US 12,282,305 B2
Building management system with generative AI-based predictive maintenance
Julie J. Brown, Yardley, PA (US); Young M. Lee, Old Westbury, NY (US); Rajiv Ramanasankaran, San Jose, CA (US); Sastry K M Malladi, Fremont, CA (US); Michael Tenbrock, Dachsen (CH); Levent Tinaz, Tampa Bay, FL (US); Samuel A. Girard, Kenosha, WI (US); David S. Elario, Hartland, WI (US); Juliet A Pagliaro Herman, Waukesha, WI (US); Miguel Galvez, Westford, MA (US); Trent M. Swanson, Wellington, FL (US); John F. Kuchler, Muskego, WI (US); Deepak Budhiraja, Ashburn, VA (US); Daniela M. Natali, Kensington, MD (US); Josip Lazarevski, Zurich (CH); Scott Deering, Milwaukee, WI (US); Gary W. Gavin, Franklin, WI (US); Kristen Sheppard-Guzelaydin, West Chester, PA (US); James Young, Cork (IE); Prashanthi Sudhakar, San Francisco, CA (US); Kaleb Luedtke, West Bend, WI (US); Karl F. Reichenberger, Mequon, WI (US); Wenwen Zhao, Santa Clara, CA (US); Adam R. Grabowski, Brookfield, WI (US); Lauren C. Dern, Fox Point, WI (US); Nicole A. Madison, Milwaukee, WI (US); Dana S. Petersen, Milwaukee, WI (US); Nevin L. Forry, York, PA (US); Pedriant Pena, Groveland, MA (US); Ghassan R. Hamoudeh, San Marcos, CA (US); and Ryan G Danielson, Castle Rock, CO (US)
Assigned to TYCO FIRE & SECURITY GMBH, Neuhausen am Rheinfall (CH)
Filed by Tyco Fire & Security GmbH, Neuhausen am Rheinfall (CH)
Filed on Jan. 22, 2024, as Appl. No. 18/419,456.
Claims priority of provisional application 63/458,871, filed on Apr. 12, 2023.
Claims priority of provisional application 63/470,123, filed on May 31, 2023.
Prior Publication US 2024/0345554 A1, Oct. 17, 2024
Int. Cl. G05B 15/02 (2006.01)
CPC G05B 15/02 (2013.01) 27 Claims
OG exemplary drawing
 
1. A method comprising:
training, by one or more processors, a generative AI model using first operating data from building equipment and natural language of a plurality of first service reports indicating a plurality of first problems associated with the building equipment, the generative AI model comprising a transformer-based neural network;
predicting, by the one or more processors using the trained generative AI model, that a control strategy that controls the building equipment is likely to cause one or more future problems to occur with the building equipment based on second operating data from the building equipment; and
automatically initiating, by the one or more processors, one or more actions to prevent the one or more future problems from occurring or mitigate an effect of the one or more future problems, the one or more actions comprising adjusting the control strategy.