Predictive prognostic modeling of refrigeration systems

• Main Author: Yong, Chang Xin
• Other Authors: Meng-Hiot Lim
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/167098
• Institution: Nanyang Technological University

Refrigeration systems are widely used in various industries, and the reliability of its operation is largely crucial for efficient processes. Predictive prognostic modelling techniques can be used to anticipate and prevent potential uprisal of issues in these systems, hence reducing maintenance costs and machinery downtime. With the combination of chiller system fundamentals, domain knowledge and a database of chiller operating information, this project aims to develop a predictive prognostic model to predict fault occurrences in refrigeration systems. The project utilizes historical data from a cooling system in an industrial building setting to develop a predictive model using Long Short-Term Memory (LSTM). The data provided includes various sensor readings, such as temperature, power, and flow rate, along with records of failure events. The model is trained to predict and classify future failure events, as well as alert users early, of any potential issues. The developed model achieved a high level of accuracy in predicting failure events which allowed us to identify warning signs early. The model's output aided the maintenance team in proactively addressing these potential issues, hence reducing chiller downtime and maintenance costs. In conclusion, the model developed in this study was effective in anticipating and preventing potential issues in the chiller system. This model can then be used to improve the reliability and efficiency of other industrial chiller systems, further reducing costs and halts due to faults. The terms “refrigeration”, “chiller” and “HVAC” systems were used synonymously in this thesis.