| I&T Solution |
Revolutionizing Fault Detection in Air Handling Units and Chiller Systems in Healthcare Institutions Proposal
(REF: S-1979) |
| Trial Project |
|
| Solution Feature |
- Utilizes AI-FDD models to analyze real-time IoT sensor data and operational logs for early fault detection and prediction.
- Features a user-friendly graphical interface that provides real-time monitoring and automatically dispatches alerts to maintenance personnel.
- Employs One-class SVM models for real-time anomaly scoring and equipment health index generation without needing prior fault records.
- Leverages advanced Transformer models trained on tokenized sensor and log data to predict faults before they occur.
- The AI model continuously improves its performance based on operator feedback through an event-driven retraining system.
|
| Trial Application and Expected Outcome |
- Reduction in Fault Occurrences: Proactive identification of potential issues leads to a significant decrease in unplanned equipment downtime.
- Enhanced Operational Efficiency: Shift from reactive, corrective maintenance to predictive maintenance, optimizing resource allocation and response times.
- Improved Equipment Health Management: Real-time health indices and alerts enable timely interventions, prolonging equipment lifespan.
- Increased Client Satisfaction: Reliable system operation minimizes disruptions to critical clinical services in healthcare institutions
- Data-Driven Continuous Improvement: The self-learning capability of the system ensures long-term accuracy and adaptability to changing conditions.
|
| Additional Solution Information |
FLAIR proposal.pdf
|
| Info on I&T Solution Provider |
| Solution Provider | : | Hong Kong Industrial Artificial Intelligence & Robotics Centre Limited | | Address | : | 5/F, HKPC Building, 78 Tat Chee Avenue, Kowloon, Hong Kong | | Contact Person | : | Ping Ying Yang |
| Position | : | Assistant Manager | | Tel | : | 852 63682827 | | Email | : |
norayang@hkflair.org | | Webpage | : | https://www.hkflair.org/ |
|
