Condition-Based Method for Malfunction Prediction

Abstract

To perform a prognostic health analysis for an asset, a plurality of independent stochastic simulations are performed using transition probabilities of a discrete Markov Chain model. A prognostic asset health state evolution is computed over a time horizon from the plurality of independent stochastic simulations. An output is generated based on the computed prognostic asset health state evolution.

1 . A method of performing a prognostic health analysis for an asset, the method comprising:
 performing a plurality of independent stochastic simulations using transition probabilities of a discrete Markov Chain model, wherein the discrete Markov Chain model has a state space that comprises a set of asset health states and wherein each of the plurality of independent stochastic simulations simulates a future evolution in the state space of the discrete Markov Chain model over a prognostic horizon;   computing a prognostic asset health state evolution over the prognostic horizon from the plurality of independent stochastic simulations, wherein computing the prognostic asset health state evolution comprises computing a time-dependent scalar function based on probabilities that the discrete Markov Chain model is in a particular state at a particular time as determined from each of the plurality of independent stochastic simulations;   generating output based on the computed prognostic asset health state evolution; and   automatically performing an action relating to the asset based on the computed prognostic asset health state evolution.   
     
     
         2 . The method of  claim 1 , wherein the asset is a power system asset or an industrial asset. 
     
     
         3 . The method of  claim 1 , wherein generating the output comprises generating an output related to a remaining useful life (RUL) or a probability of failure (PoF). 
     
     
         4 . The method of  claim 1 , wherein computing the prognostic asset health state evolution comprises computing a remaining useful life. 
     
     
         5 . The method of  claim 1 , further comprising computing confidence or variance information for the prognostic asset health state evolution as a function of time over the prognostic horizon from the plurality of independent stochastic simulations, wherein the output is further generated based on the confidence or variance information. 
     
     
         6 . The method of  claim 5 , wherein the output is further generated based on the confidence information and wherein the confidence information comprises a future evolution of a confidence interval over the prognostic horizon. 
     
     
         7 . The method of  claim 5 , wherein the output is further generated based on the variance information and wherein the variance information comprises a future evolution of a variance over the prognostic horizon. 
     
     
         8 . The method of  claim 5 , wherein the confidence or variance information comprises a time evolution of a lower boundary and a time evolution of an upper boundary, the lower boundary being associated with a first set of transition probabilities and the upper boundary being associated with a second set of transition probabilities different from the first set of transition probabilities. 
     
     
         9 . The method of  claim 1 , wherein the state space comprises:
 at least one state in which operation of the asset is not adversely affected by a failure;   at least one state in which operation of the asset is adversely affected by a failure, but the asset continues to operate; and   a state in which the asset is inoperative due to a failure.   
     
     
         10 . The method of  claim 1 , wherein computing the prognostic asset health state evolution comprises computing, for a plurality of times within the prognostic horizon, a probability distribution in the state space and mapping the probability distribution to a scalar. 
     
     
         11 . The method of  claim 10 , wherein the prognostic asset health state evolution is obtained as a time evolution of the scalar. 
     
     
         12 . The method of  claim 1 , further comprising determining the transition probabilities from historical data comprising sensor data for a plurality of assets. 
     
     
         13 . The method of  claim 12 , wherein determining the transition probabilities comprises:
 computing the time-dependent scalar function from sensor data for the plurality of assets,   identifying transitions within the state space of the discrete Markov Chain model based on the time-dependent scalar function, and   computing the transition probabilities based on the transitions within the state space of the discrete Markov Chain model.   
     
     
         14 . The method of  claim 1 , wherein the plurality of independent stochastic simulations are Markov Chain Monte Carlo simulations. 
     
     
         15 . The method of  claim 1 , further comprising:
 receiving sensor measurement data captured during operation of the asset; and   updating the prognostic asset health state evolution based on the received sensor measurement data.   
     
     
         16 . The method of  claim 1 , wherein the plurality of simulations comprise simulations for different ambient or operating scenarios. 
     
     
         17 . The method of  claim 1 , wherein:
 the asset is a power transformer, a distributed energy resource, DER, unit, or a power generator; or   the prognostic horizon is 1 year or more.   
     
     
         18 . A method of performing a prognostic health analysis for an asset, the method comprising:
 performing a plurality of independent stochastic simulations using transition probabilities of a discrete Markov Chain model, wherein the discrete Markov Chain model has a state space that comprises a set of asset health states and wherein each of the plurality of independent stochastic simulations simulates a future evolution in the state space of the discrete Markov Chain model over a prognostic horizon;   computing a prognostic asset health state evolution over the prognostic horizon from the plurality of independent stochastic simulations, wherein computing the prognostic asset health state evolution comprises computing a function based on probabilities that the discrete Markov Chain model are in particular states during the prognostic horizon as determined from each of the plurality of independent stochastic simulations;   generating output based on the computed prognostic asset health state evolution; and   automatically performing an action relating to the asset based on the computed prognostic asset health state evolution.   
     
     
         19 . The method of  claim 18 , wherein the prognostic horizon is at least one year. 
     
     
         20 . A non-transitory computer readable medium with instructions stored thereon, wherein, when executed by a processor, the instructions enable the processor to:
 perform a plurality of independent stochastic simulations using transition probabilities of a discrete Markov Chain model, wherein the discrete Markov Chain model has a state space that comprises a set of asset health states and wherein each of the plurality of independent stochastic simulations simulates a future evolution in the state space of the discrete Markov Chain model over a prognostic horizon;   compute a prognostic asset health state evolution over the prognostic horizon from the plurality of independent stochastic simulations, wherein computing the prognostic asset health state evolution comprises computing a function based on probabilities that the discrete Markov Chain model are in particular states during the prognostic horizon as determined from each of the plurality of independent stochastic simulations;   generate output based on the computed prognostic asset health state evolution; and   automatically perform an action relating to the asset based on the computed prognostic asset health state evolution.