Without measurement, there is no understanding, control or improvement. In buildings, we measure with data, but data in and of itself is worthless without the right preparation and structure. For AI to deliver its transformative potential in facilities, the data fed into the system must be accurate, consistent, and accessible. Without these attributes, any decisions made by AI will be fundamentally flawed.
At Exergenics, we've been implementing AI solutions in critical facilities for years, including data centres, airports, and district energy facilities. From our experience, even the most advanced organizations face common issues related to operational data collection and storage, though these can be easily remedied by following some fundamental principles:
Key Data Collection Practices for AI Readiness
The success of AI depends largely on the quality of the data it processes. Here are the key factors to consider when preparing your control system for AI:
- Accuracy of Data Collection
The foundation of AI success lies in accurate, reliable data. Using high-quality sensors, ensuring regular calibration, and implementing data validation and error-checking processes are all essential. AI systems rely heavily on real-time data, so errors in data collection will lead to poor outcomes. - Improve Sensors
To truly optimize processes, granular data is crucial. Collecting data at a detailed level—such as monitoring the power consumption of each individual piece of equipment—will provide a wealth of insights that can lead to more precise AI-native optimizations. - Redundant Sensors for Critical Systems
For key operational metrics, such as chilled water temperature in cooling systems, installing redundant sensors ensures continuous data accuracy. By using three sensors to monitor the same metric, inconsistencies can be quickly identified and addressed, minimizing downtime and operational errors.
Optimizing Data Storage for AI
Data is only valuable when it is accessible and well-organized. A robust data storage infrastructure is essential to support the volume and variety of data AI systems need to function effectively.
- Tagging and Labeling for Trend Analysis
Properly tagging and labeling data allows for more effective analysis. This is especially important for AI training models, which need to identify patterns and trends in large datasets. Consistency in how data is tagged across systems and locations is key to maximizing its value. - Data Usability and Accessibility
Data locked away in inaccessible storage is as good as no data at all. Ensuring your data storage solution allows easy access to large datasets and supports common protocols like OPC UA is crucial for both AI integration and human analysis. - Regular Maintenance of Data Historian
Data historians, which store and manage large amounts of time-series data, must be properly configured to handle AI requirements. Regular backups, clear data retention policies, and robust recovery procedures ensure that the AI system can rely on clean, usable data.
Conclusion
Preparing your facility for AI involves more than just deploying the latest technology. It requires a systematic approach to data collection and stoage. By refining these practices, facilities can ensure that the data being collected is not only valuable but ready to be used by AI systems to drive operational improvements.
At Exergenics, we specialize in optimizing large faciltiies operations with AI-native models. By following best practices in data preparation, facilities can unlock AI’s full potential, leading to improved efficiency, productivity, and sustainability.