We developed a hybrid algorithm for multi-objective design of water distribution networks (WDNs) in the present study. The proposed algorithm combines the global search schemes of differential evolution (DE) with the local search capabilities of harmony search (HS) to enhance the search proficiency of evolutionary algorithms. This method was compared with other multi-objective evolutionary algorithms (MOEAs) including NSGA2, SPEA2, MOEA/D and extended versions of DE and HS combined with non-dominance criteria using several metrics. We tested the compared algorithms on four benchmark WDN design problems with two objective functions, (i) the minimization of cost and (ii) the maximization of resiliency as reliability measure. The results showed that the proposed hybrid method provided better optimal solutions and outperformed the other algorithms. It also exhibited significant improvement over previous MOEAs. The hybrid algorithm generated new optimal solutions for a case study that dominated the best-known Pareto-optimal solutions in the literature

This study proposes variable balancing approaches for the exploration (diversification) and exploitation (intensification) of the non-dominated sorting genetic algorithm-II (NSGA-II) with simulated binary crossover (SBX) and polynomial mutation (PM) in the multiobjective automatic parameter calibration of a lumped hydrological model, the HYMOD model. Two objectives—minimizing the percent bias and minimizing three peak flow differences—are considered in the calibration of the six parameters of the model. The proposed balancing approaches, which migrate the focus between exploration and exploitation over generations by varying the crossover and mutation distribution indices of SBX and PM, respectively, are compared with traditional static balancing approaches (the two dices value is fixed during optimization) in a benchmark hydrological calibration problem for the Leaf River (1950 km²) near Collins, Mississippi. Three performance metrics—solution quality, spacing, and convergence—are used to quantify and compare the quality of the Pareto solutions obtained by the two different balancing approaches. The variable balancing approaches that migrate the focus of exploration and exploitation differently for SBX and PM outperformed other methods.

This study proposes a framework for improving the robustness of water distribution systems using the optimal valve installation approach with regard to system reinforcement. To improve the robustness in valve design, this study introduces critical segment selection technique and optimal valve location determination within the concept of segment. Using the segment finding algorithm, the segment and the unintentional isolation can be identified. To select the critical segment, a multicriteria decision technique is applied by considering the hydraulic, social and economic effect. Finally, the optimal valve locations and the number of additional valves is determined by pipe failure analysis through the trade-off relationship with the number of additional valves and the maximum damage under pipe failure situations. To verify the proposed technique, the real-world water distribution systems are applied and compared with the original design.

This study proposes a novel detection model for the detection of cyber-attacks using remote sensing data on water distribution systems (i.e., pipe flow sensor, nodal pressure sensor, tank water level sensor, and programmable logic controllers) by machine learning approaches. The most commonly used and well-known machine learning algorithms (i.e., k-nearest neighbor, support vector machine, artificial neural network, and extreme learning machine) were compared to determine the one with the best detection performance. After identifying the best algorithm, several improved versions of the algorithm are compared and analyzed according to their characteristics. Their quantitative performances and abilities to correctly classify the state of the urban water system under cyber-attack were measured using various performance indices. Among the algorithms tested, the extreme learning machine (ELM) was found to exhibit the best performance. Moreover, this study not only has identified excellent algorithm among the compared algorithms but also has considered an improved version of the outstanding algorithm. Furthermore, the comparison was performed using various representative performance indices to quantitatively measure the prediction accuracy and select the most appropriate model. Therefore, this study provides a new perspective on the characteristics of various versions of machine learning algorithms and their application to different problems, and this study may be referenced as a case study for future cyber-attack detection fields.

Real-time state estimation is defined as the process of calculating the state variable of interest in real time not being directly measured. In a water distribution system (WDS), nodal demands are often considered as the state variable (i.e., unknown variable) and can be estimated using nodal pressures and pipe flow rates measured at sensors installed throughout the system. Nodes are often grouped for aggregation to decrease the number of unknowns (demands) in the WDS demand estimation problem. This study proposes an optimal node grouping model to maximize the real-time WDS demand estimation accuracy. This Kalman filter-based demand estimation method is linked with a genetic algorithm for node group optimization. The modified Austin network demand is estimated to demonstrate the proposed model. True demands and field measurements are synthetically generated using a hydraulic model of the study network. Accordingly, the optimal node groups identified by the proposed model reduce the total root-mean-square error of the estimated node group demand by 24% compared to that determined by engineering knowledge. Based on the results, more pipe flow sensors should be installed to measure small flows and to further enhance the demand estimation accuracy.