IDA Lab

Urban Water-Demand (UWD) forecasting is crucial for efficient water management, improving distribution, and supporting environmental sustainability. In tourist destinations with significant seasonal variations in number of inhabitants (water consumers), accurate water-demand forecasting becomes particularly important. This work evaluates two statistical models for short-term UWD forecasting, namely, Autoregressive Integrated Moving Average (ARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA). Two different strategies for model deployment and comparison are considered: (i) a sliding window (SW) approach with one-year (1Y) and two-year (2Y) windows for training and; (ii) a expanding window (EW) approach. The ARIMA model ployed with a Sliding Window (SW) with a two-year (2Y) resolution achieved the best overall results, followed by SARIMA considering Expanding Window (EW) model. To place these outcomes in perspective, we performed a comparison with results from related work that took into account Machine Learning methods for regression for the same data. This comparison suggests that statistical methods provide results that are both competitive and robust in terms of quality for short-term forecasts.

Traditional Machine Fault Detection (MFD) techniques usually rely on multiple sensor data sources, such as vibration, temperature, force, and audio/acoustic signals. Acoustic signals, in particular, are quite appealing in the context of MFD, as they are often among the first manifestations of machine failure. Furthermore, they are associated with high sensitivity, environmental resilience, and do not require machine interference. Given these compelling characteristics, MFD based exclusively on acoustic signals can be highly beneficial. In this work, we evaluate an unsupervised MFD approach based on Autoencoders (AE) trained exclusively on features extracted from acoustic signals of a rotating machine. The data employed in this work comes from the Machine Fault Database (MaFaulDa), which includes information from vibration and velocity sensors, besides the acoustic measurements. This allows us to compare the performance of the AE models to that of supervised models (such as MLPs) trained on the same acoustic-based feature set, as well as feature sets that incorporate all sensors from MaFaulDa. Our results support that unsupervised MFD based on Autoencoders and acoustic signals is particularly appealing, as it requires only normal machine operation for training. Indeed, we obtained AUC values of 0.86 for the task.