Automated fault detection in commercial refrigeration systems

A refrigerant leak is one of the major contributors of commercial refrigeration unexpected breakdowns. Conventional methods of leak detection using physical sensors are expensive, have limited capability of square footage coverage, and incapable of detecting slow and progressive refrigerant leaks. Accordingly, the development of a smart leak detection system without the need to include additional physical sensors using AI models based on actual operating conditions could significantly reduce the overhead costs associated with system shutdowns and refrigerant fill-ups in grocery stores. The goal for this project is to analyze data collected from various systems, driving meaningful insights, and developing AI models for detecting refrigerant leaks in the form of anomalies. The outcome of achieving the project objectives would have a significant environmental impact, substantial cost-saving, and most importantly, reduce human efforts and erroneous leak maintenance and monitoring processes.

Faculty Supervisor:

Ayan Sadhu

Student:

Partner:

Kalder at Neelands

Discipline:

Engineering

Sector:

Construction and infrastructure

University:

The University of Western Ontario

Program: