US 11,868,098 B2
Chiller and pump control using customizable artificial intelligence system
Jim Jingyue Gao, Silverdale, WA (US); Vedavyas Panneershelvam, Burnaby (CA); Katherine Elizabeth Hoffman, Silverdale, WA (US); Paritosh Mohan, London (GB); and Christopher R. Vause, Austin, TX (US)
Assigned to Phaidra, Inc., Seattle, WA (US)
Filed by Phaidra, Inc., Seattle, WA (US)
Filed on Nov. 12, 2021, as Appl. No. 17/525,694.
Prior Publication US 2023/0152756 A1, May 18, 2023
Int. Cl. G05B 13/02 (2006.01); G05B 13/00 (2006.01); G06N 5/022 (2023.01)
CPC G05B 13/028 (2013.01) [G05B 13/00 (2013.01); G05B 13/024 (2013.01); G06N 5/022 (2013.01)] 33 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
generating at least one query for one or more key performance indicator (KPI) associated with an industrial process;
displaying, at a user interface, the at least one query for the KPI associated with the industrial process;
receiving, at the user interface, a first user input comprising the KPI associated with the industrial process;
generating, based on the KPI, at least one query for information associated with the KPI;
displaying, at the user interface, the at least one query for information associated with the KPI;
receiving, at the user interface, a second user input comprising a response to the at least one query for information associated with the KPI;
generating a custom objective function based on the response;
generating processed datasets based on historical sensor data and real-time sensor data from equipment of the industrial process, wherein the historical sensor data is communicated from a historical database and the real-time sensor data is streamed from sensor components in real-time, and wherein the real-time sensor data comprises a set of chiller parameters;
iteratively training, based on the objective function, an artificial intelligence agent to predict a state of the industrial process based on the processed datasets comprising the set of chiller parameters, wherein such iterative training improves an accuracy of the artificial intelligence agent;
iteratively determining, by the artificial intelligence agent, a plan comprising a set of recommended actions for optimizing the KPI based on the predicted state of the industrial process, wherein the artificial intelligence agent comprises at least one of: a reinforcement learning agent, a neural network model, or a neural network model and search, and wherein the set of recommended actions comprises a set of chiller actions; and
controlling the equipment of the industrial process based on the plan, comprising selectively activating or deactivating a chiller based on the set of chiller actions.