LAUSR

Abstract The paper discusses the use of Decision Support Systems (artificial neural networks analysis preceded by Principal Component Analysis) for the assessment of domestic sewage filtration effectiveness with four types of waste serving as filling materials in vertical flow filters. The study analyzed the effectiveness of pollution removal from wastewater by mechanically shredded waste in the form of PET flakes, polyurethane foam trims, shredded rubber tires and wadding. The organic compounds (CODcr, BOD 5 ) removal, suspended solids, biogenic compounds (N-NH 4 + , PO 4 3− ) and oxygen saturation changing compared with reference sand filling was analyzed. The paper presents the proposal for the use of artificial neural networks as a tool to support decision making on the selection of waste material, filling vertical filters cooperating with the septic tank. An analysis of the functioning of the trained neural network was performed, comparing its responses with the reduction values obtained for individual fillings under changing hydraulic conditions. Generally good agreement between the predictions of the neural model and the reduction values was obtained for the MLP 11-7-2 network.