Controlling HVAC and lighting systems

Buildings contribute to roughly 40% of the global energy use (approx. 64 PWh), of which a large portion is used for heating, cooling, ventilation, and air-conditioning (HVAC). The challenge is that current building management systems (BMS) are manually designed by human operators, which leads to energy inefficient operations. ML can help develop predictive models of building energy systems, leading to efficient building operation via advanced model-based optimal control methods. Independent studies reported that advanced HVAC control can significantly reduce energy use and mitigate greenhouse gas emissions with energy savings of 13% to 53%. Additional benefit is that advanced control strategies can provide grid flexibility services via price-responsiveness and active demand response capabilities.

Future Directions
Based on the reflections presented in  , there are several challenges for wide-spread application of advanced building controls:


 * 1) Availability of appropriate hardware and software infrastructure with compatible communication interfaces.
 * 2) Availability of data, and data-efficiency of the modeling and control design algorithms.
 * 3) User-friendly and computationally efficient building modeling and control design tools.
 * 4) Automated design, tuning, and deployment with plug-and-play implementation of advanced control strategy.
 * 5) Capability of handling multi-dimensional and hierarchical systems with dynamical coupling.
 * 6) Dealing with uncertainties in modeling and real-time measurements.
 * 7) Control security/safety/robustness by minimizing or eliminating the chance of generating control signals that might lead to undesirable outcomes during operation.
 * 8) Trained personnel to handle commissioning, and maintenance of advanced control methods in practice.
 * 9) Privacy and cyber-security issues and the user trust.
 * 10) Performance evaluation and algorithms benchmarking.