<![CDATA[Hospitals generate domestic wastewater that must be treated through a Wastewater Treatment Plant (WWTP) before discharge into the environment, particularly through a coagulation stage using alum to reduce turbidity and Total Suspended Solids (TSS). Existing automatic control systems are generally rule-based and prone to coagulant overdosing due to the non-linear nature of the coagulation process. This study aims to design an intelligent WWTP control system based on an Artificial Neural Network (ANN) with a Backpropagation learning algorithm to determine the optimal activation duration of an alum dosing pump based on pH and turbidity (NTU) sensor inputs. The research employed a Research and Development (R&D) method with a quantitative experimental approach. The ANN model was built using a Multi-Layer Perceptron (MLP) architecture with a 2-5-1 configuration and a sigmoid activation function, implemented on an ESP32 microcontroller integrated with Node-RED for real-time remote monitoring, and tested directly at the WWTP of IBI Mother and Child Hospital, Surabaya. The results show that the ANN Backpropagation model successfully determined the coagulant dosing duration with an overall regression correlation coefficient (R All) of 0.94795 and a Mean Absolute Error (MAE) of 0.203502 across 100 test data points, equivalent to a deviation of about 0.2 seconds. Furthermore, the conversion relationship between turbidity (NTU) and TSS was formulated through the linear regression equation TSS = 0.384 × NTU − 5.80, with a coefficient of determination R² ≈ 0.94. These findings demonstrate that the designed system operates accurately, efficiently, and adaptively in controlling the WWTP coagulation process while maintaining effluent water quality within regulatory standards.]]>
