Preventive Building Maintenance and Fault Detection
This project examines the optimal maintenance scheme for an entire set of building facilities such that downtime can be reduced or eliminated, the components’ life cycles can be prolonged and investment cost can be well protected.
Specifically, based on facility degradation assessment, advanced techniques such as reinforcement learning (e.g. Markov decision process, Monte Carlo methods) and stochastic dynamic programming will be suitably applied alone or in combination to minimize the cost incurred in component replacement, repair and intervention.
As a result, the project will achieve optimal maintenance policy of high adaptability and effectiveness (best practice under real world examples) at component level as well as system level over a long-term horizon.