Apparatus and method for analyzing spectrum of dissolved gas signal of extra-high voltage oil-immersed transformer

Abstract

An apparatus for analyzing a spectrum of a dissolved gas signal of an extra-high voltage oil-immersed transformer according to one embodiment of the present invention may comprise a processor which obtains a dissolved gas signal of an extra-high voltage oil-immersed transformer, uses an analysis model for estimating disturbance components of dissolved gas signals sequentially inputted over time so as to predict a dissolved gas signal to be inputted in a next sequence, updates the analysis model to minimize the difference between an actual dissolved gas signal obtained in the next sequence and the predicted dissolved gas signal, and removes the disturbance components obtained through the analysis model from the actual dissolved gas signal.

Description

Spectrum analysis device and spectrum analysis method of oil-gas signal of ultra-high pressure oil-immersed transformer

[0001] The present invention relates to a spectrum analysis device and a spectrum analysis method for a gas signal in an ultra-high pressure inlet transformer.

[0002] The biggest problem with online diagnostic techniques for diagnosing equipment deterioration in ultra-high-voltage oil-immersed transformers is their vulnerability to external disturbances, which are various noise signals that exist in actual fields.

[0003] Therefore, considering the overall sensor noise (Sensor Background Noise) and the time-varying spectrum characteristics of temporary oil gas signals due to various abnormal voltages, spectrum analysis performance can be seen as a factor that can improve the performance of oil gas diagnostic methods.

[0004] Current mainstream methods, such as the Duval Triangle Method, Duval's Pentagon Method, and the IEC Key Gas Method, feature UI trend analysis capabilities that don't require separate software algorithms. However, these techniques directly map measured online gas-in-oil data to trend analysis UIs presented in international standards or papers without any special preprocessing, which still contain disturbance components and fail to reflect the time-varying spectral characteristics of gas-in-oil signals.

[0005] Considering the recent trend of increasing the complexity of adaptive correction technology for various power facilities based on digital protective relaying and the gradual expansion of distributed power generation in the grid, the existing preprocessing-free gas-in-oil analysis technique has many limitations in terms of efficiency and reliability.

[0006] The purpose of the present invention is to provide a spectrum analysis device and spectrum analysis method for a gas signal in an ultra-high pressure oil-immersed transformer, which can not only improve the analysis performance of a gas signal in an oil-immersed transformer, but also optimize the operating environment of an ultra-high pressure facility.

[0007] In a spectrum analysis device of an oil-gas signal of an ultra-high pressure oil-gas transformer according to one embodiment of the present invention, a processor may be included that obtains an oil-gas signal of an ultra-high pressure oil-gas transformer, predicts an oil-gas signal to be input in a next sequence using an analysis model that estimates a disturbance component of an oil-gas signal sequentially input over time, updates the analysis model so that a difference between an actual oil-gas signal acquired in the next sequence and the predicted oil-gas signal is minimized, and removes a disturbance component acquired through the analysis model from the actual oil-gas signal.

[0008] The above processor can update the analysis model by correcting the filter coefficients of each of the plurality of filters included in the analysis model.

[0009] The above processor can analyze the state of the ultra-high pressure inlet transformer from a signal from which the external disturbance component has been removed from the actual oil gas signal.

[0010] The processor can track the state change of the signal from which the disturbance component has been removed based on the state history of the ultra-high voltage input transformer.

[0011] The above processor can display the state change of the actual dissolved gas signal on a standard dissolved gas analysis (DGA) graph.

[0012] The above processor can control auxiliary devices connected to the ultra-high voltage input transformer based on the result of tracking the state change of the signal from which the external component has been removed.

[0013] The above analysis model can be trained to estimate the disturbance component of the input oil gas signal using a bandwidth determined for each noise source in the frequency domain.

[0014] A method for spectrum analysis of a gas signal in an oil-immersed transformer according to an embodiment of the present invention, performed by a spectrum analysis device, may include: a step of acquiring a gas signal in an oil-immersed transformer; a step of predicting a gas signal to be input in a next sequence using an analysis model that estimates a disturbance component of a gas signal in an oil-immersed transformer sequentially input over time; a step of updating the analysis model so that a difference between an actual gas signal in an oil-immersed transformer acquired in a next sequence and a predicted gas signal is minimized; and a step of removing a disturbance component acquired through the analysis model from the actual gas signal in an oil-immersed transformer.

[0015] The step of updating the above analysis model may include a step of updating the analysis model by correcting the filter coefficients of each of the plurality of filters included in the analysis model.

[0016] The above spectrum analysis method may further include a step of analyzing the state of the ultra-high pressure inlet transformer from a signal from which the external disturbance component has been removed from the actual oil-gas signal.

[0017] The step of analyzing the state of the ultra-high voltage inlet transformer may include a step of tracking a state change of a signal from which the disturbance component has been removed based on a history of the state of the ultra-high voltage inlet transformer.

[0018] The step of analyzing the state of the above ultra-high pressure inlet transformer may include a step of displaying the state change of the actual dissolved gas signal on a standard dissolved gas analysis (DGA) graph.

[0019] The above spectrum analysis method may further include a step of controlling auxiliary devices connected to the ultra-high voltage input transformer based on the result of tracking the state change of the signal from which the external disturbance component has been removed.

[0020] According to one embodiment of the present invention, the time-varying characteristics of disturbance elements scattered in the operating field of an ultra-high pressure inlet transformer can be automatically detected, thereby effectively removing disturbance components of a gas signal in the frequency domain.

[0021] According to one embodiment of the present invention, it is possible to provide reliability of diagnostic data that can prevent various system issues due to abnormal voltage inflow into aging equipment in advance.

[0022] According to one embodiment of the present invention, performance degradation such as deterioration can be prevented and the life of the equipment can be extended through appropriate load distribution of an ultra-high voltage input transformer.

[0023] FIG. 1 is a schematic diagram illustrating a spectrum analysis system according to one embodiment of the present invention.

[0024] FIG. 2 is a block diagram illustrating the configuration of a spectrum analysis device according to one embodiment of the present invention.

[0025] FIG. 3 is a diagram illustrating an operation flow chart of a spectrum analysis device according to one embodiment of the present invention.

[0026] FIG. 4 is a drawing illustrating the removal of a disturbance component of a gas signal according to one embodiment of the present invention.

[0027] FIG. 5 is a diagram illustrating an operation flow chart of a spectrum analysis device according to a first embodiment of the present invention.

[0028] FIG. 6 is a diagram illustrating an operation flow chart of a spectrum analysis device according to a second embodiment of one embodiment of the present invention.

[0029] Figure 7 is a diagram illustrating an example of a Duval triangle graph.

[0030] - Explanation of symbols -

[0031] 1: Spectrum Analysis System

[0032] 10: Ultra-high voltage oil-immersed transformer

[0033] 100: Spectrum Analysis Device

[0034] 110: Input section

[0035] 120: Communications Department

[0036] 130: Display

[0037] 140: Storage

[0038] 150: Processor

[0039] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. The detailed description set forth below, together with the accompanying drawings, is intended to explain exemplary embodiments of the present invention and is not intended to represent the only embodiments in which the present invention may be practiced. In the drawings, portions irrelevant to the description may be omitted for clarity in describing the present invention, and the same reference numerals may be used throughout the specification for identical or similar components.

[0040] FIG. 1 is a schematic diagram illustrating a spectrum analysis system according to one embodiment of the present invention.

[0041] A spectrum analysis system (1) (hereinafter referred to as system (1)) according to one embodiment of the present invention may include an ultra-high voltage input transformer (10) and a spectrum analysis device (100).

[0042] An ultra-high voltage input transformer (10) is a power device that is installed in a power system of an ultra-high voltage power grid, such as a power plant or substation, to step down voltage, and uses insulating oil as an insulating medium for electrical insulation.

[0043] At this time, the equipment deterioration status of the ultra-high pressure oil-immersed transformer (10) can be checked using the oil-immersed gas signal generated from the insulating oil, and a sensor can be installed in the ultra-high pressure oil-immersed transformer (10) to measure the oil-immersed gas signal.

[0044] The spectrum analysis device (100) is an electronic device that analyzes the oil gas signal obtained from the ultra-high pressure inlet transformer (10) to check the equipment deterioration status, and can be implemented as a computer, server, smart phone, tablet PC, smart pad, laptop, etc.

[0045] As described above, the oil-gas signal can be measured together with noise, i.e., disturbance components, introduced into the sensor by various factors, and it is important to remove these disturbance components from the oil-gas signal.

[0046] The external disturbance component is a general concept of the external environment of the transformer, and factors that particularly affect the oil gas signal include season, weather, temperature, atmospheric pressure, and load factor.

[0047] The present invention proposes a method for accurately removing the time-varying characteristics of a long-term acquired oil gas signal by using an adaptive control algorithm, and various utilization methods using the oil gas signal from which disturbance components have been removed.

[0048] Hereinafter, the configuration and operation of a spectrum analysis device (100) according to one embodiment of the present invention will be specifically described with reference to the drawings.

[0049] FIG. 2 is a block diagram illustrating the configuration of a spectrum analysis device according to one embodiment of the present invention.

[0050] A spectrum analysis device (100) according to one embodiment of the present invention may include an input unit (110), a communication unit (120), a display unit (130), a storage unit (140), and a processor (150).

[0051] The input unit (110) generates input data in response to a user input of the spectrum analysis device (100). For example, the user input may be a user input that initiates the operation of the spectrum analysis device (100), a user input that updates an analysis model that estimates a disturbance component of a gas-in-oil signal, a user input that redistributes the load applied to an ultra-high voltage inlet transformer (10) according to the analysis, etc. In addition, if it is a user input necessary for removing and utilizing a disturbance component from a gas-in-oil signal, it may be applied without limitation.

[0052] The input unit (110) includes at least one input means. The input unit (110) may include a keyboard, a key pad, a dome switch, a touch panel, a touch key, a mouse, a menu button, etc.

[0053] The communication unit (120) can perform communication with external devices such as an ultra-high voltage inlet transformer (10), a sensor, and a server to transmit and receive measured dissolved gas signals, predicted dissolved gas signals, analysis models, disturbance components, standard dissolved gas analysis (DGA) graphs, bandwidths determined by noise sources, etc.

[0054] To this end, the communication unit (120) can perform wireless communication such as 5G (5th generation communication), LTE-A (Long Term Evolution-Advanced), LTE (Long Term Evolution), Wi-Fi (Wireless Fidelity), Bluetooth, or wired communication such as LAN (Local Area Network), WAN (Wide Area Network), and power line communication.

[0055] The display unit (130) displays display data according to the operation of the spectrum analysis device (100). The display unit (130) can display a screen that displays a trend of a gas-in-oil signal, a screen that displays a standard gas-in-oil analysis graph, a screen that displays a change in the difference between an actual gas-in-oil signal obtained in the next sequence and a predicted gas-in-oil signal, a screen that receives user input, etc.

[0056] The display unit (130) includes a liquid crystal display (LCD), a light emitting diode (LED) display, an organic light emitting diode (OLED) display, a micro electro mechanical systems (MEMS) display, and an electronic paper display. The display unit (130) may be implemented as a touch screen by being combined with the input unit (110).

[0057] The storage unit (140) stores the operation programs of the spectrum analysis device (100). The storage unit (140) includes a non-volatile storage that can preserve data (information) regardless of whether power is supplied, and a volatile memory that loads data to be processed by the processor (150) and cannot preserve data if power is not supplied. The storage includes a flash memory, a hard-disc drive (HDD), a solid-state drive (SSD), a read-only memory (ROM), etc., and the memory includes a buffer, a random access memory (RAM), etc.

[0058] The storage unit (140) can store measured oil gas signals, predicted oil gas signals, analysis models, disturbance components, standard oil gas analysis graphs, bandwidths determined by noise sources, etc., and can store computational programs required in the process of predicting candidate oil gas signals, updating analysis models, removing disturbance signals, analyzing the status of ultra-high voltage inlet transformers (10), tracking status changes in oil gas signals, etc.

[0059] The processor (150) can control at least one other component (e.g., hardware or software component) of the spectrum analysis device (100) by executing software such as a program, and can perform various data processing or operations.

[0060] A processor (150) according to one embodiment of the present invention may obtain a submerged gas signal of an ultra-high pressure inlet transformer, predict a submerged gas signal to be input in a next sequence using an analysis model that estimates a disturbance component of a submerged gas signal sequentially input over time, update the analysis model so that a difference between an actual submerged gas signal obtained in the next sequence and a predicted submerged gas signal is minimized, and remove a disturbance signal obtained through the analysis model from the actual submerged gas signal.

[0061] At this time, the processor (150) may build an analysis model and an analysis model update algorithm, or may receive and store an existing analysis model and an analysis model update algorithm from the outside and use them, but is not limited to either one.

[0062] Meanwhile, the processor (150) may perform at least a portion of the data analysis, processing, and result information generation for performing the above operations using at least one of a machine learning, neural network, or deep learning algorithm as a rule-based or artificial intelligence (AI) algorithm. Examples of the neural network may include models such as a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), and a transformer.

[0063] FIG. 3 is a diagram illustrating an operation flow chart of a spectrum analysis device according to one embodiment of the present invention.

[0064] A processor (150) according to one embodiment of the present invention can obtain a gas signal of an ultra-high pressure inlet transformer (10) (S10).

[0065] The processor (150) can obtain a gas signal from a sensor installed in an ultra-high pressure inlet transformer (10), and the processor (150) can obtain gas data from the sensor and convert it into a digital signal. However, the method or path for obtaining the gas signal is not limited to any one.

[0066] A processor (150) according to one embodiment of the present invention can predict an oil gas signal to be input in the next sequence by using an analysis model that estimates a disturbance component of an oil gas signal sequentially input over time (S20).

[0067] The analysis model is a model trained to estimate the disturbance component of an input oil-gas signal using a bandwidth determined by each noise source in the frequency domain. The analysis model may be composed of multiple filters to compensate for the oil-gas signal input in time sequence.

[0068] The processor (150) sequentially inputs the oil-based gas signals into the analysis model by time, and analyzes the time-varying characteristics of the input oil-based gas signals to predict the oil-based gas signals to be input in the next sequence.

[0069] For example, the processor (150) can sequentially input one month's worth of daily oil-gas signals into an analysis model to estimate disturbance components related to daily load conditions, seasons, and transformer losses. At this time, the processor (150) can predict the next sequence, i.e., the oil-gas signal on the 31st day.

[0070] At this time, the analysis model may be a statistical model using a stochastic approach in the frequency domain, for example, an autoregressive (AR) model, a moving average (MA) model, an ARMA model, an ARIMA model, etc. An example of an analysis model is shown in Fig. 4.

[0071] A key feature of the statistical approach in the frequency domain is that, because it is fundamentally a time-series process, past oil-gas signals can be synthesized to generate a signal with a spectrum most similar to the current oil-gas signal. Furthermore, the internal data of the ultra-high-voltage oil-immersed transformer (10) tends to exhibit relatively little change between the past and present. Therefore, disturbance modeling is possible by selecting only the disturbance components determined by external factors of the ultra-high-voltage oil-immersed transformer (10). Furthermore, when processing in the frequency domain, modeling is possible because the spectral bandwidth is determined for each noise source.

[0072] A processor (150) according to one embodiment of the present invention can update the analysis model so that the difference between the actual oil-gas signal acquired in the next sequence and the predicted oil-gas signal is minimized (S30).

[0073] For example, the processor (150) may calculate the square of the error between the oil-based gas signal actually acquired on the 31st day and the oil-based gas signal predicted to be acquired on the 31st day. The processor (150) may update the analysis model in a direction in which the square of the error is minimized.

[0074] Specifically, the processor (150) can update the analysis model by correcting the filter coefficients of each of the plurality of filters included in the analysis model in a direction that reduces the square of the error.

[0075] At this time, the processor (150) can use the RLS (Recursive Least Squares) adaptation algorithm or the LMS (Least Mean Square) adaptation algorithm as an analysis model update algorithm. The RLS direct compensation technique can minimize the power [unit: Watt] contained in a signal and reflect time-varying characteristics with a small time delay.

[0076] Any time series analysis-based algorithm can be used to correct the filter coefficients in the direction of minimizing the difference. As an example, the update method for the RLS algorithm is shown in Table 1 below.

[0077] [Table 1]

[0078]

[0079] A processor (150) according to one embodiment of the present invention can remove a disturbance component obtained through an analysis model from an actual oil-gas signal (S40). The processor (150) can perform a direct compensation technique that uses the estimated disturbance component to remove the disturbance influence included in the oil-gas signal.

[0080] Specifically, a disturbance signal can be discovered by detecting the difference between the predicted signal and the signal acquired in the next sequence, and the χ value (ξ) shown in Fig. 4 is defined as the disturbance signal. The processor (150) can remove the disturbance by subtracting the χ value (ξ) from the currently received signal by applying an appropriate scaling.

[0081] According to one embodiment of the present invention, the time-varying characteristics of disturbance elements scattered in the operating field of an ultra-high pressure inlet transformer (10) can be automatically detected to effectively remove disturbance components of a gas signal in the frequency domain.

[0082] According to one embodiment of the present invention, it is possible to provide reliability of diagnostic data that can prevent various system issues due to abnormal voltage inflow into aging equipment in advance.

[0083] FIG. 4 is a drawing illustrating the removal of a disturbance component of a gas signal according to one embodiment of the present invention.

[0084] FIG. 4 illustrates a process of predicting an oil-gas signal to be input to the next sequence using an analysis model that estimates a disturbance component of an oil-gas signal, as previously described with reference to S20 of FIG. 3.

[0085] At this time, it is assumed that one month's worth of daily oil-gas signal data is used.

[0086] For example, the processor (150) inputs the first-day oil-gas signal into y(n-1), multiplies it by the filter coefficient w1 of the analysis model, and outputs the second-day oil-gas signal ( ) can be predicted. The processor (150) inputs the actual second-day oil-based gas signal (y(n)) and predicts the second-day oil-based gas signal ( ) can identify the difference (ξ(n)). At this time, the processor (150) can correct the filter coefficient of the analysis model through the analysis model update algorithm in the direction of minimizing the difference (ξ(n)).

[0087] Next, the processor (150) inputs the second-day oil-based gas signal into y(n-1) and the first-day oil-based gas signal into y(n-2), and adds the product of the filter coefficient w1 and y(n-1) of the analysis model and the filter coefficient w2 and y(n-2) to obtain the third-day oil-based gas signal ( ) can be predicted. The processor (150) inputs the actual 3-day oil-based gas signal (y(n)) and predicts the 3-day oil-based gas signal ( ) can identify the difference (ξ(n)). At this time, the processor (150) can correct the filter coefficient of the analysis model through the analysis model update algorithm in the direction of minimizing the difference (ξ(n)).

[0088] By continuously updating the analysis model in this way, the predicted oil-gas signal ( ) and the actual oil-gas signal (y(n)) are minimized, and the analysis model can effectively model the disturbance component of the oil-gas signal.

[0089] According to one embodiment of the present invention, a filtering technique is applied in consideration of the time-varying characteristics of the spectrum of a disturbance component, and by adaptively correcting the coefficients of a disturbance compensation filter, the performance of removing a disturbance component included in a gas-in-oil trend can be improved.

[0090] FIG. 5 is a diagram illustrating an operation flow chart of a spectrum analysis device according to a first embodiment of the present invention.

[0091] The purpose of the present invention is to estimate the internal insulation performance and overheating state of a transformer in actual substation operation by analyzing the gas signal in the oil, thereby providing overall data necessary for transformer operation.

[0092] Accordingly, the present invention can be broadly divided into a stray gas removal unit and a transformer status tracking unit. As described with reference to the preceding drawings, the stray gas removal unit acquires a stray gas signal containing a stray gas, estimates the stray gas, and then removes the stray gas.

[0093] At this time, the spectrum of disturbance components contained in the spectrum of incoming real-time oil and gas signals can be analyzed. To this end, a statistical approach in the frequency domain was adopted. This approach allows for the estimation of spectral characteristics of disturbances, enabling effective modeling of disturbances.

[0094] FIG. 6 is a diagram illustrating an operation flow chart of a spectrum analysis device according to a second embodiment of one embodiment of the present invention.

[0095] To observe the trend of the periodically measured gas-in-oil signal using online gas-in-oil measurements and predict its future direction, the following two conditions are essential:

[0096] 1) Years of gas-in-oil signals measured from hundreds of ultra-high voltage inlet transformers;

[0097] 2) A library that saves the path by displaying it on a standard DGA graph as a long-term acquired oil gas signal with the external components removed.

[0098] In the present invention, the time-varying characteristics of the disturbance component are identified for a long-term acquired oil-gas signal and removed, and the movement path of the oil-gas signal with the disturbance component removed is proposed to be loaded into a library as a default value in a standard DGA graph.

[0099] Any standard gas-in-oil graph, such as IEEE or IEC, can be adopted as a standard DGA graph. The Duval triangle graph depicted in Figure 7 is an example. The measured gas-in-oil signal at any given point in time is plotted as a single point on the graph, based on three axes representing the sizes of the major gases: CH4, C2H2, and C2H4. The more points measured, the more points are plotted on the graph, and as the measurement time increases, the points are displayed as a single trend line.

[0100] The oil gas analysis algorithm applied on-site by the installed library can accurately determine the current state of the internal electrical characteristics of the transformer and provide useful data for transformer operation.

[0101] The processor (150) can compare the oil gas signal from which the disturbance component has been removed with a library storing oil gas signals from which the disturbance component has been removed previously (S610).

[0102] The processor (150) can display the results of the oil gas analysis on a standard oil gas analysis graph (S620).

[0103] The processor (150) can receive the preprocessed signal and compare it with the library installed for the oil gas that has a high correlation with the actual deterioration of the insulating oil, and display the current internal state of the transformer on a standard DGA graph.

[0104] The processor (150) can predict which path the characteristics of the oil gas signal will take in the future based on the results of the oil gas analysis (S630).

[0105] The processor (150) can control auxiliary devices connected to the ultra-high voltage input transformer (10) when determining transformer overheating or internal arc along the predicted path (S640).

[0106] The processor (150) can provide effective information to peripheral auxiliary devices of the transformer by considering the electrical status inside the ultra-high voltage input transformer (10). The auxiliary devices can include a cooling fan, a parallel operation transformer load distribution system, a substation main line current distribution switch controller, etc.

[0107] The processor (150) receives the oil-gas signal preprocessed of the external component, removes the gas signal trend that has little correlation with the actual insulation oil deterioration, and transmits the data to a higher-level HMI (Human Machine Interface), such as an Energy Management System (EMS), a Manufacturing Execution System (MES), an Employee Assistance Program (EAP), etc. In the higher-level HMI, sophisticated operation that takes temporary oil-gas false alarms into account, etc., is possible.

[0108] In addition, the processor (150) analyzes the disturbance component based on the fact that the removed disturbance is not caused inside the transformer but is external, and can redistribute the load applied to the ultra-high voltage inlet transformer (10) during parallel operation of the transformer through objective monitoring of the facility operating environment.

[0109] The present invention is a technology for analyzing the oil gas signal spectrum of an ultra-high pressure oil-immersed transformer using natural laws, and has industrial applicability.

Claims

1. In a spectrum analysis device for the oil-gas signal of an ultra-high pressure inlet transformer, Obtain the gas signal from the ultra-high pressure oil-immersed transformer, Using an analysis model that estimates the disturbance component of the oil-gas signal sequentially input over time, the oil-gas signal to be input in the next sequence is predicted, The above analysis model is updated so that the difference between the actual oil-gas signal acquired in the next sequence and the predicted oil-gas signal is minimized, A spectrum analysis device including a processor for removing a disturbance component obtained through the analysis model from the actual oil-gas signal.

2. In paragraph 1, The above processor, A spectrum analysis device that updates the analysis model by correcting the filter coefficients of each of the plurality of filters included in the analysis model.

3. In paragraph 1, The above processor, A spectrum analysis device that analyzes the state of the ultra-high pressure inlet transformer from a signal from which the external disturbance component has been removed from the actual oil-gas signal.

4. In paragraph 3, The above processor, A spectrum analysis device that tracks the state change of a signal from which the external disturbance component has been removed based on the state history of the ultra-high voltage input transformer.

5. In paragraph 4, The above processor, A spectrum analysis device that displays the state change of the actual dissolved gas signal on a standard dissolved gas analysis (DGA) graph.

6. In paragraph 4, The above processor, A spectrum analysis device that controls auxiliary devices connected to an ultra-high voltage inlet transformer based on the results of tracking the state change of a signal from which the above external disturbance component has been removed.

7. In paragraph 1, The above analysis model is, A spectrum analysis device characterized in that it is trained to estimate the disturbance component of an input oil gas signal using a bandwidth determined by each noise source in the frequency domain.

8. In a method for analyzing the spectrum of a gas signal in an ultra-high pressure inlet transformer performed by a spectrum analysis device, Step of acquiring a gas signal from an ultra-high pressure inlet transformer; A step of predicting an oil-gas signal to be input in the next sequence using an analysis model that estimates a disturbance component of an oil-gas signal sequentially input over time; A step of updating the analysis model so that the difference between the actual oil-gas signal acquired in the next sequence and the predicted oil-gas signal is minimized; A spectrum analysis method comprising a step of removing a disturbance component obtained through the analysis model from the actual oil-gas signal.

9. In paragraph 8, The steps for updating the above analysis model are: A spectrum analysis method comprising a step of updating the analysis model by correcting the filter coefficients of each of the plurality of filters included in the analysis model.

10. In paragraph 8, A spectrum analysis method further comprising a step of analyzing the state of the ultra-high pressure inlet transformer from a signal from which the disturbance component has been removed from the actual oil-gas signal.

11. In paragraph 10, The step of analyzing the state of the above ultra-high voltage input transformer is: A spectrum analysis method comprising a step of tracing a state change of a signal from which the disturbance component has been removed based on the history of the state of the ultra-high voltage input transformer.

12. In paragraph 11, The step of analyzing the state of the above ultra-high voltage input transformer is: A spectrum analysis method comprising the step of displaying the state change of the actual dissolved gas signal on a standard dissolved gas analysis (DGA) graph.

13. In paragraph 11, A spectrum analysis method further comprising a step of controlling auxiliary devices connected to an ultra-high voltage inlet transformer based on the result of tracking the state change of the signal from which the above external disturbance component has been removed.

14. In paragraph 8, The above analysis model is, A spectrum analysis method characterized in that it is learned to estimate the disturbance component of an input oil gas signal using a bandwidth determined by each noise source in the frequency domain.

Images