Identification Mechanism of Abnormal Operating Conditions in Heat Exchange Networks Based on the KF-PINN Algorithm
Ming Shi, Ling Na Sun · 2025
Continuously optimizing the utilization of process waste heat through heat exchange networks is an important part of improving energy efficiency, reducing emissions and promoting sustainable development. Today, due to heat exchange networks are widely applied in process industries, their condition monitoring is of great importance. This paper presents a fusion algorithm of Kalman filer and physical information neural network model for anomaly detection and localization in heat exchange networks. More specifically, the proposed algorithm above to predict some coefficients of heat exchanger. Afterward, comparing them with the actual coefficients to identify and locate anomaly or leakage points in the heat exchange network. The performance of the proposed method is validated with simulation and experimental scenarios. The results indicate the proposed algorithm promising accuracy and good speed of detection and location.