Fluid Flow Monitoring and Pipeline Fault Diagnosis in Three Tank Conical System Based on Capsule Neural Network

Abstract

Fluid flow monitoring and pipeline fault diagnosis are crucial tasks in various industrial processes to ensure efficient and safe operation. Capsule Neural Network (Caps Net) used to present a novel method to ensure consistent fluid flow by diagnosis the pipeline fault in a three-tank conical tank system. Flow is continuously monitored based upon the measurement and control of pressure using Distributed Control System (DCS) and continuously monitoring its variation. A comprehensive dataset comprising sensor measurements and fault labels is used to train the Caps Net. Experimental findings show that the suggested approach is effective at identifying and categorizing problems such leaking, blockage, and valve dysfunction. The Caps Net exhibits robustness against noise and variations in operating conditions. This approach has practical implications for industrial systems, improving reliability, reducing downtime, and minimizing repairs.