Artificial intelligence (AI) has the power not only to automate buildings but also to make them truly adaptive, smart, and agile, with the use of AI analytics helping to improve operations, reduce inefficiencies, and lower costs across building platforms in a wide range of industries, writes Anna Sliwon-Stewart, an analyst with IHS Markit.
While buildings have traditionally been perceived as stiff, fixed forms that merely provided a shell for its inhabitants and for the activities being conducted inside, structures outfitted with building management systems (BMS) are becoming more interactive and responsive to their occupants, according to a new study Artificial Intelligence in Smart Buildings – 2019, published by the research firm.
The study examines the potential presented by AI in smart buildings, as well as a roadmap showing both opportunities in the market and barriers to the wider adoption of AI-based solutions across a variety of end-user sectors.
Building owners are increasingly interested in what transpires inside buildings at any given time, utilising the knowledge and insights they obtain to improve building management, create better staffing plans, and reduce the operational costs of maintaining their business infrastructure in the hope of achieving higher profits.
In an ideal scenario, BMS platforms would run building systems based on conclusions drawn through the use of artificial intelligence, machine learning, and other complex statistical methods. The platform would then automatically adjust various subsystem settings without requiring involvement of the facilities manager, except for the approval of fundamental platform-level changes in specific circumstances.
While most AI solutions currently available are incapable of operating in such a model scenario, machine learning is increasingly being used across the buildings industry for more efficient operations.
Market revenue and segments
The market for AI analytics in smart buildings was worth $220.2 million in 2018, equivalent to approximately 9% of worldwide revenue for BMS platforms. But while there is wide acknowledgement that artificial intelligence can offer numerous benefits and efficiency gains for building systems, the adoption of BMS platforms with AI analytics will be relatively slow.
This protracted adoption can be attributed to a number of obstacles, including mistrust of AI, the significantly higher cost of deploying AI-powered platforms, and the limited functions of currently deployed machine-learning algorithms. Nonetheless, the use of AI in building management is a strong trend, and adoption of AI-powered BMS platforms is projected to increase, especially in new construction projects.
AI analytics and BMS platforms
AI analytics can help improve BMS platform functions in three ways—through analytics built into the platform, via analytics embedded in equipment, and from supplemental analytics software intended to augment specific building operations, such as energy management, comfort control, or predictive maintenance.
Of the three categories, analytics in equipment was the largest, accounting for nearly 50% of global AI analytics revenue in 2018. Examples of equipment covered by this category include AI-powered video surveillance cameras connected to BMS systems, as well as HVAC controllers with machine-learning capabilities. Incorporating AI into equipment is an easy way of adding advanced analytical functionality to the BMS, especially if the end-user is mistrustful of AI or of what it can do to enhance building operations.
Meanwhile, BMS platforms can be classified into three types: energy management (EM), security and safety management (SSM), and all building systems (ABS). The largest is the SSM platform, mainly because these systems are more likely to be integrated with AI-powered video surveillance cameras.
Energy management, however, is the building function most often enhanced by AI analytics, with results from an enhanced system most evident in the form of lower electricity bills.
Benefits, concerns, and challenges
AI analytics can be used in various industries to realize specific benefits.
For facilities in general, AI-enabled systems can aid in gathering data on day-to-day site elements such as the signing in or signing out of staff, floor occupancies, contractor traffic, and logistical operations. In retail, AI analytics can help improve temperature and humidity control in shopping centres, providing more comfortable conditions to shoppers. In education, end-users are adopting smart building technologies with AI capabilities to modernize facilities for students.
And in facilities that manufacture highly sensitive IT equipment components or medical components, AI and machine learning can help with climate control. If the system is integrated with fire safety, access control, and video surveillance, control of various building processes and operations could be achieved, preventing incidents that could disrupt production and negatively affect business. Such a comprehensive solution could be highly beneficial for sensitive installations like nuclear power plants, averting accidents that may have potentially devastating or lethal consequences.
Even so, concerns abound on the suitability of using AI analytics for critical infrastructure projects or in the daily operation of a building in which end-users process sensitive data. Proponents of AI analytics argue, however, that deploying the technology can go a long way toward enhancing situational awareness in managers responsible for the safety of facilities, thereby improving security overall.
A big challenge in creating AI-powered solutions that can work with all integration protocols and the various types of equipment used in smart buildings is the standardization of data gathered from every system. Machine learning algorithms and more advanced statistical algorithms can perform increasingly complex learning processes, but the data must be presented through a uniform standard. For the data to be ready for feeding to the smart algorithm, manufacturers may have to add an element to their software that pre-processes the data in real-time