Forecasting Dubai City water demand using the fuzzy logic approach

The United Arab Emirates is one of the largest global consumers of desalinated water. Therefore, it is imperative to properly manage current water sources while planning for future demand and operation. One important management tool is to forecast future demand from past and present water demand. One potential prediction tool is the fuzzy logic method that can be used to model nonlinear data. One of the main advantages of the fuzzy logic method is that it does not carry many assumptions, ergodicity and stationarity of other statistical techniques. This research utilises the Mamdani approach to predict the water demand from three antecedent water consumption values, with the model analysed using the MATLAB software for four different membership functions, namely Triangular, Trapezoidal, Gaussian and the Generalised bell-shaped membership function. The analysis highlighted that the triangular, trapezoidal and generalised bell-shaped membership functions indicated a minimum error, with the Gaussian membership function demonstrating results removed from the other three membership functions. The research concludes that utilising hybrid models, improving the quality of data and utilising a robust set of rules can improve the model's performance in predicting water consumption.