Hourly Water Demand Research Articles

Improving urban water demand forecast using conformal prediction-based hybrid machine learning models

This paper presents a probabilistic forecasting method that predicts the future water consumption patterns, taking into account the inherent uncertainty in the system. The proposed method leverages statistical techniques to estimate the probability distribution of future water demand scenarios. The findings of this study will provide decision-makers with a range of possible alternatives that facilitate better planning and management of water resources. We developed a conformal prediction-based hybrid demand forecast model, which combines convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM) while comparing it with other machine learning models for probabilistic hourly water demand forecasting. Additionally, we address crucial considerations when implementing a probabilistic forecasting system, including selecting appropriate data and choosing model parameters. The performance of the proposed model is validated for probabilistic water demand forecasting in real-world settings. Results indicate noteworthy improvement in deterministic and probabilistic predictions by 10 % and 26.7 %, respectively. The findings underscore the potential benefits of this approach for improved decision-making and resource management. The results also demonstrate that the proposed hybrid model outperforms the compared machine learning models with better prediction performance and increased robustness for handling prediction uncertainties.

Two years of COVID restrictions: A lesson from water demand data

Abstract The long period of great fragility experienced by the lockdown has strained entire countries. In response to COVID-19, Italy performed stay-at-home orders to attenuate disease spread, provoking drastic changes in all aspects of users' behaviour, particularly affecting water demand. A dataset of hourly water demand for 2019, 2020, and 2021, related to five Italian towns permitted to observe water consumption changes. Trends highlight a general decrease in water consumption, linked to the strict restrictions imposed and a morning peak shift. At the end of the strict quarantine regulations, water consumption did not return to pre-pandemic values because COVID-19 has led to a change in lifestyle.

Unsupervised anomaly detection in hourly water demand data using an asymmetric encoder–decoder model

Water demand forecasting (WDF) is essential for the design and optimal management of water distribution systems (WDS). Historical water demand data contribute significantly to WDF. Yet the obtained water demand data contain anomalies on occasions due to failures in WDS or monitoring systems. The contaminated water demand data are invalid to describe the actual water demand, thus degrading the performance of WDF models. However, the importance of anomaly detection in water demand data is underestimated, or at least not explicitly described in many published papers. To fill the gap, we propose an unsupervised anomaly detection method based on an asymmetric encoder–decoder (asyED) model. Different from the symmetric structure of a traditional autoencoder where a signal is reproduced from itself, asyED is asymmetric where a signal is reproduced from the upstream and downstream information of the signal, while the signal itself does not participate in the reconstruction. In light of this feature, asyED is powerful in identifying anomalies. The proposed method is employed to detect anomalies in hourly water demand data which exhibit trend and seasonality. The results show the superiority of the proposed method over the other four commonly used anomaly detection methods: Z-score, isolation forest, local outlier factor, and seasonal hybrid ESD (Extreme Studentized Deviate test). • An anomaly detection method based on an asymmetric encoder-decoder model is proposed. • The method applies to hourly water demand data which exhibit trend and seasonality. • A cluster-based metric is proposed for model evaluation. • The metric considers the continuity mechanism of anomalies found in the data. • Measures to quantify the strengths of different components in the data are proposed.

Optimal Control of a Reverse Osmosis Plant for Brackish Water Desalination Driven by Intermittent Wind Power

This work addresses the design of an online control system that continuously regulates a reverse osmosis (RO) desalination plant driven by wind power aided by water storage tanks. The control objective is to produce the exact hourly water demand in the presence of wind power intermittency, disturbances, and operational limitations. The manipulated variables are the RO feed pressure and the active number of RO vessels. The control system helped to decrease the annual water deficit by 20% under nominal conditions and when the plant is under the influence of disturbances. Moreover, the control system managed to decrease the annual water deficit by 73% when the plant operated under a shortage of an active number of wind turbines and RO vessels. The loss of redistributed production ratio (LPb) and the loss of raw production ratio (LP) were used as the controlled variables representing the proposed control objective. LPb was superior to LP by creating conservative control actions that produce the required water demand without violating the required water purity.

Hourly water demand forecasting using a hybrid model based on mind evolutionary algorithm

Abstract A hybrid model based on the mind evolutionary algorithm is proposed to predict hourly water demand. In the hybrid model, hourly water demand data are first reconstructed to generate appropriate samples so as to represent the characteristics of time series effectively. Then, the mind evolutionary algorithm is integrated into a back propagation neural network (BPNN) to improve prediction performance. To investigate the application potential of the proposed model in hourly water demand forecasting, real hourly water demand data were applied to evaluate its prediction performance. In addition, the performance of the proposed model was compared with a traditional BPNN model and another hybrid model where the genetic algorithm (GA) is used as an optimization algorithm for BPNN. The results show that the proposed model has a satisfactory prediction performance in hourly water demand forecasting. On the whole, the proposed model outperforms all other models involved in the comparisons in both prediction accuracy and stability. These findings suggest that the proposed model can be a novel and effective tool for hourly water demand forecasting.

An Innovative Hourly Water Demand Forecasting Preprocessing Framework with Local Outlier Correction and Adaptive Decomposition Techniques

Accurate forecasting of hourly water demand is essential for effective and sustainable operation, and the cost-effective management of water distribution networks. Unlike monthly or yearly water demand, hourly water demand has more fluctuations and is easily affected by short-term abnormal events. An effective preprocessing method is needed to capture the hourly water demand patterns and eliminate the interference of abnormal data. In this study, an innovative preprocessing framework, including a novel local outlier detection and correction method Isolation Forest (IF), an adaptive signal decomposition technique Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN), and basic forecasting models have been developed. In order to compare a promising deep learning method Gated Recurrent Unit (GRU) as a basic forecasting model with the conventional forecasting models, Support Vector Regression (SVR) and Artificial Neural Network (ANN) have been used. The results show that the proposed hybrid method can utilize the complementary advantages of the preprocessing methods to improve the accuracy of the forecasting models. The root-mean-square error of the SVR, ANN, and GRU models has been reduced by 57.5%, 27.8%, and 30.0%, respectively. Further, the GRU-based models developed in this study are superior to the other models, and the IF-CEEMDAN-GRU model has the highest accuracy. Hence, it is promising that this preprocessing framework can improve the performance of the water demand forecasting models.

Probabilistic Multi-Step-Ahead Short-Term Water Demand Forecasting with Lasso

Water demand is a highly important variable for operational control and decision making. Hence, the development of accurate forecasts is a valuable field of research to further improve the efficiency of water utilities. Focusing on probabilistic multi-step-ahead forecasting, a time series model is introduced, to capture typical autoregressive, calendar and seasonal effects, to account for time-varying variance, and to quantify the uncertainty and path-dependency of the water demand process. To deal with the high complexity of the water demand process a high-dimensional feature space is applied, which is efficiently tuned by an automatic shrinkage and selection operator (lasso). It allows to obtain an accurate, simple interpretable and fast computable forecasting model, which is well suited for real-time applications. The complete probabilistic forecasting framework allows not only for simulating the mean and the marginal properties, but also the correlation structure between hours within the forecasting horizon. For practitioners, complete probabilistic multi-step-ahead forecasts are of considerable relevance as they provide additional information about the expected aggregated or cumulative water demand, so that a statement can be made about the probability with which a water storage capacity can guarantee the supply over a certain period of time. This information allows to better control storage capacities and to better ensure the smooth operation of pumps. To appropriately evaluate the forecasting performance of the considered models, the energy score (ES) as a strictly proper multidimensional evaluation criterion, is introduced. The methodology is applied to the hourly water demand data of a German water supplier.

Unsupervised Bayesian Nonparametric Approach with Incremental Similarity Tracking of Unlabeled Water Demand Time Series for Anomaly Detection

In this paper, a fusion of unsupervised clustering and incremental similarity tracking of hourly water demand series is proposed. Current research using unsupervised methodologies to detect anomalous water is limited and may possess several limitations such as a large amount of dataset, the need to select an optimal cluster number, or low detection accuracy. Our proposed approach aims to address the need for a large amount of dataset by detecting anomaly through (1) clustering points that are relatively similar at each time step, (2) clustering points at each time step by the similarity in how they vary from each time step, and (3) to compare the incoming points with a reference shape for online anomalous trend detection. Secondly, through the use of Bayesian nonparametric approach such as the Dirichlet Process Mixture Model, the need to choose an optimal cluster number is eliminated and provides a subtle solution for ‘reserving’ an empty cluster for the future anomaly. Among the 165 randomly generated anomalies, the proposed approach detected a total of 159 anomalies and other anomalous trends present in the data. As the data is unlabeled, identified anomalous trends cannot be verified. However, results show great potential in using minimally unlabeled water demand data for a preliminary anomaly detection.

Hourly Urban Water Demand Forecasting Using the Continuous Deep Belief Echo State Network

Effective and accurate water demand prediction is an important part of the optimal scheduling of a city water supply system. A novel deep architecture model called the continuous deep belief echo state network (CDBESN) is proposed in this study for the prediction of hourly urban water demand. The CDBESN model uses a continuous deep belief network (CDBN) as the feature extraction algorithm and an echo state network (ESN) as the regression algorithm. The new architecture can model actual water demand data with fast convergence and global optimization ability. The prediction capacity of the CDBESN model is tested using historical hourly water demand data obtained from an urban waterworks in Zhuzhou, China. The performance of the proposed model is compared with those of ESN, continuous deep belief neural network, and support vector regression models. The correlation coefficient (r2), normalized root-mean-square error (NRMSE), and mean absolute percentage error (MAPE) are adopted as assessment criteria. Forecasting results obtained in the testing stage indicate that the CDBESN model has the largest r2 value of 0.995912 and the smallest NRMSE and MAPE values of 0.027163 and 2.469419, respectively. The prediction accuracy of the proposed model clearly outperforms those of the models it is compared with due to the good feature extraction ability of CDBN and the excellent feature learning ability of ESN.

A Comparison of Short-Term Water Demand Forecasting Models

This paper presents a comparison of different short-term water demand forecasting models. The comparison regards six models that differ in terms of: forecasting technique, type of forecast (deterministic or probabilistic) and the amount of data necessary for calibration. Specifically, the following are compared: a neural-network based model (ANN_WDF), a pattern-based model (Patt_WDF), two pattern-based models relying on the moving-window technique (αβ_WDF and Bakk_WDF), a probabilistic Markov chain-based model (HMC_WDF) and a naive benchmark model. The comparison is made by applying the models to seven real-life cases, making reference to the water demands observed over 2 years in district-metered areas/water distribution networks of different sizes serving a different number and type of users. The models are applied in order to forecast the hourly water demands over a 24-h time horizon. The comparison shows that a) models based on different techniques provide comparable, medium-high forecasting accuracies, but also that b) short-term water demand forecasting models based on moving-window techniques are generally the most robust and easier to set up and parameterize.

Committee Machines for Hourly Water Demand Forecasting in Water Supply Systems

Prediction models have become essential for the improvement of decision-making processes in public management and, particularly, for water supply utilities. Accurate estimation often needs to solve multimeasurement, mixed-mode, and space-time problems, typical of many engineering applications. As a result, accurate estimation of real world variables is still one of the major problems in mathematical approximation. Several individual techniques have shown very good estimation abilities. However, none of them are free from drawbacks. This paper faces the challenge of creating accurate water demand predictive models at urban scale by using so-called committee machines, which are ensemble frameworks of single machine learning models. The proposal is able to combine models of varied nature. Specifically, this paper analyzes combinations of such techniques as multilayer perceptrons, support vector machines, extreme learning machines, random forests, adaptive neural fuzzy inference systems, and the group method for data handling. Analyses are checked on two water demand datasets from Franca (Brazil). As an ensemble tool, the combined response of a committee machine outperforms any single constituent model.

Short-term forecasting of hourly water demands - a Portuguese case study

Predicting future water demands is becoming essential for the efficient management of water supply systems (WSS). To improve the operations of a Portuguese network, short-term water demand forecasting models are applied to a number of datasets collected from distinct locations in the network. Traditional forecasting models, such as exponential smoothing and naive models, and artificial neural network (ANN)-based models are developed and compared. Additionally, the influence of anthropic and weather variables in the ANN-based models is also analysed. Results demonstrate that, for this case-study, ANN-based models outperform the traditional models when external predictors such as anthropic and weather variables are included in the models. However, the inappropriate choice of such variables may lead to worse forecasting performances.

Short-term forecasting of hourly water demands - a Portuguese case study

Predicting future water demands is becoming essential for the efficient management of water supply systems (WSS). To improve the operations of a Portuguese network, short-term water demand forecasting models are applied to a number of datasets collected from distinct locations in the network. Traditional forecasting models, such as exponential smoothing and naïve models, and artificial neural network (ANN)-based models are developed and compared. Additionally, the influence of anthropic and weather variables in the ANN-based models is also analysed. Results demonstrate that, for this case-study, ANN-based models outperform the traditional models when external predictors such as anthropic and weather variables are included in the models. However, the inappropriate choice of such variables may lead to worse forecasting performances.

Tuning hyperparameters of a SVM-based water demand forecasting system through parallel global optimization

Recently, the number of machine learning based water demand forecasting solutions has been significantly increasing. Different case studies have already reported practical results proving that accurate forecasts may support optimization of operations in Water Distribution Networks (WDN). However, tuning the hyper-parameters of machine leaning algorithms is still an open problem.This paper proposes a parallel global optimization model to optimize the hyperparameters of Support Vector Machine (SVM) regression trained to provide accurate water demand forecasts in the short-time horizon (i.e. 24 h). Every SVM has the first 6 hourly water consumptions as input features and a specific hourly water demand as target to be predicted, among the remaining 18. The Mean Average Percentage Error (MAPE), computed on leave-one-out validation, is the black-box objective function optimized.Moreover, a preliminary time-series clustering has been applied in order to evaluate if this can improve the accuracy of the forecasting mechanism. Time-series clustering implies that the overall number of SVMs, whose hyperparameters are optimized through parallel global optimization, increases, with a SVM trained for each cluster identified and for each hourly water demand to be predicted, making even more critical a quick tuning of the hyperparameters.Results on the urban water demand data in Milan prove that forecasting error is significantly low and that preliminary clustering allows for further reducing error while also improving computational performances.

Forecasting Hourly Water Demands With Seasonal Autoregressive Models for Real‐Time Application

AbstractConsumer water demands are not typically measured at temporal or spatial scales adequate to support real‐time decision making, and recent approaches for estimating unobserved demands using observed hydraulic measurements are generally not capable of forecasting demands and uncertainty information. While time series modeling has shown promise for representing total system demands, these models have generally not been evaluated at spatial scales appropriate for representative real‐time modeling. This study investigates the use of a double‐seasonal time series model to capture daily and weekly autocorrelations to both total system demands and regional aggregated demands at a scale that would capture demand variability across a distribution system. Emphasis was placed on the ability to forecast demands and quantify uncertainties with results compared to traditional time series pattern‐based demand models as well as nonseasonal and single‐seasonal time series models. Additional research included the implementation of an adaptive‐parameter estimation scheme to update the time series model when unobserved changes occurred in the system. For two case studies, results showed that (1) for the smaller‐scale aggregated water demands, the log‐transformed time series model resulted in improved forecasts, (2) the double‐seasonal model outperformed other models in terms of forecasting errors, and (3) the adaptive adjustment of parameters during forecasting improved the accuracy of the generated prediction intervals. These results illustrate the capabilities of time series modeling to forecast both water demands and uncertainty estimates at spatial scales commensurate for real‐time modeling applications and provide a foundation for developing a real‐time integrated demand‐hydraulic model.

Influence of a season on hourly and daily variations in water demand patterns in a rural water supply line – case study

Abstract This article presents the results of a study on hourly and daily variations in water demand patterns, depending on a season. The study was conducted in the years 2014-2015 on a selected rural water supply line. The analysis was based on values of hourly water demand as measured by a water meter coupled with a recording device. The research showed that both the volume and fluctuations in daily water demand were higher in the spring and the summer, versus the autumn and the winter. This was most probably caused by water consumption for additional purposes, specific for rural areas. Individual water demand was the highest in the summer, and the lowest in the winter. Two peaks for hourly water demand were determined for the analyzed seasons. The morning peak always occurred at 7 a.m. on working days, while on days off work it fell at 9 a.m. or 10 a.m., depending on the season. The evening peak always fell at 8 p.m., regardless of a season or a day of a week. On working days, the evening peak was always higher than the morning one, while on days off work the morning peak was higher than the evening one in the autumn and in the winter, and both peaks were the same in the spring and in the summer.

A comparison between pattern-based and neural network short-term water demand forecasting models

In this paper, two models are set up in order to forecast hourly water demands up to 24 h ahead and are contrasted with each other. The first model (hereinafter referred to as the Patt model) is based on the representation of the periodic patterns that typically characterize water demands, such as seasonal and weekly patterns of daily water demands and daily patterns of hourly water demands. The second model is based on artificial neural networks (hereinafter referred to as ANN models). Both the models have been applied to three case studies, representing water distribution systems managed by HERA S.p.A., characterized by very different numbers of users served, and consequently very different average water demands, ranging from 900 L/s for the first case study (CS1) to about 8 L/s and1.5 L/s for the second (CS2) and third (CS3) case studies, respectively. The results show that in general, both the models, Patt and ANN, provide good accuracy for the CS1. The performances of both the models tend to decrease for CS2 and, particularly, for CS3. In particular, in the validation phase, the Patt model is more accurate than the ANN model for the CS1; for the CS2, the accuracy of the two models are very similar, and for the CS3 the accuracy of the ANN model is slightly higher than that of the Patt model.

A Short-Term Water Demand Forecasting Model Using a Moving Window on Previously Observed Data

In this article, a model for forecasting water demands over a 24-h time window using solely a pair of coefficients whose value is updated at every forecasting step is presented. The first coefficient expresses the ratio between the average water demand over the 24 h that follow the time the forecast is made and the average water demand over the 24 h that precede it. The second coefficient expresses the relationship between the average water demand in a generic hour falling within the 24-h forecasting period and the average water demand over that period. These coefficients are estimated using the information available in the weeks prior to the time of forecasting and, therefore, the model does not require any actual calibration process. The length of the time series necessary to implement the model is thus just a few weeks (3–4 weeks) and the model can be parameterized and used without there being any need to collect hourly water demand data for long periods. The application of the model to a real-life case and a comparison with results provided by another model already proposed in the scientific literature show it to be effective, robust, and easy to use.

Hourly water demand forecasting for micro water grids

Research Article| August 03 2015 Hourly water demand forecasting for micro water grids Juneseok Lee; Juneseok Lee 1Department of Civil and Environmental Engineering, San José State University, San José, CA 95112, USA and Department of Environmental Health and Safety, Eulji University, 212, Yangji-dong, Sujeong-gu, Seongnam-si, Gyeonggi-do 461-713, South Korea Search for other works by this author on: This Site PubMed Google Scholar Soo-Kwon Chae Soo-Kwon Chae 1Department of Civil and Environmental Engineering, San José State University, San José, CA 95112, USA and Department of Environmental Health and Safety, Eulji University, 212, Yangji-dong, Sujeong-gu, Seongnam-si, Gyeonggi-do 461-713, South Korea E-mail: cskwen@eulji.ac.kr Search for other works by this author on: This Site PubMed Google Scholar Journal of Water Supply: Research and Technology-Aqua (2016) 65 (1): 12–17. https://doi.org/10.2166/aqua.2015.144 Article history Received: October 28 2014 Accepted: July 02 2015 Views Icon Views Article contents Figures & tables Video Audio Supplementary Data Share Icon Share Facebook Twitter LinkedIn MailTo Tools Icon Tools Cite Icon Cite Permissions Search Site Search Dropdown Menu toolbar search search input Search input auto suggest filter your search All ContentAll JournalsThis Journal Search Advanced Search Citation Juneseok Lee, Soo-Kwon Chae; Hourly water demand forecasting for micro water grids. Journal of Water Supply: Research and Technology-Aqua 16 February 2016; 65 (1): 12–17. doi: https://doi.org/10.2166/aqua.2015.144 Download citation file: Ris (Zotero) Reference Manager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex Many countries have problems related to water scarcity and are thus seeking to promote greater water efficiency. A micro water grid (MWG) is a high-efficiency water management system that integrates information and communication technologies (ICT) for the water distribution systems in individual buildings. More accurate forecasting of hourly water demand is necessary if these systems are to function correctly and thus is the focus of this paper. Autoregressive integrated moving average (ARIMA) variant models were developed to create 24-hour lead-time forecasts. The forecast and observed values for both variant and traditional models were compared and developed models seem to perform well. It is therefore recommended to use linear stochastic models when developing MWGs for forecasting water demand to ensure sustainable water resource planning and management in MWG projects. autoregressive integrated moving average (ARIMA), hourly water demand, micro water grid (MWG) © IWA Publishing 2016 You do not currently have access to this content.

Assessment of predictive uncertainty within the framework of water demand forecasting using the Model Conditional Processor (MCP)

This paper presents an application of the Model Conditional Processor (MCP), originally proposed by Todini (2008) within the hydrological framework, to assess the predictive uncertainty in water demand forecasting related to water distribution systems. The MCP enables us to assess the probability distribution of the future water demand conditional on the forecasts provided by two or more deterministic forecasting models. In the numerical application described here, where two years of hourly water demand data for a town in northern Italy are considered, two forecasting models are applied in order to forecast hourly water demands from 1 to 24 hours ahead: the first model has a modular structure comprising a periodic component which reflects the long-term effects and a persistence component which represents the short-term memory of the process; the latter is based on neural networks. The results highlight the effectiveness of the approach, provided that the data set used for the MCP parameterization is properly selected so as to be actually representative of the accuracy of the real-time water demand forecasting models.