Optimization of Coagulant Dosing Process for Water Purification Based on Artificial Neural Networks

Abstract

AbstractThe article reveals the essence and features of the neural network model used to regulate water purification processes. The peculiarities, principles and main stages of water purification are determined, which consist of the use of rectification and deposition of separated particles, pre-filtration. It is noted that each process is closely related to the other. The stages of modeling an artificial neural network with descriptions of each step sequentially are revealed. The approach to the use of artificial neural networks during dosing of the mixture for water purification is substantiated. The process of dosing the mixture for water purification and related indicators that are influential for the implementation of the main process are analyzed. A number of factors that directly affect the coagulation process and, as a consequence, the structure of the neural network include turbidity, flow rate and working pumps. However, it is emphasized that there are other parameters, such as pH, conductivity and water temperature, which also have a slow effect on rectification and coagulation, but are not critical. The coagulant regulates turbidity, minimizing the cost of production, which is a very important factor in the economic efficiency of the enterprise. It is emphasized that determining the dose of coagulant is necessary to minimize time, implementation without an intermittent process, stabilize variations in the observations of the operator and improve the quality of the end result. Taking into account the parameters of influence, it is proposed to establish a special regime of coagulant dose control, on the basis of which a control scheme is developed, which is presented schematically with the separation of the main processes occurring during coagulant dosing. Given the complexity of building a neural network model, a neural network model with uncontrolled learning is proposed, which is used to construct multidimensional space in space with a lower dimension to optimize coagulant dosing in water purification.KeywordsArtificial neural networkAutomatic dosingCoagulantWaterPurificationSuspended solidsConcentrationIntelligent information system