US-12619904-B2 - Apparatus and method for predicting transformer state in consideration of whether oil filtering is performed

US12619904B2US 12619904 B2US12619904 B2US 12619904B2US-12619904-B2

Abstract

A method of predicting a transformer state in consideration of whether oil filtering is performed includes receiving, by a transformer state prediction apparatus, transformer data of a transformer, determining, by the transformer state prediction apparatus, whether an oil of the transformer is filtered on the basis of the transformer data, and predicting, by the transformer state prediction apparatus, a state of the transformer on the basis of different prediction models depending on whether the oil is filtered.

Inventors

Jin Shi CUI
Jae Kyung Shin
Bo Seong Seo

Assignees

OnePredict Co., Ltd

Dates

Publication Date: 20260505
Application Date: 20201116
Priority Date: 20200720

Claims (12)

1 . A method of predicting a state of a power transformer, based on machine learning, in consideration of whether oil filtering of the power transformer is performed, the method comprising: generating prediction models including an oil filtering prediction model and an oil non-filtering prediction model, the oil filtering and oil non-filtering prediction models being long short term models, the oil filtering and oil non-filtering prediction models each including a plurality of sub-prediction models, the prediction models generated via: receiving, by a transformer data input receiver of a transformer state prediction apparatus, transformer training data of a transformer, the transformer training data including dissolved gas data of the transformer measured at n previous time points, n being greater than or equal to a threshold number of times for predicting the state of the transformer, separating the transformer training data into pieces of training data based on a determination, by a processor of the transformer state prediction apparatus, whether oil of the transformer has been filtered on the basis of the transformer training data, the pieces of training data including oil filtering training data and oil non-filtering training data, training the oil filtering prediction model with the oil filtering training data including dissolved gas data and transformer state data of a case in which oil filtering has been performed, the oil filtering prediction model trained to predict a transformer state of the transformer in the case in which oil filtering has been performed for the transformer, training the oil non-filtering prediction model with the oil non-filtering training data including dissolved gas data and transformer state data of a case in which oil filtering has not been performed, the oil non-filtering prediction model trained to predict a transformer state of the transformer in the case in which oil filtering has not been performed for the transformer, and training each sub-prediction model individually to distinguish transformer states and learn weights for distinguishing the transformer states, the transformer states including a normal state, a warning state, a critical state, and a fault state, wherein the oil filtering prediction model includes a first layer to which first dissolved gas data corresponding to year t−3 is input as an input value and first transformer state data corresponding up to year t−2 is input as an output value, a second layer to which second dissolved gas data corresponding to the year t−2 is input as an input value and second transformer state data corresponding up to year t−1 is input as an output value, a third layer to which third dissolved gas data corresponding to the year t−1 is input as an input value and the third transformer state data corresponding up to year t is input as an output value, and a fourth layer to which fourth dissolved gas data corresponding to the year t is input as an input value and fourth transformer state data corresponding up to year t+1 is input as an output value, and wherein the oil non-filtering prediction model includes a first layer to which first dissolved gas data corresponding to year t−3 is input as an input value and first transformer state data corresponding up to year t−2 is input as an output value, a second layer to which second dissolved gas data corresponding to the year t−2 is input as an input value and second transformer state data corresponding up to year t−1 is input as an output value, a third layer to which third dissolved gas data corresponding to the year t−1 is input as an input value and the third transformer state data corresponding up to year t is input as an output value, and a fourth layer to which fourth dissolved gas data corresponding to the year t is input as an input value and fourth transformer state data corresponding up to year t+1 is input as an output value; and generating a prediction, by the processor of the transformer state prediction apparatus, of the transformer state at an n+1 time point on the basis of the determination and the prediction models via: receiving, by the transformer data input receiver of the transformer state prediction apparatus, transformer data of the transformer comprising dissolved gas data of n previous time points, determining, by the processor of the transformer state prediction apparatus, an oil filtering determination of whether oil of the transformer has been filtered on the basis of the transformer data, determining a selected model of the prediction models to make the prediction based on the oil filtering determination by: selecting the oil filtering prediction model as the selected model responsive to the oil filtering determination indicating oil filtering has been performed for the transformer, and selecting the oil non-filtering prediction model as the selected model responsive to the oil filtering determination indicating oil filtering has not been performed for the transformer, and providing the dissolved gas data of n previous time points as input to the selected model to generate the prediction of the transformer state at the n+1 time point.
2 . The method of claim 1 , wherein whether the oil filtering has been performed is determined based on reduction ratios of oil-filtering determination gases.
3 . The method of claim 1 , wherein the transformer training data include information on six kinds of dissolved gases and component ratio information of the six kinds of dissolved gases, the six kinds of dissolved gases including hydrogen (H 2 ), methane (CH4), ethylene (C 2 H 4 ), ethane (C 2 H 6 ), acetylene (C,H,), and carbon monoxide (CO), the component ratio information including information on a value of a specific dissolved gas versus the sum of values of the six kinds of dissolved gases.
4 . The method of claim 2 , wherein the oil-filtering determination gases comprise methane (CH 4 ), ethylene (C 2 H 4 ) and ethane (C 2 H 6 ), and wherein it is determined that the oil filtering has been performed when the oil-filtering determination gases are simultaneously reduced by a threshold percentage or more at a subsequent measuring time point and wherein H 2 , CO and C 2 H 2 are not selected as the oil-filtering determination gas.
5 . The method of claim 1 , wherein a first sub-prediction model is a model for distinguishing the normal state from the warning state, the critical state and the fault state of the transformer, wherein a second sub-prediction model is a model for distinguishing the warning state from the normal state, the critical state and the fault state of the transformer, the third sub-prediction model is a model for distinguishing the critical state from the normal state, the warning state and the fault state of the transformer, and wherein the fourth sub-prediction model is a model for distinguishing the fault state from the normal state, the warning state and the critical state of the transformer.
6 . The method of claim 1 , wherein a first sub-prediction model of the oil filtering prediction model is a model for distinguishing the critical state from the normal state, the warning state and the fault state of the transformer and the first sub-prediction model is trained based on a first oil-filtering data set including dissolved gas data and transformer state data with which the transformer is determined to be in the critical state, and a second oil-filtering data set including dissolved gas data and transformer state data with which the transformer is determined to be in the normal state, the warning state or the fault state, the first oil-filtering data set and the second oil-filtering data set each including dissolved gas data and transformer state data of a case in which the oil filter has been performed.
7 . The method of claim 6 , wherein a second sub-prediction model of the oil filtering prediction model is a model for distinguishing the fault state from the normal state, the warning state and the critical state of the transformer based on a third oil-filtering data set labeled as the first information and a fourth oil-filtering data set labeled as the second information, wherein a third sub-prediction model of the oil filtering prediction model is a model for distinguishing the normal state from the warning state, the critical state and the fault state based on a fifth oil-filtering data set labeled with the first information and a sixth oil-filtering data set labeled with the second information, and wherein a fourth sub-prediction model of the oil filtering prediction model is a model for distinguishing the warning state from the normal state, the critical state and the fault state of the transformer based on a seventh oil-filtering data set labeled with the first information and an eight oil-filtering data set labeled with the second information.
8 . The method of claim 1 , wherein a first sub-prediction model of the oil non-filtering prediction model is a model for distinguishing the normal state from the warning state, critical state and the fault state of the transformer and the first sub-prediction model of the oil non-filtering prediction model is trained based on a first oil non-filtering data set and a second oil non-filtering data set as input in a case in which the oil filtering has not been performed.
9 . The method of claim 8 , wherein a second sub-prediction model of the oil non-filtering prediction model is a model for distinguishing the warning state from the normal state, the critical state and the fault state based on a third oil non-filtering data set labeled with the first information and a fourth oil non-filtering data set labeled with the second information, wherein a third sub-prediction model of the oil non-filtering prediction model is a model for distinguishing the critical state from the normal state, the warning state and the fault state based on a fifth oil non-filtering data set labeled with the first information and a sixth oil non-filtering data set labeled with the second information, and wherein a fourth sub-prediction model of the oil non-filtering prediction model is a model for distinguishing the fault state from the normal state, the warning state and the critical state based on a seventh oil non-filtering data set labeled with the first information and an eighth oil non-filtering data set labeled with the second information.
10 . The method of claim 1 , generating the prediction further comprising: determining whether the transformer data satisfy a condition for generating the prediction, the condition comprising at least one of n pieces of consecutive dissolved gas data or n−1 pieces of the transformer state data corresponding to the n pieces of the consecutive dissolved gas data in a time series.
11 . The method of claim 1 , wherein a sequence of the sub-prediction models of at least one of the oil filtering prediction model or the oil non-filtering prediction model are sequentially arranged to determine the prediction based on a prediction accuracy.
12 . The method of claim 11 , wherein the sequence of the sub-prediction models is adaptively changed based on the prediction accuracy.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS The present application claims priority to and the benefit from Korean Patent Application No. 10-2020-0089642, filed Jul. 20, 2020, and Korean Patent Application No. 10-2020-0121939, filed Sep. 22, 2020, each of which are hereby incorporated by reference in their entirety. BACKGROUND 1. Field of the Invention The present invention relates to a method of predicting a state of an oil-filled transformer in consideration of whether oil filtering is performed and an apparatus for performing the method. More particularly, the present invention relates to a method of determining whether oil filtering is performed and predicting a state of an oil-filled transformer on the basis of a trained prediction model and an apparatus for performing the method. 2. Discussion of Related Art With rapid industrial development, the demand for electric energy has drastically increased, leading to an increase in use of power transformers. Accordingly, many currently installed transformers are aged, and unpredictable equipment accidents frequently occur. Since the capacities of power transformers have been increased and power systems have been complicated, an accident caused by an equipment failure involves a widespread power outage, and an economic loss increases due to difficulties in power recovery and supply. To minimize such a loss, it is required to diagnose a current state of a transformer as accurately as possible. It is necessary to minimize unpredictable accidents of transformers by performing required management and maintenance. The largest share of cases in transformer accidents is related to the degradation of dielectric strength. The dielectric breakdown of a transformer may involve an explosion due to characteristics thereof. As the most effective method of analyzing insulation degradation characteristics, dissolved gas analysis (DGA) is frequently used. Organic insulating materials, such as insulating oil and insulating paper, used in transformers are increased in temperature due to operation and cause local overheats. Also, degraded products including various gases are generated through pyrolysis caused by an electric discharge and the like. Gases among the degraded products are dissolved in the insulating oil. For this reason, it is possible to estimate whether there is an abnormality in a transformer by regularly sampling the insulating oil of the transformer in operation and analyzing the concentrations of dissolved gases. However, when a transformer state is simply determined on the basis of the pattern of a specific gas, whether a specific gas exceeds a reference value, etc., it is difficult to make an accurate diagnosis so as to choose management, maintenance, or replacement of the transformer. Consequently, there is necessity for a method of not only diagnosing the cause of an abnormality in a transformer but also diagnosing a transformer state more clearly than existing methods. SUMMARY OF THE INVENTION The present invention is directed to solving all of the above-described problems. The present invention is also directed to accurately predicting a transformer state on the basis of different learning models depending on whether oil filtering is performed. The present invention is also directed to accurately predicting a transformer state of a transformer on the basis of learning models separately generated depending on states of a transformer. Representative configurations of the present invention for achieving the above objects are as follows. One aspect of the present invention provides a method of predicting a state of a transformer in consideration of whether oil filtering is performed. The method comprises receiving, by a transformer state prediction apparatus, transformer data of a transformer, determining, by the transformer state prediction apparatus, whether an oil of the transformer is filtered on the basis of the transformer data; and predicting, by the transformer state prediction apparatus, a state of the transformer on the basis of different prediction models depending on whether the oil is filtered. Also, the transformer data includes dissolved gas data of the transformer measured a threshold number of times or more. Also, the different prediction models include a first prediction model (oil filtering) and a second prediction model (oil non-filtering). One aspect of the present invention provides an apparatus for predicting a state of a transformer in consideration of whether oil filtering is performed. The apparatus comprises a transformer data input part configured to receive transformer data of a transformer; and a processor operatively connected to the transformer data input part, wherein the processor determines whether an oil of the transformer is filtered on the basis of the transformer data and predicts a state of the transformer on the basis of different prediction models depending on whether the oil is filtered. Also, the transformer data inclu