Rapid and inexpensive estimation of biochemical oxygen demand and sludge volume index using AI-based soft sensors

Sustainable wastewater management is essential for protecting aquatic ecosystems and conserving resources. However, traditional methods for monitoring critical parameters, like Biochemical Oxygen Demand (BOD) and Sludge Volume Index (SVI), are time-intensive, requiring chemical reagents and manual processes that increase treatment costs and environmental impact. This study introduces an artificial intelligence (AI)-driven soft sensor to enable rapid estimation of BOD and SVI, enhancing sustainable process control in wastewater treatment plants (WWTPs). The proposed framework utilizes Artificial Neural Networks (ANNs) driven by rapid-measurement variables: total dissolved solids (TDS), dissolved oxygen (DO), and the TDS/DO ratio for effluent BOD estimation, and sludge volume (SV), TDS, and temperature for SVI estimation, alongside a numerical plant identifier. Unlike prior soft sensor approaches that rely on complex, laboratory-intensive input parameters, this work demonstrates that accurate estimation is achievable using only simple, rapidly measurable inputs, with a single model generalized across multiple treatment plants. The model demonstrated high estimation accuracy, outperforming conventional regression techniques in terms of RMSE, R2, MSE, and MAE. By reducing reliance on frequent laboratory testing and enabling proactive operational adjustments, it can potentially contribute to optimized energy consumption, reduced chemical use, and support for circular water management. This AI-powered framework contributes to the development of smart, sustainable WWTPs, aligning global sustainability goals for resource efficiency and environmental protection.