Establishment of a projection-pursuit-regression-based prediction model for the filtration performance of a micro-pressure filtration and cleaning tank for micro-irrigation

Abstract

Filtration is a key process of a micro-irrigation system. To achieve water conservation, improve the energy efficiency, and meet the low-carbon environmental requirements, this study investigates a new type of pre-pump filtration measure—the micro-pressure filtration and cleaning tank, and the sediment removal (SR), mass of intercepted sediment (Ma), and filtration time (t) are used as evaluation indicators. A uniform orthogonal test was designed to evaluate the influences of the initial sediment concentration, flow rate, and filter area. We carried out simulations using the projection pursuit regression (PPR) method based on physical measurement data, and then NSGA-II (fast elitist non-dominated sorting genetic algorithm) algorithm was used to find the optimal parameter values. The results showed that the order of the factors influencing the SR and Ma of the micro-pressure filtration and cleaning tank was the filter area > initial sediment concentration > flow rate. The order of the factors influencing the filtration time was flow rate > initial sediment concentration > filter area. A PPR prediction model was built based on the filter area, initial sediment concentration, flow rate, and assessment indicators, and the optimal operating conditions were determined (initial sediment concentration: 0.36 kg/m3, flow rate: 7 m3/h, and filter area: 2000 cm2). The research results demonstrated the filtration performance of the micro-pressure filtration and cleaning tank and provide a theoretical basis for the popularization and application of such tanks as well as a reference for sustainable new irrigation technologies.