Using extreme learning machines for short-term urban water demand forecasting

This study explores the ability of various machine learning methods to improve the accuracy of urban water demand forecasting for the city of Montreal (Canada). Artificial Neural Network (ANN), Support Vector Regression (SVR) and Extreme Learning Machine (ELM) models, in addition to a traditional model (Multiple linear regression, MLR) were developed to forecast urban water demand at lead times of 1 and 3 days. The use of models based on ELM in water demand forecasting has not previously been explored in much detail. Models were based on different combinations of the main input variables (e.g., daily maximum temperature, daily total precipitation and daily water demand), for which data were available for Montreal, Canada between 1999 and 2010. Based on the squared coefficient of determination, the root mean square error and an examination of the residuals, ELM models provided greater accuracy than MLR, ANN or SVR models in forecasting Montreal urban water demand for 1 day and 3 days ahead, and can be considered a promising method for short-term urban water demand forecasting.     

A DECISION SUPPORT SYSTEM FOR DROUGHT CHARACTERIZATION AND MANAGEMENT

A decision support system has been developed for drought characterization and management. The purpose of the decision support system is to assist the operators and water managers of the water supply system of the City of Lexington, Kentucky. The motivation of this study was a severe drought that occurred in the state of Kentucky during the summer of 1988. The data derived from the City of Lexington, Kentucky and the Kentucky River Basin were employed in this study. The developed decision support system consists of three components: a water demand forecasting component, a streamflow forecasting component, and an integrated expert system component. The water demand and streamflow forecasting components of the decision support system predict the water consumption for the City of Lexington and the flow in Kentucky River at Lock 10 near Winchester, Kentucky, respectively. The lead time of the forecasting models was taken as five days as they were intended to be employed in developing a short-term drought management policy. Various modeling techniques ranging from regression and time series analysis to the relatively new technique of expert systems and artificial neural networks were explored for forecasting both water demand and streamflow. The integrated expert system component consists of five sub-components. Each sub-component entails developing a knowledge base for a specific purpose. The expert system component integrates all sub-components and characterizes the drought potential in the coming five days and recommends a drought management policy for the week to come. The developed decision support system is capable of running on a persona] computer “and provides a user-friendly platform for decision-makers to explore a wide range of drought management alternatives.     

城市供水管网漏损时间的预测模型研究

城市供水企业迫切需要加强给水管网的漏损管理，以减少漏损水量和提高经济效益。在对华北某市供水管网漏损数据进行统计和分析的基础上，按照管段实际发生漏损次数分两种情况建立了供水管网漏损时间的预测模型，对漏损次数≤4次的管段采用基于SAS系统的多元线性回归方法，对漏损次数〉4次的管段则采用灰色预测方法。经实例验证，多元线性回归方法预测的平均相对误差为21％，灰色预测方法预测的平均相对误差〈6％，整套模型的精度可满足城市供水管网漏损宏观管理的需要，能够提高管网漏损防治的效率。     

An ARIMA-NARX hybrid model for forecasting urban water consumption (case study: Tehran metropolis)

ABSTRACT

In this research, an ARIMA-NARX (Autoregressive Integrated Moving Average-Nonlinear Auto-Regressive eXogenous) hybrid model is proposed to forecast daily Urban Water Consumption (UWC) for Tehran Metropolis. The linear and nonlinear component of the UWC was forecast by ARIMA as a linear forecasting model and the artificial neural network as a nonlinear forecasting model, respectively. An alternative hybrid model including sunshine hour in addition to the previous studies’ predictors (the minimum, maximum and average temperature, relative humidity and precipitation) was selected as the superior alternative model. Then, the performance of proposed model was compared with ARIMA and NARX models. The results showed that the hybrid model, which benefits from capability of both linear and nonlinear models, has a higher accuracy than the other two models in forecasting UWC. Therefore, the proposed hybrid model has better results in UWC forecasting and, as a consequence, better urban water reservoir management will be provided.     

Large metropolitan water demand forecasting using DAN2, FTDNN,and KNN models: A case study of the city of Tehran,Iran

Efficient operation of urban water systems necessitates accurate water demand forecasting. We present daily, weekly, and monthly water demand forecasting using dynamic artificial neural network (DAN2), focused time-delay neural network (FTDNN), and K-nearest neighbor (KNN) models for the city of Tehran. The daily model investigates whether partitioning weekdays into weekends and non-weekends can improve forecast results; it did not. The weekly model yielded good results by using the summation of the daily forecast values into their corresponding weeks. The monthly results showed that partitioning the year into high and low seasons can improve forecast accuracy. All three models offer very good results for water demand forecasting. DAN2, the best model, yielded forecasting accuracies of 96%, 99%, and 98%, for daily, weekly, and monthly models respectively.     

Generation of synthetic water demand time series at different temporal and spatial aggregation levels

This paper presents a procedure for the generation and spatial-temporal aggregation of synthetic water demand time series which reproduce the main statistics - mean, variance and (spatial and temporal) covariance - of the corresponding observed series. Starting from observed historical time series taken at low levels of temporal aggregation (e.g., one minute) and relating to individual users, the procedure enables a) the generation of synthetic water demand time series for every individual user with a time step of one minute, b) the temporal aggregation of these synthetic series in order to obtain synthetic water demand time series with a time step, for example, of one hour, and which are such as to reproduce the hourly mean, variance and temporal covariances of the corresponding temporally aggregated historical time series, and c) the spatial aggregation of the synthetic hourly water demand time series of every user in order to generate a synthetic water demand time series that is representative of the entire group of users considered, and is such as to reproduce the mean, variance and temporal covariance observed at that level of spatial aggregation; The entire procedure was parameterized and applied to a case study on the water demands of 21 users of the water distribution system of Milford (Ohio). The results obtained show that the temporal aggregation procedure is effective in generating hourly water demand time series that preserve the mean, variance and temporal correlation of the historical time series for every individual user, while the spatial aggregation method shows good level of effectiveness in preserving the statistics of the aggregated series. Overall, the proposed procedure is demonstrated to be a valid tool for the bottom-up generation of synthetic water demand time series at various levels of spatial-temporal aggregation which reproduce the mean, variance and covariance statistics of the historical time series.     

Forecasting construction industry demand,price and productivity in Singapore: the BoxJenkins approach

In academic research, the traditional Box-Jenkins approach is widely acknowledged as a benchmark technique for univariate methods because of its structured modelling basis and acceptable forecasting performance. This study examines the versatility of this approach by applying it to analyse and forecast three distinct variables of the construction industry, namely, tender price, construction demand and productivity, based on case studies of Singapore. In order to assess the adequacy of the Box-Jenkins approach to construction industry forecasting, the models derived are evaluated on their predictive accuracy based on out-of-sample forecasts. Two measures of accuracy are adopted, the root mean-square-error (RMSE) and the mean absolute percentage error (MAPE). The conclusive findings of the study include: (1) the prediction RMSE of all three models is consistently smaller than the model's standard error, implying the models' good predictive performance; (2) the prediction MAPE of all three models consistently falls within the general acceptable limit of 10%; and (3) among the three models, the most accurate is the demand model which has the lowest MAPE, followed by the price model and the productivity model.     

An ensemble wavelet bootstrap machine learning approach to water demand forecasting: a case study in the city of Calgary,Canada

This paper explores a hybrid wavelet, bootstrap and neural network (WBNN) modeling approach for daily (1, 3 and 5 day) urban water demand forecasting in situations with limited data availability. This method was tested using 3 years of daily water demand and meteorological data for the city of Calgary, Alberta, Canada. The performance of the WBNN method was compared to that of three other methods: traditional neural networks (NN), wavelet NNs (WNN), and bootstrap-based NN (BNN) models. While the hybrid WBNN and WNN models equally provided 1-day lead-time forecasts of greater accuracy than those obtained with other methods, for longer lead-time (3- or 5-day) forecasts the WBNN model alone outperformed the other models. The confidence bands generated using the WBNN model displayed the uncertainty associated with the forecasts.     

基于小波分解的城市供水管网短期水量负荷预测

利用小波分解和人工神经网络相结合的方法建立了城市供水管网短期水量负荷的组合预测模型。该方法首先利用小波分解技术将时负荷水量分解为相对简单的带通分量信号，然后根据各分量信号的特点分别建立独立的神经网络预测模型，最后将预报结果集成。由于小波分解后各分量的频率相对单一，因而可有效缩短网络训练时间，提高计算速度。仿真计算结果表明，该方法建模合理、计算量适中，可准确预测管网水量。     

Application of modified complementary reservoir approach in analysis of water distribution networks under pressure-deficient conditions

Hydraulic simulation models which simulate water distribution systems in different operating conditions are essential tools to evaluate network reliability. Simulation models found in most commercial software are not effectively capable of analyzing demand nodes under critical conditions such as fire-fighting demand or network pipe breakage. In the current study, a combination of hydraulic model and complementary reservoir solution (CRS) method is used for solving network's problems in critical conditions for both series-looped networks and a part of water network in Ilam city (Iran). Obtained results show that CRS provide more than actual need on the demand node for some reservoir total pressure heads. Thus, two modified versions of CRS method are proposed to deal with failures of CRS in such cases. Obtained results demonstrate improved efficiency in the combined model for analyzing networks in abnormal conditions.     

Evaluating the performance of combining neural networks and genetic algorithms to forecast construction demand: the case of the Singapore residential sector

In recent years, forecasting demand for residential construction in Singapore has become more vital, since it is widely perceived that the next trough of the real estate cycle is approaching. This paper evaluates the use of a combination of neural networks (NNs) and genetic algorithms (GAs) to forecast residential construction demand in Singapore. Successful applications of NNs, especially in solving complex non-linear problems, have since stimulated interest in exploring the capabilities of other biological-based methods such as GAs, and in exploiting the synergy of these two techniques to create more problem-solving power. In the study, a basic NN model is used as a benchmark to gauge the performance of the combined NN-GA model. A relative measure of forecasting accuracy, known as the mean absolute percentage error (MAPE), is used for the comparison. The models are checked also for internal validity by allowing each to be trained twice and having a set of forecasts generated after each training. Both models are found to produce accurate forecasts, because their MAPE values consistently fall within the acceptable limit of 10%. However, the combined model out-performs the basis model remarkably by reducing the average MAPE from about 6% to a mere 1%. For each model, the marginal difference in the MAPE values (i.e., 0.5% for the NN model and 0.06% for the NN-GA model) of its two forecasts indicates consistency in performance, hence establishing internal validity as well. The findings reinforce the reliability of using NNs to model construction demand and reveal the benefit of combining NNs and GAs to produce more accurate models.     

A model of the filling process of an intermittent distribution network

In many countries, private tanks are acquired by users to reduce their vulnerability to intermittent supply. The presence of these local reservoirs modifies the user demand pattern and usually increases user water demand at the beginning of the service period depending on the tank filling process. This practice is thus responsible for the inequality that occurs among users: those located in advantaged positions of the network are able to obtain water resources soon after the service period begins, while disadvantaged users have to wait much longer, after the network is full. This dynamic process requires the development of ad hoc models in order to obtain reliable results. This paper discusses a numerical model used for evaluating this complex process as well as the application of model to an Italian case study. The model agreed with calibration data and provided interesting insights into the network filling process.     

Forecasting short-term air passenger demand using big data from search engine queries

Forecasting air passenger demand is a critical aspect of formulating appropriate operation plans in airport operation. Airport operation not only requires long-term demand forecasting to establish long-term plans, but also short-term demand forecasting for more immediate concerns. Most airports forecast their short-term passenger demand based on experience, which provides limited forecasting accuracy, depending on the level of expertise. For accurate short-term forecasting independent of the level of expertise, it is necessary to create reliable short-term forecasting models and to reflect short-term fluctuations in air passenger demand. This study aims to develop a forecasting model of short-term air passenger demand using big data from search queries to identify these short-term fluctuations. The suggested forecasting model presents an average forecast error of 5.3% and indicates that an increase of approximately 195,000 air passengers is to be expected 8 months later, as the key query frequencies increase by 0.1%.     

分区计量供水管网水力模型的流量分配

结合天津市供水管网实例,通过分析其计量水量的特点与水力模型的构建要求,按照水量数据来源分别侧重于小区表、户表和在线流量计,提供了分区计量供水管网水力模型的三个流量分配方案;从数据健全度、流量分配准确度、实施难度、流量分配校正依据、漏损考察功能、模型动态模拟、模型维护与应用难度和模型构建平台八个方面对三个流量分配方案进行了多角度评价,可为水力模型项目的实施提供参考。     

The influence of coping appraisals on the adoption of decentralised water systems in Australia

In Australia, water scarcity has resulted in the need for re-evaluating demand management policies, as well as the identification of alternative water supplies. Specifically, water utilities have been focusing on increasing the adoption of household level decentralised water systems (DWS). While such engineering solutions may be effective, understanding the factors which influence adoption is crucial for widespread uptake. Protection Motivation (PM) theory was used to assess DWS adoption in a sample of 295 homeowners across South East Queensland, Australia. Results provided good support for the application of PM theory to understanding whether people cope adaptively to water shortage threats. The model’s hypothesised link between adaptive coping and behavioural intention was also supported, suggesting PM theory can be used to understand people’s intention to adopt DWS in the context of the drought. The predictive ability of the PM model improved significantly when demographic variables – age and perceived water – use were included.     

改进单指数平滑法预测某市日用水量

为实现供水管网经济、可靠、科学的优化调配用水量,给出一种基于改进单指数平滑预测方法,该预测方法引进"追踪信号"来反应时间序列的变化,通过重新修正平滑常数a以建立改进单指数预测模型。以东北某城市日用水量为原始数据进行了实际预测,模型精度检验的结果满足Y市用水量要求,该预测模型应用于Y市的日用水量预测,为Y市供水优化调配提供有效依据。     

Comparison of the performance of four measure-correlate-predict algorithms

Measure-correlate-predict (MCP) algorithms are used to predict the wind resource at target sites for wind power development. This paper describes some of the MCP approaches found in the literature and then compares the performance of four of them, using a common set of data from a variety of sites (complex terrain, coastal, offshore). The algorithms that are compared include a linear regression model, a model using distributions of ratios of the wind speeds at the two sites, a vector regression method, and a method based on the ratio of the standard deviations of the two data sets. The MCP algorithms are compared using a set of performance metrics that are consistent with the ultimate goals of the MCP process. The six different metrics characterize the estimation of (1) the correct mean wind speed, (2) the correct wind speed distribution, (3) the correct annual energy production at the target site, assuming a sample wind turbine power curve, and (4) the correct wind direction distribution. The results indicate that the method using the ratio of the standard deviations of the two data sets and the model that uses the distribution of ratios of the wind speeds at the two sites work the best. The linear regression model and the vector regression model give biased estimates of a number of the metrics, due to the characteristics of linear regression.