US 12,001,180 B2
Condition-based method for malfunction prediction
Pawel Stano, Cracow (PL); and Frank Kirschnick, Adliswil (CH)
Assigned to HITACHI ENERGY LTD, Zurich (CH)
Filed by HITACHI ENERGY LTD, Zürich (CH)
Filed on Jun. 8, 2021, as Appl. No. 17/342,133.
Claims priority of application No. 20178840 (EP), filed on Jun. 8, 2020.
Prior Publication US 2021/0382447 A1, Dec. 9, 2021
Int. Cl. G05B 17/02 (2006.01); G01N 15/06 (2006.01); G06N 7/01 (2023.01)
CPC G05B 17/02 (2013.01) [G01N 15/0618 (2013.01); G06N 7/01 (2023.01)] 21 Claims
OG exemplary drawing
 
1. A method, the method being performed by at least one integrated circuit and comprising:
determining a future evolution of an asset health state of an asset, the asset being a power transformer, a distributed energy resource, or a power generator, wherein determining the future evolution of the asset health state of the asset comprises iteratively repeating:
performing a stochastic simulation to obtain a prognosis for the future evolution of the asset health state, wherein the stochastic simulation uses a model having a discrete state space;
after performing the stochastic simulation to obtain the prognosis for the future evolution of the asset health state, reading sensor measurements from one or more sensors; and
updating the prognosis based on the read sensor measurements using a particle filter; and
generating output based on the determined future evolution of the asset health state; and
automatically performing an action, wherein the action comprises at least one of scheduling a down-time of the asset based on the determined future evolution of the asset health state, scheduling maintenance work based on the determined future evolution of the asset health state, scheduling replacement work based on the determined future evolution of the asset health state, or changing maintenance intervals based on the determined future evolution of the asset health state,
wherein the stochastic simulation uses a model having a discrete state space, wherein updating the prognosis comprises:
computing, based on the prognosis, an expectation value for an observable quantity included in the sensor measurements;
performing a comparison of the expectation value to the observable quantity included in the sensor measurements; and
computing an updated initial state for a subsequent iteration of the stochastic simulation based on the comparison.