CONTROL OF REAGENT OXIDATION-REDUCTION PROCESSES OF IMPURITIES IN INDUSTRIAL WASTEWATER USING ELEMENTS OF ARTIFICIAL INTELLIGENCE

Abstract

The article presents the results of research on the creation of an automated system for reagent oxidation-reduction of toxic impurities found in multicomponent wastewaters of enterprises. It is proposed to control the dosing of reagents and the purification process using such parameters as the active reaction (pH) and the oxidation-reduction potential (Eh) of the medium. A characteristic feature of multicomponent wastewater is the non-stationarity of wastewater in terms of composition and concentration of impurities, as well as a change in the kinetics of the pH and Eh parameters during the reagent dosing process. It should be taken into account that the reagents themselves affect the parameters of the medium, which change during the course of chemical reactions, and pH and Eh also mutually influence each other in opposite directions. As a result, the purification process occurs under conditions of significant uncertainty of the parameters. Automated control of the oxidation/reduction process, which includes a step-by-step pulse dosing of reagents into water, is recommended. At each stage, a reagent is dosed into the water to change the pH, then an oxidant or reductant to change Eh to some intermediate value and a holding time to complete the chemical reaction. The process leads to the gradual achievement of the final Eh, necessary for complete oxidation/reduction of impurities. Since the cleaning conditions are constantly changing and it is necessary to make decisions with a significant number of pH and Eh parameter values, it is proposed to use artificial intelligence elements to control the cleaning process. AI technologies make it possible to analyze large data sets, identify patterns and adaptively control the dosing of reagents. In such conditions, a combined approach to the use of AI may be the most effective. The basic regulation is performed by a classic PID or MPC controller, and the superstructure in the form of artificial intelligence is RL. Fuzzy logic or expert rules can be used to ensure the adaptability of the system to changes in the composition of wastewater and parameters of the cleaning process. The key to implementing such systems is the use of digital twins for preliminary training and testing of algorithms.