Water Distribution Network Monitoring Research Articles

Interpretable SAM-kNN Regressor for Incremental Learning on High-Dimensional Data Streams

ABSTRACT In many real-world scenarios, data are provided as a potentially infinite stream of samples that are subject to changes in the underlying data distribution, a phenomenon often referred to as concept drift. A specific facet of concept drift is feature drift, where the relevance of a feature to the problem at hand changes over time. High-dimensionality of the data poses an additional challenge to learning algorithms operating in such environments. Common scenarios of this nature can for example be found in sensor-based maintenance operations of industrial machines or inside entire networks, such as power grids or water distribution systems. However, since most existing methods for incremental learning focus on classification tasks, efficient online learning for regression is still an underdeveloped area. In this work, we introduce an extension to the SAM-kNN Regressor that incorporates metric learning in order to improve the prediction quality on data streams, gain insights into the relevance of different input features and based on that, transform the input data into a lower dimension in order to improve computational complexity and suitability for high-dimensional data. We evaluate our proposed method on artificial data, to demonstrate its applicability in various scenarios. In addition to that, we apply the method to the real-world problem of water distribution network monitoring. Specifically, we demonstrate that sensor faults in the water distribution network can be detected by monitoring the feature relevances computed by our algorithm.

Complexities of the urban drinking water systems in Ethiopia and possible interventions for sustainability.

The online version contains supplementary material available at 10.1007/s10668-022-02901-7.

Using mobile and fixed sensors for optimal monitoring of water distribution network under dynamic water quality simulations

Urban water distribution networks (WDNs) are strategic infrastructures, which have critical roles in preserving public health and social growth. This study designs a water quality monitoring network (WQMN) in the WDN Example 3 of EPANET2.0, contaminated with potassium cyanide (KCN), accounting for the dynamic changes of chlorine as a common water-quality parameter. The uncertainties in the contaminant events, such as the amount, location, start-time, and duration of contaminant injection, are considered in the problem formulations. A simulation-optimization framework based on EPANET-MSX and Particle Swarm Optimization algorithm is developed to determine the optimal location of the fixed sensor(s) and/or time and location of releasing mobile sensor(s). The results indicate that increasing the available mobile sensors from 1 unit to 3 or 5 units, decreased the contaminant detection time (DT) by 14% and 22%, increased the contaminant detection likelihood by 102% and 159%, and reduced the contaminated water consumption by 52% and 70%, respectively. The sensitivity analysis on the number of available sensors implies that changes in fixed sensors are more effective than mobile sensors on contaminant DT. The results showed the uncertainty in sink nodes' coverage area and mobile sensors’ battery life become more important as the number of mobile sensors increases.

Efficient k-means clustering and greedy selection-based reduction of nodal search space for optimization of sensor placement in the water distribution networks

Monitoring of water distribution network (WDN) requires placement of sensors at strategic locations to detect maximum contamination events at the earliest. The multi-objective optimization (MOO) of sensor placement is a complicated problem owing to its combinatorial nature, interconnected and large WDN sizes, and temporal flows producing complex outcomes for a given set of contamination events. In this study, a new method is proposed to reduce the complexity of the problem by condensing the nodal search space. This method first segregates the nodes based on intrusion events detected, using k-means clustering, followed by selecting nodes from each group based on the improvement observed in the objectives, namely, contamination event detection, expected detection time, and affected population. The selected nodes formed the decision variable space for the MOO study. The developed strategy was tested on two benchmark networks: BWSN Network1 and C-town network, and its performance is compared with the traditional method in terms of hypervolume contribution rate (CR) indicator and the number of Pareto points. The optimal subset of nodes generated twice the number of Pareto points than the complete set of nodes set for placing 20 sensors and had 10% more than CR indicator than the traditional method. For the placement of 5 sensors, the proposed solutions were better at the higher detection likelihood values, which is required to achieve maximum detection. The proposed sensor placement algorithm can be easily scaled to large WDNs. It is expected to provide a better optimal sensor placement solution irrespective of network size as compared to the traditional approach.

Optimal Number of Pressure Sensors for Real-Time Monitoring of Distribution Networks by Using the Hypervolume Indicator

This article proposes a novel methodology to determine the optimal number of pressure sensors for the real-time monitoring of water distribution networks based on a quality hypervolume indicator. The proposed methodology solves the optimization problem for different numbers of pressure sensors, assesses the gain of installing each set of sensors by means of the hypervolume indicator and determines the optimal number of sensors by the variation of the hypervolume indicator. The methodology was applied to a real case study. Several robustness analyses were carried out. The results demonstrate that the methodology is hardly influenced by the method parameters and that a reasonable estimation of the optimal number of sensors can be easily achieved.

Multiscale Gaussian process regression-based generalized likelihood ratio test for fault detection in water distribution networks

This paper proposes a new leak/contaminant detection approach that aims to enhance the monitoring of water distribution network (WDN). The developed method relies on using machine learning (e.g Gaussian process regression (GPR)) as a modeling framework and generalized likelihood ratio (GLRT) for detection purposes. To improve the performances of the developed GPR model even further, multiscale representation of data will be used to develop multiscale extension of these method. Multiscale representation is a powerful data analysis technique that presents efficient separation of deterministic characteristics from random noise. Therefore, the multiscale GPR method, that combines the advantages of the GPR method with those of multiscale representation, will be developed to enhance the WDN modeling performance. We develop a new technique for detecting leak/contaminant in WDN using GLRT. For further enhance, the performance of GLRT, an exponentially weighted moving average (EWMA)-GLRT (EWMA-GLRT) chart is developed. The simulation results show that the MSGPR-based EWMA-GLRT method outperforms MSGPR-based GLRT and that both of them provide clear advantages over the neural networks (NN)- and support vector regression (SVR)- and GPR-based GLRT techniques.

Energy Efficient Sensor Activation for Water Distribution Networks Based on Compressive Sensing

The recent development of low cost wireless sensors enables novel internet-of-things (IoT) applications, such as the monitoring of water distribution networks. In such scenarios, the lifetime of the wireless sensor network (WSN) is a major concern, given that sensor node replacement is generally inconvenient and costly. In this paper, a compressive sensing-based scheduling scheme is proposed that conserves energy by activating only a small subset of sensor nodes in each timeslot to sense and transmit. Compressive sensing introduces a cardinality constraint that makes the scheduling optimization problem particularly challenging. Taking advantage of the network topology imposed by the IoT water monitoring scenario, the scheduling problem is decomposed into simpler subproblems, and a dynamic-programming-based solution method is proposed. Based on the proposed method, a solution algorithm is derived, whose complexity and energy-wise performance are investigated. The complexity of the proposed algorithm is characterized and its performance is evaluated numerically via an IoT emulator of water distribution networks. The analytical and numerical results show that the proposed algorithm outperforms state-of-the-art approaches in terms of energy consumption, network lifetime, and robustness to sensor node failures. It is argued that the derived solution approach is general and it can be potentially applied to more IoT scenarios such as WSN scheduling in smart cities and intelligent transport systems.

SmartPipes: Smart Wireless Sensor Networks for Leak Detection in Water Pipelines

Asset monitoring, specifically infrastructure monitoring such as water distribution pipelines, is becoming increasingly critical for utility owners who face new challenges due to an aging network. In the UK alone, during the period of 2009–2010, approximately 3281 mega litres (106) of water were wasted due to failure or leaks in water pipelines. Various techniques can be used for the monitoring of water distribution networks. This paper presents the design, development and testing of a smart wireless sensor network for leak detection in water pipelines, based on the measurement of relative indirect pressure changes in plastic pipes. Power consumption of the sensor nodes is minimised to 2.2 mW based on one measurement every 6 h in order to prolong the lifetime of the network and increase the sensor nodes’ compatibility with current levels of power available by energy harvesting methods and long life batteries. A novel pressure sensing method is investigated for its performance and capabilities by both laboratory and field trials. The sensors were capable of measuring pressure changes due to leaks. These pressure profiles can also be used to locate the leaks.

Sequential Detection of Transient Changes

This article addresses the transient change detection problem. It is assumed that a change occurs at an unknown (but nonrandom) change-point and the duration of post-change period is finite and known. A latent detection—that is, a detection that occurs after signal disappearance—is considered as a missed detection. A new optimality criterion adapted to the detection of transient changes involves the minimization of the worst-case probability of missed detection under constraint on the false alarm rate for a given period. A suboptimal sequential transient change detection algorithm is proposed. It is based on a window-limited cumulative sum (CUSUM) test. An upper bound for the worst-case probability of missed detection and a lower and an upper bound for the false alarm rate are proposed. Based on these bounds, the window-limited CUSUM test is optimized with respect to the proposed criterion. The developed algorithm and theoretical findings are applied to drinking water distribution network monitoring.

Sequential Monitoring of Water Distribution Network

Monitoring drinking water is a public health problem because water is essential for human life. A typical modern drinking water distribution system includes a water purification plant, water towers, water tanks, pumps and pipes. Many procedures are developed for monitoring water quality in water treatment plants. Monitoring of water distribution systems has received less attention.The goal of this paper is to study the problem of drinking water safety by ensuring the monitoring of the distribution network from water tower to private residences. The proposed approach is based on the observation of residual chlorine concentrations which are provided by sensors network. By using the hydraulic model, it is possible to compute the nominal residuals chlorine concentrations under assumption that quality of the water at the water tower output is known. If a significant reduction of the residual chlorine concentrations is observed then it can be interpreted as presence of a dangerous water contamination. This event should be detected as reliable as possible provided an acceptable detection delay and false alarm rate. The traditional abrupt change detection theory is well developed in the case when the duration of the post-change mode is infinite. Unfortunately, in the discussed case the transient signal duration is short. Hence, the usage of traditional change detection algorithm is compromised.A criterion based on the minimization of the missed detection probability provided that the false alarm rate is upper bounded is used in the paper. A suboptimal detection algorithm is designed. The theoretical analysis and the results of simulation are provided.

Cost-effective leakage reduction through district metering

A new stochastic water distribution network (WDN) modelling tool has been developed with the aim of providing a sound basis for cost-effective strategic planning of WDN monitoring schemes in terms of quantifiable leakage. Through an analysis of uncertainties in water demand and its propagation through a WDN, the developed computer tool recommends the most cost-effective solution in terms of the number and location of monitoring sites necessary to increase the probability of identifying the location of leaks (or other system anomalies). The methodology was tested on a medium-sized WDN in the UK and the results are presented in this paper.

An integrated approach to water distribution network monitoring and performance prediction using geographical information systems

The deterioration of water distribution systems and the subsequent need for major rehabilitation has provided researchers with a challenge: to seek new techniques to facilitate decision-making and assist network planners. Consequently, a range of methods have been developed, concerning several key performance indicators. Those with the greatest potential tend to promote a pro-active approach to pipeline rehabilitation, by predicting future performance of individual or groups of pipes. Such an approach is essential if cost-effective solutions are to be found. Prediction models are based on indicators that are perceived to be the most appropriate for measuring and comparing an aspect of performance. The level of “appropriateness” is influenced not only by the availability of data and the ease at which it can be recorded, but also its accuracy and unambiguity. Consequently, indicators such as burst rates are widely used to assess structural performance. Ideally though, a more direct measure of the condition of a pipe could be used, such as its remaining wall thickness. This has previously been largely disregarded, due to the lack of appropriate (non-destructive) measurement technology. However, technological advances are being made to enable the inline monitoring of pipes and determination of deficiencies (particularly corrosion pits in cast iron pipes). This paper illustrates how online condition monitoring of distribution pipelines can be combined with existing prediction models, to facilitate the decision-making process. Integration is achieved through the use of a Geographical Information System (GIS), which greatly enhances representation of spatial and temporal information.

Constrained state estimation with applications in water distribution network monitoring

This paper investigates the effects of imposing bounds on the measurements used in weighted least-squares (WLS) state estimation. The active limits for such bounds are derived and algorithms based on linear and quadratic programming kernels are presented. Using the lower limit for the bounds, the constrained WLS scheme becomes an adaptive maximally constrained scheme: M-WLS. For some networks, the poor prior knowledge of the global error characteristic results in some measurements having less influence than would be expected from the local error characteristics of their transducers. By using M-WLS estimation, the influence of such measurements on state estimation may be improved. Analysis of the adaptive bounding of the scheme can also lead to identification of critical measurement discrepancies. For the purpose of illustration, results are presented using simulated measurements; the head measurements (pressures) are consistent with nominal demands (nodal flows) and the demand measurements are generated by superimposing random errors of 2.5ls-1 rms on the nominal demands.

Constrained state estimation with applications in water distribution network monitoring

This paper investigates the effects of imposing bounds on the measurements used in weighted least-squares (WLS) state estimation. The active limits for such bounds are derived and algorithms based on linear and quadratic programming kernels are presented. Using the lower limit for the bounds, the constrained WLS scheme becomes an adaptive maximally constrained scheme: M-WLS. For some networks, the poor prior knowledge of the global error characteristic results in some measurements having less influence than would be expected from the local error characteristics of their transducers. By using M-WLS estimation, the influence of such measurements on state estimation may be improved. Analysis of the adaptive bounding of the scheme can also lead to identification of critical measurement discrepancies. For the purpose of illustration, results are presented using simulated measurements; the head measurements (pressures) are consistent with nominal demands (nodal flows) and the demand measurements are generated by superimposing random errors of 2.5ls-1 rms on the nominal demands.