Document Type
Patent
Publication Date
3-31-2026
Abstract
A computer-implemented method, system and computer program product for detecting or predicting system faults in cooling systems. A model (deep learning model) is built and trained to detect or predict system faults in a cooling system based on acoustic emission signals (both in temporal and frequency domains) and/or imaging signals. Upon training the model to detect or predict system faults in a cooling system, acoustic emission signals may be obtained non-intrusively from the cooling system using acoustic emission sensors, hydrophones and/or microphones. Additionally, upon training the model to detect or predict system faults in a cooling system, imaging signals (e.g., boiling images) may be obtained non-intrusively from the cooling system using optical sensors (e.g., high-speed camera). The trained model may then detect or predict a system fault in the cooling system based on such information (acoustic emission signals, including in temporal and frequency domains, and/or the imaging signals).
Department
Mechanical Engineering
Patent Number
US12591230
Application Number
US 20230195094
Application Published
6-22-2023
Application Filed
12-9-2022
Assignee
Board of Trustees of the University of Arkansas (Little Rock, AR)
Citation
Hu, H., Pandey, H., & Dunlap, C. (2026). Detecting or predicting system faults in cooling systems in a non-intrusive manner using deep learning. Patents Granted. Retrieved from https://scholarworks.uark.edu/pat/514
Comments
Han Hu, Department of Mechanical Engineering, University of Arkansas, Fayetteville, AR
Hari Pandey, Department of Mechanical Engineering, University of Arkansas, Fayetteville, AR
Christy Dunlap, Department of Mechanical Engineering, University of Arkansas, Fayetteville, AR