US 12,492,837 B2
Process for adaptable health, degradation and anomaly detection of systems using benchmarks
Vaclav Slimacek, Valasske Mezirici (CZ); and Jiri Rojicek, Badeniho (CZ)
Assigned to HONEYWELL INTERNATIONAL INC., Charlotte, NC (US)
Filed by Honeywell International Inc., Charlotte, NC (US)
Filed on Dec. 30, 2021, as Appl. No. 17/565,775.
Application 17/565,775 is a continuation of application No. 16/432,479, filed on Jun. 5, 2019, granted, now 11,248,819.
Prior Publication US 2022/0120463 A1, Apr. 21, 2022
Int. Cl. F24F 11/49 (2018.01); F24F 11/38 (2018.01); F24F 11/52 (2018.01); G06N 20/00 (2019.01)
CPC F24F 11/49 (2018.01) [F24F 11/38 (2018.01); F24F 11/52 (2018.01); G06N 20/00 (2019.01)] 15 Claims
OG exemplary drawing
 
1. A method for monitoring the performance of a system, the method comprising:
defining a benchmark condition for a system, the benchmark condition corresponding to a first set of operating conditions for the system;
operating the system under the first set of operating conditions that correspond to the benchmark condition;
collecting data while operating the system under the first set of operating conditions that correspond to the benchmark condition;
training a model using the collected data, the model is trained to produce a measure of performance of the system operating under the benchmark condition;
applying the trained model to data collected while operating the system under the first set of operating conditions that correspond to the benchmark condition to obtain a current measure of performance of the system under the benchmark condition;
operating the system under another set of operating conditions that do not correspond to the benchmark condition;
returning to operate the system under the first set of operating conditions that correspond to the benchmark condition;
collecting new data while operating the system under the first set of operating conditions that correspond to the benchmark condition;
repeating the steps of operating the system under the first set of operating conditions that correspond to the benchmark condition, collecting new data while operating the system under the first set of operating conditions that correspond to the benchmark condition, applying the trained model to the new data collected while operating the system under the first set of operating conditions that correspond to the benchmark condition, operating the system under another set of operating conditions that do not correspond to the benchmark condition, and returning to operate the system under the first set of operating conditions that correspond to the benchmark condition; and
evaluating an evolution of the current measure of performance of the system under the benchmark condition.