System and method for proactive handling of multiple faults and failure modes in an electrical network of energy assets

Patent metadata

Field	Value
Publication number	US-12585265-B2
Application number	US-202418659645-A
Country	US
Kind code	B2
Filing date	May 9, 2024
Priority date	Dec 27, 2018
Publication date	Mar 24, 2026
Grant date	Mar 24, 2026

Abstract

An example method comprises receiving historical sensor data of a renewable energy asset for a first time period, identifying historical log data in one or more log sources, retrieving dates of the identified historical log data, retrieving sequences of historical sensor data using the dates, training hidden Markov models using the sequences of historical sensor data to identify probability of shifting states of one or more components of the renewable energy asset, receiving current sensor data of a second time period, identifying current log data in the one or more log sources, retrieving dates of the identified current log data, retrieving sequences of current sensor data using the dates, applying the hidden Markov models to the sequences of the current sensor data to assess likelihood of the one or more faults, creating a prediction of a future fault, and generating a report including the prediction of the future fault.

First claim

The invention claimed is:

1 . A non-transitory computer-readable medium comprising executable instructions, the executable instructions being executable by one or more processors to perform a method, the method comprising:
receiving sensor data of a first time period, the sensor data from one or more first sensors of a renewable energy asset; identifying log data in one or more log sources, the log data indicating one or more faults in the renewable energy asset, the one or more log sources storing log data of the renewable energy asset; retrieving dates of the identified log data that indicates the one or more faults; retrieving sequences of sensor data using the dates of the identified log data; applying one or more trained Markov models to the sequences of sensor data to assesses one or more likelihoods of the one or more faults and create a prediction of a future fault in the renewable energy asset, the one or more trained Markov models being previously trained on sensor data of a second time period, the sensor data of the second time period from one or more second sensors of the renewable energy assets; generating a report including the prediction of the future fault in the renewable energy asset; and controlling the renewable energy asset based on the prediction of the future fault in the renewable energy asset.

2 . The non-transitory computer-readable medium of claim 1 , wherein applying the one or more trained Markov models to the sequences of the sensor data to assess the one or more likelihoods of the one or more first faults and create the prediction of the future fault in the renewable energy asset includes comparing one or more maximum log likelihoods of one or more different failure states from one or more different components of the renewable energy asset.

3 . The non-transitory computer-readable medium of claim 1 , wherein applying the one or more trained Markov models to the sequences of the sensor data to assess the one or more likelihoods of the one or more faults and create the prediction of the future fault in the renewable energy asset includes determining probability of one or more different faults of two or more different components of the renewable energy asset using an iterative Expectation-Maximization (EM) algorithm.

4 . The non-transitory computer-readable medium of claim 1 , the method further comprising:
comparing the future fault against one or more criteria to determine a significance of the future fault; generating an alert based on the significance, the alert identifying the future fault; and providing the alert.

5 . The non-transitory computer-readable medium of claim 1 , the method further comprising:
receiving the sensor data of the second time period; identifying an other log data in the one or more log sources, the other log data indicating other faults in the renewable energy asset; retrieving dates of the other log data that indicates the other faults; retrieving sequences of the sensor data of the second time period using the dates of the other log data; and training one or more Markov models using the sequences of the sensor data of the second time period to obtain the one or more trained Markov models.

6 . The non-transitory computer-readable medium of claim 5 , the method further comprising extracting features from the sequences of the sensor data of the second time period, and wherein training the one or more Markov models using the sequences of the sensor data of the second time period to obtain the one or more trained Markov models includes training the one or more Markov models using the features extracted from the sequences of the sensor data of the second time period to obtain the one or more trained Markov models.

7 . The non-transitory computer-readable medium of claim 6 , wherein the extracting the features from the sequences of the sensor data of the second time period includes dimension reduction of the sequences of the sensor data of the second time period.

8 . The non-transitory computer-readable medium of claim 1 , the method further comprising extracting features from the sequences of the sensor data of the first time period, and wherein applying the one or more trained Markov models to the sequences of the sensor data to assess the one or more likelihoods of the one or more faults and create the prediction of the future fault in the renewable energy asset includes applying the one or more trained Markov models to the features extracted from the sequences of the sensor data to assess the one or more likelihoods of the one or more first faults and create the prediction of the future fault in the renewable energy asset.

9 . The non-transitory computer-readable medium of claim 1 , wherein the one or more first sensors are a subset of the one or more second sensors.

10 . The non-transitory computer-readable medium of claim 1 , wherein the first time period is a current time period and the second time period is a historical time period.

11 . A system comprising:
at least one processor; and memory containing executable instructions, the executable instructions being executable by the at least one processor to:
receive sensor data of a first time period, the sensor data from one or more first sensors of a renewable energy asset;
identify log data in one or more log sources, the log data indicating one or more faults in the renewable energy asset, the one or more log sources storing log data of the renewable energy asset;
retrieve dates of the identified log data that indicates the one or more faults;
retrieve sequences of sensor data using the dates of the identified log data;
apply one or more trained Markov models to the sequences of sensor data to assess one or more likelihoods of the one or more faults and create a prediction of a future fault in the renewable energy asset, the one or more trained Markov models being previously trained on sensor data of a second time period, the sensor data of the second time period from one or more second sensors of the renewable energy assets;
generate a report including the prediction of the future fault in the renewable energy asset; and
control the renewable energy asset based on the prediction of the future fault in the renewable energy asset.

12 . The system of claim 11 , wherein applying the one or more trained Markov models to the sequences of the sensor data to assess the one or more likelihoods of the one or more first faults and create the prediction of the future fault in the renewable energy asset includes comparing one or more maximum log likelihoods of one or more different failure states from one or more different components of the renewable energy asset.

13 . The system of claim 11 , wherein applying the one or more trained Markov models to the sequences of the sensor data to assess the one or more likelihoods of the one or more faults and create the prediction of the future fault in the renewable energy asset includes determining probability of one or more different faults of two or more different components of the renewable energy asset using an iterative Expectation-Maximization (EM) algorithm.

14 . The system of claim 11 , wherein the executable instructions being further executable by the at least one processor to:
compare the future fault against one or more criteria to determine a significance of the future fault; generate an alert based on the significance, the alert identifying the future fault; and provide the alert.

15 . The system of claim 11 , wherein the executable instructions being further executable by the at least one processor to:

Assignees

• Utopus Insights Inc

Inventors

• Srinivasan Guruprasad
• Kim Younghun

Classifications

• H02J2103/30

  Simulating, planning, modelling, reliability check or computer assisted design [CAD] of electric power networks · CPC title

• H02J2101/20

  Dispersed power generation using renewable energy sources · CPC title

• H02J3/381

  Dispersed generators · CPC title

• H02J3/001

  Arrangements for handling faults or abnormalities, e.g. emergencies or contingencies · CPC title

• H02J3/00

  Circuit arrangements for AC mains or AC distribution networks · CPC title