Profiling transformer of power system
View Patent ↗Method and system for predicting an oil temperature of a transformer for a desired load and/or predicting a load that a transformer can support for a desired time. A machine learning algorithm is developed using historical data of a transformer. After the algorithm is developed, historical data corresponding to the transformer are input into the algorithm to develop a profile of the transformer describing how the temperature of oil within the transformer is expected to change as a function of a desired load. Using the profile, the of temperature of the transformer is predicted for a desired load. In this way, a prediction is made as to whether and/or for how long a transformer may support a desired load before the oil temperature reaches a specified threshold and/or before the transformer fails due to the load.
1. A method for a power system, comprising:
predicting an oil temperature of a transformer of the power system for a desired load based upon a profile of the transformer developed via a machine-learning algorithm; and
predicting a length of time that the transformer can support the desired load as a function of the predicted oil temperature.
2. The method of claim 1 , comprising:
acquiring heat run test data for the transformer, wherein the predicting an oil temperature comprises:
predicting the oil temperature by applying the heat run test data to the machine-learning algorithm.
3. The method of claim 1 , comprising:
responsive to the predicted oil temperature indicating that the transformer cannot support the desired load for a desired length of time, identifying a load that can be supported by the transformer for the desired length of time.
4. The method of claim 1 , comprising:
providing the machine-learning algorithm with historical data corresponding to the transformer to develop the profile of the transformer.
5. The method of claim 1 , comprising:
training the machine-learning algorithm utilizing historical data from the power system.
6. The method of claim 1 , wherein the machine-learning algorithm is a feed-forward artificial neural network algorithm.
7. The method of claim 6 , wherein the feed-forward artificial neural network algorithm comprises two hidden layers.
8. The method of claim 4 , wherein the historical data corresponding to the transformer is derived from at least one of:
one or more sensors operably coupled to the transformer; or
a heat run test performed on the transformer.
9. The method of claim 1 , wherein the predicting an oil temperature comprises:
predicting the oil temperature of the transformer for the desired load based upon present operating conditions of the transformer, the present operating conditions comprising at least one of:
a present load of the transformer;
a present ambient air temperature proximate the transformer; or
a present oil temperature of the transformer.
10. The method of claim 1 , wherein the profile describes how the transformer is expected to perform at various operating conditions.
11. A method for a power system, comprising:
identifying a failure of a first transformer of the power system, the first transformer supporting a first load;
identifying a second transformer of the power system that can support at least some of the first load for a desired length of time, comprising:
estimating a load that the second transformer can support for the desired length of time based upon a profile of the second transformer; and
transferring, to the second transformer, at least some of the first load, a total load of the second transformer not exceeding the estimated load.
12. The method of claim 11 , wherein the profile of the second transformer is developed via a machine learning algorithm.
13. The method of claim 12 , wherein at least one of:
the profile describes how the second transformer is expected to perform at various operating conditions; or
historical data corresponding to the second transformer is utilized as an input to the machine learning algorithm.
14. The method of claim 13 , wherein the historical data is derived from at least one of:
one or more sensors operably coupled to the second transformer; or
a heat run test performed on the second transformer.
15. The method of claim 12 , comprising:
training the machine-learning algorithm utilizing historical data from the power system.
16. The method of claim 11 , wherein the profile of the second transformer is developed via a feed-forward neural network algorithm.
17. The method of claim 11 , wherein the estimating is based upon present operating conditions of the second transformer, the present operating conditions comprising at least one of:
a present load of the second transformer;
a present ambient air temperature proximate the second transformer; or
a present oil temperature of the second transformer.
18. A system for a power system, comprising:
a profile generation component configured to develop a profile of a transformer of the power system using a neural network algorithm, the neural network algorithm receiving, as an input, historical data corresponding to the transformer; and
a prediction component configured to predict an oil temperature of the transformer at a desired load based upon the profile of the transformer and predict a length of time that the transformer can support the desired load as a function of the predicted oil temperature.
19. The system of claim 18 , wherein the prediction component is configured to predict the oil temperature of the transformer at the desired load based upon present operating conditions of the transformer.
20. The system of claim 18 , comprising a training component configured to train the neural network algorithm based upon historical data from the power system.