2020/07/07 by Kevin Shen Hoong Ong, Dusit Niyato, Ong, Kevin Shen Hoong +3
Computer Science · Environmental Science · #Air Quality Monitoring and Forecasting #Anomaly Detection Techniques and Applications #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML)
paper · pdf · doi:10.48550/arxiv.2007.03313
openalex publication_date 2020/07/07 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Failure of mission-critical equipment interrupts production and results in\nmonetary loss. The risk of unplanned equipment downtime can be minimized\nthrough Predictive Maintenance of revenue generating assets to ensure optimal\nperformance and safe operation of equipment. However, the increased\nsensorization of the equipment generates a data deluge, and existing\nmachine-learning based predictive model alone becomes inadequate for timely\nequipment condition predictions. In this paper, a model-free Deep Reinforcement\nLearning algorithm is proposed for predictive equipment maintenance from an\nequipment-based sensor network context. Within each equipment, a sensor device\naggregates raw sensor data, and the equipment health status is analyzed for\nanomalous events. Unlike traditional black-box regression models, the proposed\nalgorithm self-learns an optimal maintenance policy and provides actionable\nrecommendation for each equipment. Our experimental results demonstrate the\npotential for broader range of equipment maintenance applications as an\nautomatic learning framework.\n