An Artificial Intelligence Based Self-Adaptive Dynamic Process Control System for Enhancing In-Situ Bioremediation of Benzene-Contaminated Groundwater

Abstract

This study develops a new AI-based Self-Adaptive DPC (SADPC) system based on stepwise inference combing with genetic algorithm optimization technologies, including a filtered-clustering inference prediction model (FCI simulator), a stepwise inference controller (SI emulator), a model predictive control controller (MPC controller), a 1st-stage optimizer, and a 2nd-stage optimizer. This system effectively reflects the dynamics and complexity of the biodegradation process and realizes the control for the remediation system based on the feedback information. To achieve this goal, a statistical model for simulating the bioremediation process through the FCI simulator is proposed, which can predict the resulting contamination situation based on the previous contamination situation and control action. Then a bridge between control actions and contamination situations is established through the SI emulator, which can generate a control action based on a given contamination situation. Through running the SADPC system, the desired control action can be identified. Results show that The SADPC system increases the removal rate of benzene and arrives at the remediation goal earlier than other systems. This suggested decision makers that guidelines and policies on remediation-oriented SADPC systems could be tentatively investigated, developed, and applied in the future effort.