How artificial intelligence and machine learning tools deliver smart building benefits ?
For many commercial building owners, the ability to capture and analyze facility performance data is rapidly emerging as a high priority and as a competitive differentiator. By focusing on data that pertains to both energy consumption and building asset maintenance, building owners, thanks to new artificial intelligence (AI) and machine learning capabilities, can now drastically reduce these two principal operational cost drivers. The development of IT and artificial intelligence tools are enabling data to become a more strategic business asset, and for the smart building vision to become a reality. Data collection, consolidation, and analysis can now be used to improve operational uptime, carbon footprint, economic performance, and occupant comfort and safety.
At Automatique & Industrie, a Schneider Electric Certified EcoXpert partner with extensive experience in smart building management systems (BMS), heating, ventilation and air conditioning (HVAC), and supervision and regulation, we often encounter building owners who have yet to take advantage of the new artificial intelligence and machine learning tools that are now available. They only capture a small portion of useful building performance data using rudimentary Excel spreadsheets that only enable cursory data analysis. As a result, they leave behind millions of dollars in potential cost savings every year.
To fully leverage the data that their smart building assets and energy systems are generating every day, it becomes essential for building owners to centralize their building operational data, and to observe cause and effect on both energy consumption rates and building asset maintenance performance.
How artificial intelligence and machine learning tools deliver smart building benefits
Let’s look at both energy consumption and asset maintenance to determine how AI and machine learning tools can help.
Model energy consumption in smart buildings
AI tools now make it possible to precisely model energy consumption. By analyzing the gap between actual consumption rates and projected artificial intelligence model consumption, it is possible to detect underperforming areas. By identifying the sources of energy waste, the root causes of building efficiency gaps can then be rectified.
Some tools go even further by controlling the facility equipment, such as the heating and air conditioning system, based on occupancy schedules and weather forecasts. The building then consumes only the amount of energy necessary to achieve the necessary comfort for the occupants, at the right time and at the lowest cost.
These tools also allow for the comparison of the energy performance across your buildings and similar buildings managed by other people. This provides building owners with a deeper appreciation for what is possible in the realm of energy performance and energy savings.
A major issue, however, remains the quality of the data gathered. An algorithm cannot perform well if the input data is missing or incorrect. Our approach at Automatique & Industrie is to help our clients in the gathering and analysis of quality data as opposed to big data. We developed a patented tool that automatically detects and reconstructs missing or erroneous energy data. Other tools we use help to make AI methods and algorithms understandable to the human technical expert who validate that the AI data models accurately reflect the physical reality of the in-building systems.
Predictive asset maintenance in smart buildings
Using algorithms, building infrastructure systems can learn on their own through analysis of historical data and trending what the likely future behaviors of building assets will be. In this way, the cost of developing a system with the ability to predict when key components will break down and need replacement goes down. PIBCV smart actuators capture analytics and communicate any issues via alerts to be able to plan maintenance, ensure the health of the building and enhancing safety and comfort while reducing inconvenience.
Predictive maintenance drastically reduces the number of maintenance interventions required, and, when maintenance is performed, it can be scheduled at a time when the impact of a technical breakdown is minimized (like on weekends) and when the financial impact of a production stoppage is low. This is very different from traditional reactive and calendar-based maintenance methods where levels of disruption to operations are more drastic. Predictive maintenance aims to provide the decision support tools that enable maintenance to be performed only when needed, saving on unnecessary technician visits while minimizing unanticipated downtime.
In most cases, for machine learning and AI to have a major impact on both energy consumption and asset maintenance, it is necessary to analyze the various operating modes within a building. We recommend a minimum of one year of data history to be able to observe the building’s behavior across all four seasons. However, it is possible to leverage machine learning with only a few weeks of data. Although the application will not immediately achieve its full potential, it can immediately begin to add value through some initial cost savings and then scale up over the following months.
For more information
Automatique & Industrie supports energy management and predictive maintenance initiatives through the integration of smart building operation technologies (OT) systems (e.g., power, cooling, ventilation, building management systems) that allow for optimized facility operation and maintenance. We deploy predictive maintenance tools that analyze building infrastructure asset behavioral anomalies and that forecast when breakdowns are likely to occur.
We work in conjunction with experts from Schneider Electric and AVEVA who provide sophisticated tools such as PRISM for predictive maintenance and Resource Advisor for building energy management. To learn more about how smart building systems can increase your facility’s performance, visit us at the Automatique & Industrie or the Schneider Electric websites.