Modeling and Experimental Verification of Solid-Liquid Two-Phase Flow Long-Distance Pipeline Friction Drag Loss Based on LS-SVM

Abstract

The mechanism model needs to assume a lot of prerequisites, but it is lack of these conditions in the solid-liquid two-phase flow in long-distance pipeline, therefore there will be a deviation between the friction drag loss of mechanism model and the actual value. This paper adopted the least square support vector machine (LS-SVM) to fix the value of the mechanism model, and increase the prediction accuracy. Meanwhile, for improving real-time online LS-SVM performance, introducing the local LS-SVM. The experiment result shows, LS-SVM and local LS-SVM greatly improved the forecast accuracy, compared with the mechanism model correction.