This article proposes a new general approach in short-term water demand forecasting based on a two-stage learning process that couples time-series clustering with gene expression programming (GEP). The approach was tested on the real life water demand data of the city of Milan, in Italy. Moreover, multi-scale modeling using a series of head-time was deployed to investigate the optimum temporal resolution under study. Multi-scale modeling was performed based on rearranging hourly based patterns of water demand into 3, 6, 12, and 24 h lead times. Results showed that GEP should receive more attention among the emerging nonlinear modelling techniques if coupled with unsupervised learning algorithms in detailed spherical k-means.

Water distribution systems (WDSs) face numerous challenges in providing service to their customers. An especially difficult challenge occurs when several pipes in the system fail simultaneously. Simultaneous failure is more likely to occur in older networks or systems located in harsh climatic conditions. Previous studies have focused on WDS reliability when pipes fail individually. The current research proposes a technique to determine the reliability of a WDS experiencing different degrees of simultaneous pipe failure and to assess errors in reliability that occur when an inappropriate level of simultaneous failure is assumed. The model was applied to two case studies including a hypothetical small system and an actual WDS. Comparing the various states of reliability provided valuable information about system sensitivity to simultaneous multipipe failures. Results demonstrated that a system may be able to achieve a higher level of reliability if more realistic expectations of simultaneous failure are assumed.

Intelligent Soft Computing Models in Water Demand Forecasting | InTechOpen, Published on: 2016-08-24. Authors: Sina Shabani, Peyman Yousefi, Jan Adamowski, et

Studying uncertainties in hydrological modeling is necessary because of data scarcity or abundance and quality issues. These uncertainties can have significant effects on environmental decision making. Traditionally, probabilistic methods have been used to study uncertainties; however, recently, more comprehensive methods are used in the treatment of uncertainty. These methods are capable of addressing uncertainty in the form of vagueness, ambiguity, and conflict, which cannot be studied efficiently using probabilistic frameworks. The Dempster-Shafer theory of evidence (DST) is one of the popular methods that can provide a unified platform to address data conflict and incompleteness. In this paper, the use of DST to model and propagate the uncertainty arising from two snow water equivalent data sets with a high degree of conflict (DST conflict $k=0.74$) is demonstrated. In DST, on the basis of the nature of data, e.g., the degree of conflict, different combination rules are applicable. Here, four DST combination rules are applied including Dempster-Shafer, Yager, mixture, and the proportional conflict redistribution rule number 6 (PCR6). The outcomes from these rules are compared, and their effects on subsequent decision-making are discussed. Considering the specific condition of the data used, i.e., high-conflict data with limited quality information, results indicate that mixture and PCR6 rules are more appropriate. The resultant uncertainty-driven data set is subsequently used as input into an illustrative hydrologic model demonstrating a method for propagating uncertainty. In addition, the issues of resolving conflict for less contradicting data sets, the dependency between bodies of evidence, and modeling incompleteness are also discussed.

Considering all possible input (predictor) variables in a predictive water quality model is impractical owing to computational workload and complexity of the problem. Computational efficiency, as well as complexity of a model, is greatly increased if the most influential variables are determined using an input variable selection technique. In this study, the multilayer perceptron artificial neural network was implemented in order to predict total dissolved solids in the Sufi Chai river (Iran). The research studied the impacts of chemical composition (salinity, potassium, sodium, magnesium, calcium, sulfate, chloride, bicarbonate, carbonate, pH, and sodium adsorption ratio) in source water, climatic variables (rainfall, air temperature, wind speed, and evaporation), and hydrometric variables (river discharge and suspended sediment) on the predictions. Garson’s equation was used to find the relative importance of each input variable and to select the most influential variables. A correlation method was applied and the results were compared with those of the Garson method. A set of 12-year data (1999–2010) was used to calibrate, validate, and test the models. The results indicated that input variable selection before modeling can improve both accuracy and simplicity of the models. Although Garson and correlation methods both improved the accuracy of the models, the Garson method was found to be more accurate. As well, the research showed including climatic and hydrologic variables improved the accuracy of the models with fewer variables considered.