Estimation

This paper proposes a method for optimal placement of phasor measurement units (PMUs) for measuring the states of a power system. A method to compute the measurement uncertainty associated with the estimated states is also illustrated in the paper. The PMU placement strategy ensures complete observability of the power system states for normal operating conditions, as well as under the loss of a single transmission line or even a single measurement unit. An integer quadratic programming approach is used to minimize the total number of PMUs required to make the system completely observable, and to maximize the measurement redundancy at the power system busses. The goal of the research is to take into account the measurement uncertainty while determining the optimal number and locations of the PMUs for state estimation. Simulation results on the IEEE 14-bus test system are presented in this paper.

Fault detection for service mobile robots using model-based method

Detection of faults is a topic of high importance because it increases robot dependability, a requirement for the wide acceptance of service robots in domestic environments. This work takes a model-based approach for detecting and identifying actuator faults on differential-drive mobile robots in an indoor environment. An error-bound is calculated between the estimated and measured robot states which is constantly adapted based on the current state and input signals. A fault is detected when the estimation error is outside this bound. The model parameters are learned by the robot using an adaptive law, after the robot deployment in the target environment. Model uncertainties have an important impact on the fault detection performance, and are dealt with by considering the uncertainty bounds in the bound calculations. This ensures no false alarms occur when the uncertainty remains bounded during normal operation. Furthermore an extension to the method is proposed that addresses the problem of detecting small faults. The method is experimentally validated on a iRobot Roomba autonomous robot.

Water Contamination Impact Evaluation and Source-Area Isolation Using Decision Trees

AbstractThe security of drinking water distribution operation is an important issue that has received increasing interest within the last few years. The U.S. EPA has issued guidelines for water utilities regarding which qualitative and quantitative metrics to monitor, and what response actions to take from the moment a contamination event alarm has been triggered, until the contamination has been accommodated and the system has returned to normal operation. Expanded sampling is a type of response action in which the water utilities examine water quality at certain locations in the network after a contamination event has been detected to help evaluate the contamination impact and locate the source-area. In this work, we propose a computational approach, based on decision trees, for choosing a sequence of nodes in the distribution network to perform expanded sampling, such that the water contamination impact is evaluated and the source-area is isolated, with as few manual quality samplings as possible. To i...

A Matlab-CONTAM Toolbox for Contaminant Event Monitoring in Intelligent Buildings

An intelligent building should take all the necessary steps to provide protection against the dispersion of contaminants from sources (events) inside the building which can compromise the indoor air quality and influence the occupants’ comfort, health, productivity and safety. Multi-zone models and software, such as CONTAM, have been widely used in building environmental studies for predicting airflows and the resulting contaminant transport. This paper describes a developed Matlab Toolbox that allows the creation of data sets from running multiple scenarios using CONTAM by varying the different problem parameters. The Matlab-CONTAM Toolbox is an expandable research tool which facilitates the implementation of various algorithms related to contamination event monitoring. In particular, this paper describes the implementation of state-of-the-art algorithms for detecting and isolating a contaminant source. The use of the Toolbox is demonstrated through a building case-study. The Matlab-CONTAM Toolbox is released under an open-source licence, and is available at https://github.com/KIOS-Research/matlab-contam-toolbox .

Contaminant Detection in Urban Water Distribution Networks Using Chlorine Measurements

In this work we present a contamination detection methodology for water distribution networks. The proposed detection method is based on chlorine sensor measurements, which are compare to certain computed upper and lower periodic bounds. The bounds are computed using randomized simulations aimed at capturing the variations in chlorine concentration due to the significant uncertainty in the water demand patterns, average nodal consumptions, roughness parameters and reaction coefficients. The proposed method is applied to a set of high-impact contamination fault scenarios using a benchmark distribution network, for which on-line chlorine concentration sensors are assumed to have been installed at certain locations following an optimization procedure. The results indicate that by using the periodic bounds computed from the randomized simulations, for the proposed benchmark, contamination events are detected within reasonable time.

Contamination event detection in drinking water systems using a real-time learning approach

In this work we present the problem of contamination event detection in drinking water distribution networks using real-time learning approaches and a model-based event detection scheme. By using chlorine concentration measurements a contamination event detection algorithm is developed with the aim of detecting the occurrence of contaminant injection into a water tank by monitoring the change of the chlorine concentration. The proposed methodology is comprised of two steps: a) learn in real-time the unknown chlorine reaction dynamics using a Radial-Basis Function network; b) activate a contamination-event detection methodology which uses an adaptive detection threshold. A contamination event detection alarm is activated when the magnitude of the estimation error exceeds the detection threshold. To demonstrate the proposed methodology realistic case study is evaluated with Monte-Carlo simulations of contamination events of different magnitudes occurrence times and environmental characteristics. The results demonstrate the improvement in the contamination event detection performance when using the real-time learning approach.

Security of Water Infrastructure Systems

This paper formulates the security problem in critical water infrastructure systems for diagnosing quality faults. The proposed scheme is based on the discretized equations of advection and reaction of contaminant concentrations in pipes and tanks, expressed in a state-space form. Faults are signals affecting the states, and their impact is measured based on certain epidemiological dynamics. A multi-objective optimization problem is formulated for minimizing various risk-related objectives.