A Data-Driven Soft Sensor to Forecast Energy Consumption in Wastewater Treatment Plants: A Case Study

Energy consumption is vital to the global costs of wastewater treatment plants (WWTPs). With increase installed WWTPs worldwide, modeling and forecast their energy have become a critical factor in WWTP design meet environmental economic requirements. The accurate swift forecasting soft-sensors are not only supportive daily electric financial budgeting by practitioners on micro-scale, but also beneficial local municipal operation fundamental regional impact estimation macro-scale. influenced different biological factors, making it complicated challenging build soft-sensors. This article intends provide short-term based data-driven soft sensors using traditional time-series deep learning methods. Ten sensors, including ordinary least square, exponential smoothing state space, regression, auto-regressive integrated moving average (ARIMA), structural time series model, Bayesian series, non-linear auto-regressive, long memory with without updates, gated recurrent units been investigated compared for forecasting. data from membrane bioreactor-based middle east used evaluate performances proposed Results showed that ARIMA achieved slightly improved performances, among others. employment adaptive learning-based expected enhance capabilities models quickly accurately follow trend future data.