AI-Based Guidance System of Operating Parameters for Municipal Wastewater Treatment

A real-time guidance system of operating parameters in wastewater treatment process using AI-based algorithm (random forest) was developed and evaluated for its predicting power using full-scale data collected offline from a municipal wastewater treatment plant. In the course of model development, the effects of input data volume and random forest parameters on predictions were investigated and these model creation conditions were fixed individually for air flow rate and waste activated sludge withdrawal rate predictions. Using training data of 2 years with limited number of process variables, the system was able to give predictions of these operating parameters with acceptable error (MAPE) of less than 10% on annual average basis. These results demonstrated that the AI-based guidance system was, once trained with past data, able to trace the operating parameters by skilled operators with practical error range.