IoT-enabled ensemble learning for sustainable wastewater valorization: a fuzzy goal programming approach with multi-objective optimization

The rapid growth of the textile industry has led to large-scale discharge of dye-contaminated wastewater, posing serious environmental and public health risks due to the persistence and toxicity of synthetic dyes. Most existing treatment models handle real-time monitoring, prediction, and optimization separately. When health impacts are considered, they are usually folded into economic costs rather than treated as a separate goal. This study proposes an IoT-enabled framework that combines stacked ensemble learning with multi-objective optimization for textile wastewater treatment and resource valorization. Internet of Things (IoT) sensors collect real-time water quality data, feeding a stacked ensemble model that predicts pollutant removal efficiency. A three-objective model then balances environmental, economic, and social sustainability: minimizing residual pollutant load, maximizing profit through resource reuse, and minimizing health risk by measuring pollutant compliance against regulatory limits. These objectives are solved jointly using weighted maximum–minimum (max–min) fuzzy goal programming (FGP). Across four operating scenarios, the framework consistently found profitable, well-balanced solutions, with overall satisfaction between 66 and 73% and daily profits ranging from ₹162–1,461. Notably, the ML performance level maintained during operation was found to drive sustainable outcomes more strongly than a model's fixed baseline accuracy, giving operators a practical, real-time lever for advancing environmental protection, economic resilience, and public health together.