A method for detecting dynamic ice storage air conditioning cooling pipeline flow stagnation

Abstract

This invention discloses a machine learning-based method for detecting flow obstruction in dynamic ice storage air conditioning cooling pipelines. It collects various flow condition data through multiphase flow computational fluid dynamics, and establishes a flow obstruction risk level criterion based on the proportion and uneven distribution of ice particles at the top of the pipe. A logistic regression algorithm is used to obtain the mathematical correlation between pipeline operating parameters and blockage status, establishing a machine learning model suitable for detecting flow obstruction in ice slurry cooling pipelines. This machine learning model takes the pipe diameter, flow velocity, solid-phase ice particle to liquid-phase carrier density ratio, initial ice particle concentration, and particle size as inputs, and the flow obstruction risk as the output. Based on probabilistic statistical principles, it predicts the flow obstruction risk level. This method not only quickly and accurately identifies the flow obstruction state of ice slurry cooling pipelines, but also ensures that the required input parameters are easily monitored during actual operation.

Images