JP7621910B2 - Condition diagnosis device and condition diagnosis method - Google Patents

Description

An embodiment of the present invention relates to a transformer condition diagnosis device and a condition diagnosis method.

Substations that raise and lower the voltage of electricity sent from power plants to suit each purpose are often installed away from urban areas and managed unmanned. To evaluate the soundness of transformers used in such substations, workers periodically visit the substation and perform simple inspections using gas-in-oil analysis. Transformers are also highly reliable devices that can maintain stable performance over a long period of time, and there are many cases where there is little change from the previous inspection even after inspection. For this reason, it is desirable to switch from the TBM (Time Based Maintenance) method, which performs inspections at a specified interval, to the CBM (Condition Based Maintenance) method, which uses a simple method to diagnose the condition of the equipment and determines the next inspection date depending on the diagnosis results. However, in the past, there were limited methods for predicting the condition of transformers, so appropriate diagnosis was not possible in some cases.

Patent No. 4857597

The problem that this invention aims to solve is to provide a condition diagnosis device and a condition diagnosis method that can more appropriately diagnose the future condition of a transformer.

A condition diagnosis device according to an embodiment includes a preprocessing unit, a prediction unit, a condition diagnosis unit , and a correction value generating unit . The preprocessing unit performs a predetermined preprocessing on the history information based on history information related to past inspections of the transformer and equipment information of the transformer. The prediction unit predicts a change trend of the amount of gas in oil of the transformer based on the information processed by the preprocessing unit. The condition diagnosis unit diagnoses at least one of a future inspection time, a failure time, or a lifespan of the transformer based on the change trend of the amount of gas in oil of the transformer predicted by the prediction unit. The correction value generating unit generates a correction value based on a prediction error between a predicted value of the change trend of the amount of gas in oil of the transformer by the prediction unit and an actual measured value of the change trend of the amount of gas in oil based on the inspection result of the transformer. The condition diagnosis unit corrects the change trend of the amount of gas in oil of the transformer predicted by the prediction unit based on the correction value generated by the correction value generating unit, and diagnoses at least one of a future inspection time, a failure time, or a lifespan of the transformer based on the corrected change trend.

FIG. 1 is a diagram showing an example of the configuration of a condition diagnosis device 100 according to a first embodiment. FIG. 4 is a diagram for explaining the contents of inspection history information 122. FIG. 4 is a diagram for explaining the contents of transformer information 124; 4 is a diagram for explaining the details of the functions of a prediction unit 140 and a state diagnosis unit 150; 4 is a flowchart showing the flow of a series of processes in the condition diagnosis device 100 of the first embodiment. FIG. 13 is a diagram showing an example of the configuration of a condition diagnosis device 100A according to a second embodiment. FIG. 13 is a diagram showing an example of the configuration of a condition diagnosis device 100B according to a third embodiment. 13 is a flowchart showing the flow of a series of processes in a condition diagnosis device 100B according to the third embodiment. FIG. 13 is a diagram showing an example of the configuration of a condition diagnosis device 100C according to a fourth embodiment. 13 is a flowchart showing the flow of a series of processes in a condition diagnosis device 100C according to a fourth embodiment. FIG. 13 is a diagram showing an example of the configuration of a condition diagnosis device 100D according to a fifth embodiment. FIG. 13 is a diagram for explaining deaeration determination and correction of the amount of gas in oil in the fifth embodiment. 13 is a flowchart showing the flow of a series of processes in a condition diagnosis device 100D of the fifth embodiment.

The following describes a condition diagnosis device and a condition diagnosis method according to an embodiment with reference to the drawings. The following describes a condition diagnosis device and a condition diagnosis method for diagnosing the condition of an oil-immersed transformer that has insulating oil inside a tank as an insulating and cooling medium. The transformer is, for example, a transformer installed in a power plant, a substation, or other substation equipment.

(First embodiment)
FIG. 1 is a diagram showing an example of the configuration of a condition diagnosis device 100 of the first embodiment. The condition diagnosis device 100 shown in FIG. 1 includes, for example, an inspection result acquisition unit 110, a storage unit 120, a preprocessing unit 130, a prediction unit 140, a condition diagnosis unit 150, and an output unit 160. The inspection result acquisition unit 110, the preprocessing unit 130, the prediction unit 140, and the condition diagnosis unit 150 are each realized by, for example, a hardware processor such as a CPU (Central Processing Unit) executing a program (software). Some or all of these components may be realized by hardware (including circuitry) such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a GPU (Graphics Processing Unit), or may be realized by cooperation between software and hardware. Some or all of the functions of these components may be realized by a dedicated LSI. The program may be stored in a storage device (storage device with a non-transient storage medium) such as a hard disk drive (HDD) or flash memory provided in condition diagnosis device 100, or may be stored in a removable storage medium (non-transient storage medium) such as a digital versatile disc (DVD) or CD-ROM, and installed in the HDD or flash memory provided in condition diagnosis device 100 by attaching the storage medium to a drive device of condition diagnosis device 100.

The storage unit 120 may be realized by the above-mentioned various storage devices, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a ROM (Read Only Memory), or a RAM (Random Access Memory). For example, inspection history information 122, transformer information 124, a program, and various other information are stored in the storage unit 120. The storage unit 120 may also store the results predicted by the prediction unit 140, the results diagnosed by the condition diagnosis unit 150, and information output by the output unit 160. The inspection history information 122 is an example of "history information". The transformer information 124 is an example of "equipment information". The storage unit 120 may be an external device (for example, a database server) accessible from the condition diagnosis device 100.

The inspection result acquisition unit 110 acquires the inspection result of the transformer whose condition is to be diagnosed. For example, the inspection result acquisition unit 110 acquires the inspection result by an input operation using an input unit (not shown) of the condition diagnosis device 100 by a worker who inspected the transformer or an administrator of the condition diagnosis device 100. The inspection result acquisition unit 110 may also acquire the inspection result transmitted from an external device by communicating with the external device connected via a network. The network may be, for example, a LAN (Local Area Network) or may include a WAN (Wide Area Network) such as the Internet or a cellular network. The inspection result acquisition unit 110 may also acquire information on the state of the transformer from the transformer via the network at a predetermined period (for example, daily or weekly). The information on the state of the transformer includes, for example, an analysis result of gas (hereinafter, gas in oil) contained in insulating oil for insulating the transformer from the outside. The inspection result acquisition unit 110 stores the acquired inspection result in the inspection history information 122 of the storage unit 120.

Figure 2 is a diagram for explaining the contents of the inspection history information 122. In the inspection history information 122, for example, the reason for inspection and the results of the gas-in-oil analysis are associated with the date. The date is information indicating the date on which the gas-in-oil was analyzed. The reason for inspection is information indicating the reason (e.g., regular inspection, replacement of equipment, etc.) when an operator or the like inspected, maintained, or tested the transformer. If the reason for inspection is not stored in the inspection history information 122, this indicates that the results of the gas-in-oil analysis are stored for a date other than the inspection date.

The results of the gas-in-oil analysis include, for example, the analysis results of decomposition gases generated when insulating oil or insulation is thermally decomposed due to local overheating or discharge inside the transformer. Since the decomposition gases dissolve in the insulating oil, the decomposition gases dissolved in the oil (gases in oil) are extracted to analyze the type of gas and the amount of gas for each type of gas. The gases in oil to be analyzed include, for example, hydrogen, methane, ethane, ethylene, acetylene, carbon monoxide, etc. The results of the gas-in-oil analysis may also include the total amount of flammable gas in the oil. The inspection history information 122 may also include information on other inspection results of the transformer (for example, operating status, load factor, utilization rate). For example, the information on the inspection results includes, for each transformer, information on the operating time at the time of performing the gas-in-oil analysis and information on the overload rate, which indicates the percentage of time or number of times the transformer's allowable load is exceeded. For transformers that are not equipped with devices for measuring the operating time or overload rate, the information on the inspection results includes information for obtaining these information. The above-mentioned information on the inspection results may be obtained by the preprocessing unit 130, which will be described later, from the data stored in the inspection history information 122. The inspection history information 122 may be stored for each transformer ID, which serves as identification information for identifying the transformer. By storing the above-mentioned history information in the inspection history information 122, it is possible to more appropriately predict the future state of the transformer (for example, the trend of change in the amount of gas for each gas type from the present time onwards), etc.

The inspection result acquisition unit 110 may also acquire information about the transformer whose condition is to be diagnosed, and store the acquired information in the storage unit 120 as transformer information 124. FIG. 3 is a diagram for explaining the contents of the transformer information 124. In the transformer information 124, for example, equipment data is associated with a transformer ID. The equipment data may include, for example, a manufacturer type that identifies the manufacturer that manufactured or sold the transformer, and a transformable voltage capacity that is determined by the size and performance of the transformer. The equipment data may also include information that identifies the management company that manages the transformer, and information about the parts that make up the transformer (for example, manufacturer, performance information, etc.).

The preprocessing unit 130 executes a predetermined preprocessing based on the inspection history information 122 and the transformer information 124 stored in the storage unit 120. The predetermined preprocessing is, for example, generating information to be used by the prediction unit 140 from the information included in the inspection history information 122. For example, the preprocessing unit 130 corrects or removes the amount of gas in oil that is determined to be inappropriate based on the change trend of the amount of gas in oil in the transformer from the past to the present. The gas in oil analysis result of the inspection history information 122 includes the amount of gas for each gas type obtained by component analysis of the insulating oil in the transformer, but the amount of gas may contain outliers (noise) that clearly exceed the allowable range of increase or decrease due to failure of measuring equipment, measurement method errors, input errors of measurement results by workers, etc. In addition, if the insulating oil is not circulating sufficiently in the transformer, a bias in the components may occur, and the amount of gas in the insulating oil collected during the component analysis may become a higher or lower value than the actual value, resulting in an outlier. In addition, the amount of gas in the insulating oil collected during the component analysis may become a higher or lower value than the actual value due to the influence of noise during the component analysis, resulting in an outlier. Including such outliers in the input data to the prediction unit 140 can have a significant impact on the prediction results.

Therefore, the preprocessing unit 130 detects outliers in the oil-in-gas analysis results included in the inspection history information 122 for the transformer ID to be diagnosed, where the amount of gas in oil is outside a predetermined range, and excludes the detected outliers from the data to be input to the prediction unit 140. The predetermined range is, for example, a range from a predetermined lower limit value to an upper limit value of the gas amount. The predetermined range may be set for each gas type, and may be set based on the equipment data of the transformer included in the transformer information 124. The predetermined range may also be a fixed range regardless of the gas type.

The preprocessing unit 130 may detect outliers in the gas amount contained in the inspection history information 122 by applying the Smirnov-Grubbs test, which tests particularly large and small values in the target data, and may exclude the detected outliers from the data to be input to the prediction unit 140. In addition, any method that can properly determine the gas amount analyzed by the component analysis may be applied to detect outliers, such as judgment using a filter with upper and lower limits of the gas amount change set by an expert as thresholds. Furthermore, for transformer information in which many gas amounts are determined to be outliers, or for transformers that are not determined to be outliers but have a larger increase or decrease in the gas amount compared to others, it is possible that the insulating oil or the transformer itself is in a unique condition, so such transformer information may be excluded from the learning data when generating the prediction model described later.

The preprocessing unit 130 may also exclude, as outliers, gas amounts that are outside the distribution of the gas amounts in the entire data included in the inspection history information 122 from the data to be input to the prediction unit 140. Instead of excluding the outliers from the data to be input, the preprocessing unit 130 may adjust the gas amount so that the outliers are included in a predetermined range. The preprocessing unit 130 may also interpolate with a gas amount estimated from a past change trend instead of the outlier. Specifically, the preprocessing unit 130 linearly interpolates the gas amount at the time the outlier was detected based on the gas amount detected earlier than the detected outlier. Note that the preprocessing unit 130 excludes the outliers when the increase or decrease in the past gas amount is significantly large (when the increase or decrease is greater than a threshold value) because the interpolation cannot be performed appropriately. The above-mentioned processing for outliers is applied not only to the judgment of the appropriateness of new gas-in-oil analysis results, but also to the judgment of outliers for past inspection history information, and excludes or corrects the outliers. The preprocessing unit 130 may also analyze the gas-in-oil generation algorithm from a physical or chemical perspective, and generate input target data that does not include outliers using a filter generated by the analyzed generation algorithm.

In addition, if the transformer is maintained such that at least a portion of the insulating oil is replaced, the amount of gas in the oil will be significantly reduced during the subsequent inspection, and it is highly likely that the amount will be regarded as an outlier. Therefore, the pre-processing unit 130 may refer to the inspection reason in the inspection history information 122 associated with the transformer to be diagnosed, and if the inspection reason is a reason that is predicted to cause a change in the amount of gas in the oil, such as "insulating oil replacement," the pre-processing unit 130 may not exclude the outlier from the data to be input to the prediction unit 140. In addition, the pre-processing unit 130 may adjust the amount of gas in the oil that is regarded as an outlier, depending on the amount of insulating oil replaced and its proportion to the total. In addition, the pre-processing unit 130 may perform a ratio conversion of the gas contained in the oil, etc.

The prediction unit 140 predicts the change trend of the gas-in-oil amount in the transformer using the data processed by the pre-processing unit 130. For example, the prediction unit 140 predicts the future change trend of the gas-in-oil amount for the gas-in-oil analysis results of the transformer to be diagnosed, using a prediction model that inputs the gas-in-oil analysis results and outputs the change trend of the gas-in-oil amount in the transformer. The change trend includes, for example, trend information such as the amount of gas-in-oil itself and the rate at which the amount of gas will increase in the future (or whether it will remain the same). The future refers to the period from the present time until a predetermined time has passed.

The prediction model may be obtained from an external device via a network or may be learned by the prediction unit 140. For example, the prediction unit 140 compares the change trend of the gas-in-oil amount predicted in the past with the change trend of the gas-in-oil amount derived from the actual measured value, and performs control (feedback control) to approximate the predicted change trend to the actual measured value, thereby learning the prediction model. The prediction model may be generated for each type of gas contained in the gas-in-oil, for example. The prediction model may also be learned based on the correlation between the change trends of multiple different gases.

The prediction unit 140 may also generate a prediction model by learning using the gas-in-oil analysis results of transformers, which are the same equipment data, based on the equipment data of the transformer information 124. The same equipment data is, for example, equipment data containing the same manufacturer type and voltage capacity, and the voltage capacity may include a predetermined error range. The prediction unit 140 may also learn the prediction model using only the gas-in-oil analysis results of transformers of the same manufacturer type, may also learn using only the gas-in-oil analysis results of transformers of the same voltage capacity, or may also learn using the gas-in-oil analysis results of the same transformer. The prediction unit 140 can select which gas-in-oil amount to use for learning, for example, depending on the characteristics of the transformer, the characteristics of the gas type to be predicted, the characteristics of the prediction method, and even the number of data available for learning.

The prediction unit 140 may learn a prediction model using various machine learning methods or neural networks such as random forest or LSTM (Long Short Term Memory), generate a prediction model based on a clustering method, or generate a prediction model using a VAR model (Vector Auto Regression Model) or a VARMA model (Vector Auto Regression Moving Average model). The prediction unit 140 may select a prediction model (prediction method) to be used for prediction, for example, according to the type of transformer to be predicted and the type and characteristics of the gas in the oil. For example, for a transformer in which no correlation is recognized between gas types and in which the changes in other gas types become noise and appropriate prediction becomes difficult when a prediction model is constructed by including other gas types as learning data, the prediction unit 140 generates a prediction model by learning about one gas type of multiple transformers using random forest or LSTM. Furthermore, when predicting the gas amount of a gas type in which a strong correlation is recognized between the gas types in the transformer, the prediction unit 140 generates a prediction model by learning about all gas types in multiple transformers using a VAR model, which is a time series model related to feature vectors, a VARMA model, or the like. Then, the prediction unit 140 predicts the future change trend of the gas amount for one or more gas types based on the generated prediction model. Furthermore, the prediction unit 140 may predict the future change trend of the gas amount by comprehensively determining the change trend of the gas amount for multiple gas types.

The condition diagnosis unit 150 diagnoses the future condition of the transformer based on the change trend of the amount of gas in the oil of the transformer predicted by the prediction unit 140. The future condition includes, for example, at least one of the next inspection time, future failure time, and lifespan. The lifespan is, for example, the remaining period (remaining life) during which the transformer can be operated. The functions of the prediction unit 140 and the condition diagnosis unit 150 will be described in detail later.

The output unit 160 generates information corresponding to the destination device based on the diagnosis result of the condition diagnosis unit 150, and outputs the generated information to the destination. For example, if the destination device is a display device, the output unit 160 generates an image indicating the diagnosis result, and if the destination device is an audio output device, the output unit 160 generates audio indicating the diagnosis result.

Next, the details of the functions of the prediction unit 140 and the state diagnosis unit 150 will be specifically described. FIG. 4 is a diagram for explaining the details of the functions of the prediction unit 140 and the state diagnosis unit 150. In the example of FIG. 4, the actual gas amount (actual measurement value) of a specific gas (hereinafter referred to as gas A) contained in the oil gas of the transformer over time and the change tendency of gas A are shown. The horizontal axis of FIG. 4 indicates time t, and the vertical axis indicates the gas amount [ppm] of gas A. Time t may be a date. In the example of FIG. 4, time t0 indicates the start time of operation of the transformer, and times t1, t2, and t3 indicate the past inspection dates. In addition, in the example of FIG. 4, the gas amount (actual measurement value) of gas A is continuously acquired from the transformer at a predetermined cycle during the period from time t0 to t3. In addition, in the example of FIG. 4, inspection is performed at a predetermined cycle (for example, time △Ta) from time t0. In the example of FIG. 4, when the gas amount of gas A is between 0 and G1, the transformer is considered to be normal, and when the gas amount is between G1 and G2, the transformer is considered to be in a section requiring maintenance (a section requiring caution). The gas amounts G1 and G2 may differ depending on the type of gas to be diagnosed, for example, or may be the same regardless of the gas type.

The prediction unit 140 obtains the past (before time t3) change trend based on the gas volume of gas A from time t0 to t3. For example, the prediction unit 140 generates an approximation line that minimizes the error of the gas volume from time t0 to t3 by the least squares method. The prediction unit 140 also uses this line as the past change trend of gas A and predicts the future change trend (after time t3) of the gas volume of gas A based on the past change trend. For example, as shown in FIG. 4, the prediction unit 140 may predict the future change trend by applying the line of the past change trend to time t3 and after, or may predict the future change trend by changing the slope of the line of the past change trend shown in FIG. 4 according to the gas type and transformer information 124. The prediction unit 140 may also obtain the past change trend of the gas volume or predict the future change trend using an approximation curve based on a predetermined function instead of a straight line. The prediction unit 140 may also input the gas volume of gas A from time t0 to t3 into the above-mentioned prediction model to predict the future change trend. Then, the prediction unit 140 predicts the time t10 at which the amount of gas A will reach G1 based on the change trend.

The gas amount G1 at the boundary between the normal zone and the caution zone may include an allowable range in which the prediction reliability is equal to or greater than a threshold. For example, when a predetermined allowable amount ΔG is set based on the prediction reliability with the gas amount G1 as the standard, the prediction unit 140 predicts a period ΔP in which the gas amount will reach the range of the allowable amount ΔG based on the future change trend of the gas amount of gas A. The period ΔP is the range of times centered on time t10 in which the prediction accuracy is equal to or greater than a threshold.

The condition diagnosis unit 150 diagnoses at least one of the future inspection timing, failure timing, or lifespan of the transformer based on the prediction result by the prediction unit 140. In the example of FIG. 4, the condition diagnosis unit 150 diagnoses time t5, which is a period included in the period ΔP in the inspection cycle (time ΔTa) from time t3, as the next inspection timing (one example of a future inspection timing). That is, in the example of FIG. 4, the condition diagnosis unit 150 diagnoses that inspection is not performed at time t4, when time ΔTa has passed from time t3, and that the transformer is in a state where inspection is to be performed at time t5, when time ΔTa has further passed. Note that it is preferable for the condition diagnosis unit 150 to set the future inspection timing within the period ΔP so as not to exceed time t10. Therefore, if the above-mentioned time t5 is later than time t10, the condition diagnosis unit 150 sets time t10 as the inspection timing. This prevents the gas volume from exceeding the gas volume G1, and a safer inspection timing can be set.

Furthermore, the condition diagnosis unit 150 may diagnose that it is necessary to replace the insulating oil in the transformer or replace the transformer before a condition requiring attention is reached if the amount of gas A exceeds G1 at time t4, which is a time ΔTa that has elapsed since time t3, based on the future change trend. Furthermore, the condition diagnosis unit 150 may diagnose time t10 as the end of the transformer's life if time t10 is a predetermined time or more from the start of operation of the transformer. In this case, the condition diagnosis unit 150 may diagnose the time associated with the life as the time to replace the transformer.

Depending on the type of gas, the condition diagnosis unit 150 may diagnose time t10 as the time for a transformer failure or for changing the insulating oil, rather than as the inspection period. For example, if the gas in the oil is gas A, the condition diagnosis unit 150 may diagnose time t5 as the time for inspection, if the gas in the oil is gas B different from gas A, the condition diagnosis unit 150 may diagnose time t5 as the time for a transformer failure, and if the gas is gas C different from gases A and B, the condition diagnosis unit 150 may diagnose time t5 as the time for changing the insulating oil. This allows the transformer to be diagnosed more appropriately based on the future trend of changes in the gas volume for each gas type.

The condition diagnosis unit 150 may also diagnose that a fault has occurred in the transformer if the future change trend of the gas amount exceeds a predetermined amount and is on a downward trend, if the gas amount does not reach the allowable amount ΔG even after a predetermined time has passed since time t3, or if the past gas amount has already exceeded gas amount G1 or G2.

Figure 5 is a flowchart showing a series of processing steps in the condition diagnosis device 100 of the first embodiment. In the example of Figure 5, it is assumed that the inspection history information 122 and the transformer information 124 are already stored in the storage unit 120. The same applies to the explanation of the flowcharts in the following embodiments. In the example of Figure 5, the pre-processing unit 130 acquires the inspection history information 122 and the transformer information 124 stored in the storage unit 120 (step S100), and performs pre-processing on the gas-in-oil analysis results included in the inspection history information 122 based on the acquired information (step S102).

Next, the prediction unit 140 predicts the past and future trends of change in gas volume based on the data processed by preprocessing (step S104). Next, the condition diagnosis unit 150 diagnoses the future condition of the transformer based on the prediction result (step S106). Next, the output unit 160 outputs the future condition of the transformer diagnosed by the condition diagnosis unit 150 (step S108).

As described above, according to the first embodiment, the future state of the transformer can be diagnosed more appropriately. Specifically, according to the first embodiment, by diagnosing the state of the transformer based on the analysis results of gas in the insulating oil of the transformer, it is possible to provide information regarding the next inspection timing and the timing of equipment replacement.

In addition, according to the first embodiment, it is only necessary to perform inspections at more appropriate times, and the workload of workers can be reduced compared to the TBM method. For example, in a situation where transformer inspections are performed every six months, if the future trend shows that the amount of gas A will reach G1 in five years, the next inspection can be postponed to four and a half years later, assuming six-month intervals.

In addition, according to the first embodiment, if the result shows that the gas volume of Gas A will exceed G1 six months later, it is possible to take measures such as changing the oil in the transformer or replacing the transformer as transformer maintenance before it becomes a condition requiring attention. Therefore, it is possible to efficiently reduce the risk of power outages and the like due to transformer failure.

Second Embodiment
Next, a condition diagnosis device of the second embodiment will be described. For example, the amount of gas in oil of a transformer basically tends to increase due to chemical changes and aging inside the transformer. However, the amount of increase may vary depending on the equipment characteristics and usage conditions of the transformer. In addition, there are prediction methods that are good at predicting the long-term change trend of the amount of gas in oil, those that are good at short-term prediction, and those that wait for high reliability (precision) for a specific gas. Therefore, in the second embodiment, in order to predict the change trend of the amount of gas in oil, prediction is performed using multiple prediction methods, and the multiple prediction results are comprehensively judged to diagnose the state of the transformer. In the following description, among the components of the condition diagnosis device 100 in the first embodiment, components that have the same function are assigned the same reference numerals as in the first embodiment, and further specific descriptions of common matters are omitted. The same applies to the descriptions of the following embodiments.

Figure 6 is a diagram showing an example of the configuration of a condition diagnosis device 100A of the second embodiment. The condition diagnosis device 100A shown in Figure 6 includes, for example, an inspection result acquisition unit 110, a storage unit 120, a pre-processing unit 130A, multiple prediction units 140A-1 to 140-n (n is an integer equal to or greater than 2), a condition diagnosis unit 150, and an output unit 160. Compared to the condition diagnosis device 100 of the first embodiment, the condition diagnosis device 100A differs in that, instead of the pre-processing unit 130, the prediction unit 140, and the condition diagnosis unit 150, the condition diagnosis device 100A includes a pre-processing unit 130A, multiple prediction units 140A-1 to 140A-n, and a condition diagnosis unit 150A. The following mainly describes the above-mentioned differences.

The preprocessing unit 130A of the second embodiment outputs preprocessed information to each of the multiple prediction units 140A-1 to 140A-n. In this case, the preprocessing unit 130A may adjust the format of the input data to correspond to the prediction method of each prediction unit 140A.

Each of the multiple prediction units 140A-1 to 140A-n predicts the change trend of the gas in the oil using a prediction method different from that of the other prediction units based on the data input from the pre-processing unit 130A. For example, a prediction method with different characteristics is applied based on the transformer or gas type to be diagnosed. The characteristics are, for example, characteristics that are good at predicting long-term change trends in gas volume, good at short-term change trends, or have high reliability for specific gas volumes, depending on the characteristics of the algorithm of the prediction model.

The types and number of prediction methods executed by prediction units 140A-1 to 140A-n may be selected, for example, based on the equipment data (manufacturer type, voltage capacity), transformer equipment characteristics, and usage status included in transformer information 124, or may be selected based on the type of gas to be predicted. For example, prediction method A is selected when predicting the change trend of the amount of gas A contained in oil-in-gas in a transformer made by manufacturer A, and prediction method B is selected when predicting the change trend of the amount of gas A contained in oil-in-gas in a transformer made by manufacturer B. Furthermore, prediction method C is selected when the transformer has been in operation for 10 years or more, and prediction method D is selected when the transformer has been in operation for 20 years or more.

The change tendency of the gas amount of a transformer differs depending on the voltage class and the load condition. The voltage class is the magnitude of the available voltage set for each transformer. The load condition is the degree of load on the transformer based on the usage conditions such as the continuous operation time and frequency of use of the transformer, the change tendency of the amount of gas in oil to date, and the ratio of the amount of gas in oil to the type of gas in oil. Therefore, if a prediction model is generated using only a specific method, it may not be possible to properly learn the change tendency. Therefore, in the second embodiment, the prediction units 140A-1 to 140A-n predict the change tendency of the amount of gas in oil of the transformer by switching the prediction method of the amount of gas in oil of the transformer or by combining multiple prediction methods based on one or both of the voltage class and the load condition of the transformer to be diagnosed.

For example, in the second embodiment, the preprocessing unit 130A calculates the range of increase and decrease in the change in the gas amount up to the present, and if the calculated range (change) varies widely (if the range is equal to or greater than a threshold), it is determined that the transformer is prone to noise influence or bias in the insulating oil. Then, the preprocessing unit 130A selects a prediction method suitable for trend prediction that can determine whether the change in the gas amount will tend to increase or decrease, and if the range of increase and decrease in the change in the gas amount is stable, selects a method for predicting the gas amount itself and makes the prediction unit make a prediction. In addition, the preprocessing unit 130A may set a prediction method when the range of increase and decrease in the change is equal to or greater than a threshold, and a prediction method when the range of increase and decrease in the change is less than the threshold, and make the prediction unit make a prediction by switching the prediction method according to the range of increase and decrease in the transformer. In addition, depending on the transformer, the change trend of the gas amount may differ depending on the gas type. Therefore, the prediction units 140A-1 to 140A-n may make predictions using prediction methods associated with each gas type, and combine the prediction results to predict the change in the overall gas amount. The type and number of prediction methods may be selected by the user of the condition diagnosis device 100A, or a fixed type and number may be set. Each of the multiple prediction units 140A-1 to 140A-n outputs a prediction result to the condition diagnosis unit 150A.

The condition diagnosis unit 150A predicts the future change trend of the gas-in-oil amount by comprehensively judging the prediction results of the change trend of the gas amount obtained from one or more of the multiple prediction units 140A-1 to 140A-n. For example, the condition diagnosis unit 150A may diagnose the future condition of the transformer using the average result of the future change trends of the multiple gas-in-oil amounts, or may diagnose the future condition using the change trend that maximizes or minimizes the change.

The condition diagnosis unit 150A may also assign a predetermined weight to each prediction method according to the gas type, and increase the priority of the results diagnosed based on the change trend predicted by a prediction method with a larger weight. In this case, the weight according to the gas type is set so that, for example, the closer the prediction result of the change trend of the gas amount in the past is to the change trend of the actual gas amount, the heavier the weight for that gas type.

For example, the condition diagnosis unit 150A may diagnose the transformer based on future change trends that combine prediction results from a prediction method that excels at long-term prediction with prediction results from a prediction method that excels at short-term prediction. The condition diagnosis unit 150A may also predict the change trend of the gas volume of gas C and diagnose the transformer based on the correlation between the change trend of the gas volume of gas A performed by the prediction unit 140A-1 and the change trend of the gas volume of gas B performed by the prediction unit 140A-2.

In the case of newly installed transformers, there is little past inspection history, and therefore little learning data for generating a prediction model. As a result, it may not be possible to properly predict the future trend of changes in the amount of gas in the oil. In this case, at least one of the multiple prediction units 140A-1 to 140A-n makes predictions using a prediction method that uses either or both of information based on the physical perspective of the gas in the transformer's oil (hereinafter, physical perspective information) and information based on the chemical perspective (hereinafter, chemical perspective information).

The physical perspective information includes, for example, the correlation between each gas type, the amount of change in the transformer over time, the load factor of the transformer, etc. At least one of the multiple prediction units 140A-1 to 140A-n inputs the above-mentioned physical perspective information instead of (or in addition to) the results of the gas-in-oil analysis, and makes a prediction based on a prediction logic that outputs the future trend of change in the gas volume.

The chemical information also includes information on the type and amount of gas generated by, for example, overheating or discharge of insulating oil, or decomposition of insulating paper (cellulose). For example, when a transformer overheats or discharges, insulating oil decomposes and generates a specific gas. For example, in the case of local overheating, methane is generated, and as the temperature increases, hydrogen increases and acetylene is also generated. In the case of partial discharge, mainly hydrogen and acetylene are generated. In the case of arc discharge, hydrogen and acetylene are generated in large amounts, and ethylene and methane are also generated. In addition, insulating paper used as a conductor coating for windings inside a transformer is gradually decomposed by heat generated by the large current flowing through the winding and moisture adsorbed on the organic material, generating carbon dioxide, furfural, acetone, etc. The amount of these gases generated is determined by the current value and the amount of moisture. At least one of the multiple prediction units 140A-1 to 140A-n inputs the above-mentioned chemical perspective information instead of (or in addition to) the results of the gas-in-oil analysis, and makes a prediction based on a prediction logic that outputs the future change trend of the gas amount. In this way, by making a prediction based on physical perspective information and chemical perspective information, it is possible to predict the change trend even when there are no or few gas-in-oil analysis results.

The prediction unit 140A may predict the change trend of the gas amount using both the physical perspective information and the chemical perspective information described above. For example, the likelihood of overheating or discharge of insulating oil, and decomposition of insulating paper (cellulose), and the amount of gas generated vary depending on the age and load factor of the transformer. Therefore, by combining these pieces of information, it is possible to more appropriately predict the change trend of the gas amount.

The condition diagnosis unit 150A can perform a diagnosis that fully reflects the inherent characteristics of the transformer being predicted by comprehensively diagnosing the future change trend of the gas amount, including the results predicted based on the above-mentioned physical perspective information and chemical perspective information. Note that predictions using physical perspective information or chemical perspective information are not limited to newly installed transformers, and may be applied, for example, to transformers that have been in use for many years and for which sufficient learning data can be obtained.

The flow of the series of processes in the condition diagnosis device 100A of the second embodiment can be performed in a similar manner to the processes in steps S100 to S108 in the condition diagnosis device 100 of the first embodiment described above, so a detailed description will be omitted here. In the case of the second embodiment, in the process of step S102, the pre-processing unit 130A performs pre-processing, and in the process of step S104, the prediction units 140A-1 to 140A-n each predict the change trend of the gas-in-oil amount in the transformer using a different prediction method. In the process of step S106, the condition diagnosis unit 150A comprehensively judges the results predicted by the multiple prediction units 140A-1 to 140A-n to diagnose the condition of the transformer.

As described above, the second embodiment not only achieves the same effects as the first embodiment, but also uses multiple prediction units 140A-1 to 140A-n to perform predictions using different prediction methods, and then comprehensively evaluates each prediction result to derive a final prediction result, thereby enabling a more appropriate diagnosis of the transformer's condition.

Third Embodiment
Next, a condition diagnosis device according to a third embodiment will be described. For example, the prediction model used in the prediction unit 140 is generated by learning the past change trends of the amount of gas in oil, but since the learning is performed including the time element, factors such as aging deterioration are also included. However, since the aging deterioration differs depending on the usage conditions of the transformer and the influence of the surrounding environment, it is possible that the aging deterioration has progressed since the previous inspection. Therefore, in the third embodiment, the amount of gas in oil of the transformer predicted at the previous inspection (predicted value) is compared with the newly acquired amount of gas in oil of the transformer (actual value), and if an error equal to or greater than a threshold occurs compared to the past prediction error, it is determined that a change has occurred in the trend of aging deterioration, and the change trend of the amount of gas in oil of the transformer is corrected.

Figure 7 is a diagram showing an example of the configuration of a condition diagnosis device 100B of the third embodiment. The condition diagnosis device 100B includes, for example, an inspection result acquisition unit 110, a storage unit 120, a preprocessing unit 130, a prediction unit 140, a condition diagnosis unit 150B, an output unit 160, and a correction value generation unit 170. Compared to the condition diagnosis device 100 of the first embodiment, the condition diagnosis device 100B differs in that it includes a condition diagnosis unit 150B instead of the condition diagnosis unit 150, and further includes a correction value generation unit 170. The following mainly describes the above-mentioned differences.

The correction value generating unit 170 compares the amount of gas in the oil of the transformer predicted at the time of the previous inspection with the amount of gas in the oil of the transformer newly acquired this time, and obtains the prediction error of the amount of gas in the oil (e.g., the error for each gas amount for each gas type). Furthermore, if the prediction error is equal to or greater than a threshold, the correction value generating unit 170 assumes that a change has occurred in the tendency of deterioration over time, and generates a correction value for the change trend of the amount of gas in the oil of the transformer. For example, the correction value generating unit 170 generates a correction value for the change trend of the amount of gas in the oil so that the prediction error of the amount of gas in the oil is less than the threshold.

The condition diagnosis unit 150B corrects the change trend of the amount of gas in the oil predicted by the prediction unit 140 with the correction value generated by the correction value generation unit 170, and diagnoses the future state of the amount of gas in the oil. Note that if a correction value is not generated by the correction value generation unit 170, the condition diagnosis unit 150B does not correct the change trend of the amount of gas in the oil predicted by the prediction unit 140.

Figure 8 is a flowchart showing the flow of a series of processes in the condition diagnosis device 100B of the third embodiment. The example in Figure 8 differs from the process of steps S100 to S108 of the condition diagnosis device 100 of the first embodiment shown in Figure 5 in that the process of steps S200 to S204 has been added. Therefore, the following description will mainly focus on the above-mentioned differences, and descriptions of other processes will be omitted.

In the example of FIG. 8, after processing in step S104, the correction value generating unit 170 compares the amount of gas in oil predicted during the previous inspection with the amount of gas in oil of the transformer acquired this time (step S200). Next, the correction value generating unit 170 determines whether the error between the amount of gas in oil predicted during the previous inspection and the amount of gas in oil of the transformer acquired this time is equal to or greater than a threshold value (step S202). If it is determined that the error is equal to or greater than the threshold value, a correction value is generated based on the amount of error (step S204).

In the third embodiment, after the processing of step S204 or in the processing of step S202, if the error is less than the threshold value, the condition diagnosis unit 150B diagnoses the future condition of the transformer (step S106). In this case, if the error is equal to or greater than the threshold value, the condition diagnosis unit 150B diagnoses the current condition of the transformer or predicts the future condition based on the predicted change trend of the amount of gas in the oil and the correction value. If the error is less than the threshold value, the condition diagnosis unit 150B diagnoses the current condition of the transformer or predicts the future condition based on the predicted change trend of the amount of gas in the oil.

As described above, the third embodiment not only achieves the same effects as the second embodiment, but also corrects the prediction information according to the amount of error obtained by comparing the predicted value with the actual measured value, thereby making it possible to more appropriately predict the future state of the transformer.

(Fourth embodiment)
Next, a fourth embodiment of the condition diagnosis device will be described. For example, when predicting the change trend of the amount of gas in oil in a transformer, a prediction model is generated based on the past amount of gas in oil in the transformer and transformer equipment data. However, if some phenomenon (a sign of transformer failure) occurs in the internal state of the transformer just before inspection, this phenomenon may cause a change in the amount of gas in oil in the transformer. Therefore, there are cases where the condition of the transformer cannot be properly diagnosed based only on the change in the amount of gas in oil in the transformer at the time of inspection. Therefore, in the fourth embodiment, the change trend of the amount of gas in oil predicted by the prediction unit 140 is corrected based on the sensing data detected by a sensor that detects the state of the transformer.

Figure 9 is a diagram showing an example of the configuration of a condition diagnosis device 100C of the fourth embodiment. The condition diagnosis device 100C shown in Figure 9 includes, for example, an inspection result acquisition unit 110, a memory unit 120C, a preprocessing unit 130, a prediction unit 140, a condition diagnosis unit 150C, an output unit 160, a sensing data acquisition unit 180, and a condition estimation unit 190. Compared to the condition diagnosis device 100 of the first embodiment, the condition diagnosis device 100C differs in that it includes a memory unit 120C and a condition diagnosis unit 150C instead of the memory unit 120 and the condition diagnosis unit 150. Furthermore, compared to the condition diagnosis device 100, the condition diagnosis device 100C differs in that it includes a sensing data acquisition unit 180 and a condition estimation unit 190. Therefore, the following description will mainly focus on the above-mentioned differences.

As shown in FIG. 9, one or more sensors are installed on the transformer 200 to be diagnosed. The sensor 210 includes, for example, a sound sensor such as a microphone that detects sound output from the transformer 200, a vibration sensor such as an acceleration sensor that detects vibration of the transformer 200, a temperature sensor that detects the temperature of the transformer (for example, the internal temperature or the external temperature), and an odor sensor that detects the odor of a specific part or the entire transformer 200. The odor sensor may detect the degree (level) of an abnormal odor or odor, or may detect the type of gas that is the source of the odor. The sensor 210 may be a permanent sensor attached to the transformer 200, or a sensor that is temporarily installed only during inspection (hereinafter referred to as a temporary sensor). The temporary sensor is a sensor that can be installed detachably on the transformer 200 or its surroundings. The temporary sensor is installed, for example, on a transformer that is determined to have a high possibility of failure in the near future based on the previous diagnosis results, a transformer whose state is to be obtained in detail, or a transformer whose state estimation and state diagnosis accuracy is to be improved. By installing a temporary sensor, more information can be detected, improving the estimation accuracy of the condition estimation unit 190 and the diagnosis accuracy of the condition diagnosis unit 150. The sensor 210 transmits the detected data (sensing data) to the condition diagnosis device 100C via a network such as a LAN or WAN at a predetermined interval or at a predetermined timing.

The sensing data acquisition unit 180 acquires the sensing data transmitted from the sensor 210. The sensing data acquisition unit 180 also associates the acquired sensing data with the transformer ID and time information at which the sensing data was detected, and stores the associated sensing data in the memory unit 120C as sensing data 126. The sensing data acquisition unit 180 may also store the acquired data in the inspection history information 122, or output the data to the condition estimation unit 190. The memory unit 120C stores the sensing data 126 in addition to the information stored in the memory unit 120 of the condition diagnosis device 100.

The state estimation unit 190 acquires sensing data of the transformer ID to be diagnosed from the sensing data 126 stored in the memory unit 120C, and performs estimation processing on the internal state of the transformer 200 based on the acquired sensing data. For example, the state estimation unit 190 performs frequency analysis of the sound output from the transformer included in the sensing data, and estimates the state of the transformer 200 by comparing the frequency characteristics of the sound obtained by the analysis with the frequency characteristics of the sound output when the transformer is malfunctioning. For example, if the similarity of the frequency characteristics of the sounds is less than a predetermined value, it estimates that the transformer 200 is in a normal state, and if it is equal to or greater than the predetermined value, it estimates that the transformer 200 is malfunctioning or is highly likely to malfunction in the near future (a predetermined short period of time).

In addition, instead of (or in addition to) sound-based estimation, the state estimation unit 190 may perform frequency analysis of the vibrations of the transformer 200 contained in the sensing data. In this case, the state estimation unit 190 estimates the state of the transformer by comparing the frequency characteristics of the vibrations of the analysis results with the frequency characteristics of the vibrations output when the transformer 200 is malfunctioning. In addition, instead of (or in addition to) sound or vibration, the sensing data acquisition unit 180 may estimate the state of the transformer by comparing the temperature contained in the sensing data with the temperature when the transformer is malfunctioning.

In addition, the state estimation unit 190 may estimate the state of the transformer 200 based on the level and type of odor of the transformer 200 contained in the sensing data, instead of (or in addition to) sound, vibration, or temperature. In this case, the state estimation unit 190 may estimate that the greater the level of odor, the more likely it is that a malfunction will occur in the transformer 200, and may estimate that a malfunction has occurred in the transformer 200 when the level of odor of a specified gas type is equal to or greater than a threshold value.

The state estimation unit 190 may also estimate the state of the transformer by selectively using one or more of the data indicating sound, vibration, temperature, and smell contained in the sensing data depending on the type and characteristics of the transformer 200 and the surrounding environment of the transformer 200. The state estimation unit 190 may also periodically monitor the state of the transformer 200 based on the sensing data acquired by the sensing data acquisition unit 180 from the sensor 210.

The condition diagnosis unit 150C diagnoses the condition of the transformer 200 based on the change trend of the amount of gas in the oil predicted by the prediction unit 140 and the condition estimated by the condition estimation unit 190. For example, when the condition estimation unit 190 estimates that a malfunction has occurred or may occur in the near future, based on the change trend of the amount of gas in the oil predicted by the prediction unit 140, the condition diagnosis unit 150C performs a diagnosis by correcting the increase trend of the amount of gas in the oil to be larger. Furthermore, when the condition diagnosis unit 150C estimates that a malfunction has occurred, it may perform a diagnosis to perform an inspection immediately.

Figure 10 is a flowchart showing the flow of a series of processes in the condition diagnosis device 100C of the fourth embodiment. The example in Figure 10 differs from the process of steps S100 to S108 of the condition diagnosis device 100 of the first embodiment shown in Figure 5 in that steps S210 to S214 have been added instead of step S106. Therefore, the following description will mainly focus on the above-mentioned differences, and descriptions of other processes will be omitted.

In the example of FIG. 10, after processing of step S104, the sensing data acquisition unit 180 acquires sensing data detected by the sensor 210 installed in the transformer 200 to be diagnosed (step S210). Next, the state estimation unit 190 estimates the state of the transformer based on the sensing data (step S212). Next, the state diagnosis unit 150C diagnoses the state of the transformer 200 based on the predicted result of the change trend of the amount of gas in the oil and the estimation result of the state estimation unit 190 (step S214).

As described above, the fourth embodiment not only achieves the same effects as the first embodiment, but also makes it possible to comprehensively diagnose the state of the transformer using the estimated results of the transformer state based on the sensing data. This makes it possible to more appropriately predict the future state of the transformer. Furthermore, according to the fourth embodiment, it is possible to more accurately diagnose the future state of nearby transformers based on the sensing data.

Fifth Embodiment
Next, a fifth embodiment of the state diagnosis device will be described. For example, in the maintenance of a transformer, a degassing operation may be performed. The degassing operation is, for example, a process of removing gas in oil (dissolved gas) by replacing at least a part of the insulating oil, or a process of reducing the amount of gas in the oil by other methods. Since the amount of gas in the insulating oil is greatly reduced by performing this degassing operation, the continuity of the gas amount before and after the degassing operation is lost. Therefore, when the change trend of the gas amount is predicted using information on the gas amount after the degassing operation, the discontinuity is also reflected in the prediction model, and an inappropriate prediction result may be obtained. Therefore, in the fifth embodiment, the location where the degassing operation is estimated to have been performed is determined based on the history information (the change trend of the gas amount in the transformer from the past to the present) stored in the inspection history information 122, and the amount of gas in oil, which is data that is not appropriate for prediction due to the degassing operation, is corrected based on the change and continuity of the gas amount before that location.

Figure 11 is a diagram showing an example of the configuration of a condition diagnosis device 100D of the fifth embodiment. The condition diagnosis device 100D includes, for example, an inspection result acquisition unit 110, a storage unit 120, a preprocessing unit 130D, a prediction unit 140, a condition diagnosis unit 150, and an output unit 160. Compared to the condition diagnosis device 100 of the first embodiment, the condition diagnosis device 100D differs in that it includes a preprocessing unit 130D instead of the preprocessing unit 130. The following description will mainly focus on the above-mentioned differences.

The pre-processing unit 130D performs the following processing in addition to (or instead of) the processing of the pre-processing unit 130 of the first embodiment. The pre-processing unit 130D includes, for example, a degassing determination unit 132 and a correction unit 134. The degassing determination unit 132 compares the amount of gas in the oil at two consecutive inspection times based on the history information stored in the inspection history information 122, and determines whether or not degassing work was performed between the previous inspection and the subsequent inspection. The above-mentioned degassing work may also be performed during the previous inspection. For example, the degassing determination unit 132 determines that degassing work was performed if the amount of gas in the oil of the transformer at the subsequent inspection has decreased by more than a threshold value since the previous inspection, and determines that degassing work was not performed if the amount has not decreased by more than the threshold value. The amount of gas may be one predetermined gas component, or the average of the amount of decrease of multiple gas components may be compared with the threshold value.

Figure 12 is a diagram for explaining deaeration judgment and correction of the amount of gas in the oil in the fifth embodiment. In Figure 12, the horizontal axis indicates time t, and the vertical axis indicates the amount of gas (amount of gas in the oil) [ppm] of a certain gas component in the oil. Times t11 to t16 indicate each inspection time point in a predetermined cycle. The graph on the left side of Figure 12 indicates the amount of gas in the oil based on the history information stored in the inspection history information 122, and the graph on the right side of Figure 12 indicates the amount of gas in the oil after correction.

In the graph on the left side of FIG. 12, the amount of gas in the oil GA1 at the time of inspection at time t13 decreases to the amount of gas in the oil GA2 at the time of the subsequent inspection (time t14) (ΔGA) equal to or greater than a threshold value. In this case, the degassing determination unit 132 determines that the degassing operation was performed between times t13 and t14.

In addition, when an inflection point based on the change trend of the amount of gas in the oil from time t11 to t16 included in the inspection history information 122 is extracted, the degassing determination unit 132 may determine that degassing work was performed between the inspection points before and after the extracted inflection point. An inflection point based on the change trend is, for example, a point where the sign of the second derivative changes when the change trend of the amount of gas in the oil is represented by a curve (a point where the curve switches between an upward convex state and a downward convex state).

When the degassing determination unit 132 determines that the degassing operation has been performed, the correction unit 134 performs a correction to increase the amount of gas in the oil after the degassing operation has been performed, based on the amount of gas in the oil before the degassing operation was performed. For example, the correction unit 134 obtains the change trend of the amount of gas in the oil from time t11 to time t13, and corrects the amount of gas in the oil from time t14 to t16 based on the obtained change trend so that the amount of gas in the oil changes with the same change trend as before time t13, using the amount of gas in the oil GA1 at time t13 as a reference.

Specifically, the correction unit 134 calculates the average increase between the amount of gas in the oil from time t11 to time t12 (first increase) and the amount of gas in the oil from time t12 to time t13 (second increase), and corrects the amount of gas in the oil by adding the average increase to the actual amount of gas in the oil GA1 at time t13 as the amount of gas at time t14. The correction unit 134 may also add only the second increase, which is the increase immediately before, to the amount of gas in the oil GA1 to obtain the amount of gas in the oil at time t14. The correction unit 134 may also add the value obtained by subtracting the first increase from the second increase (second increase - first increase) to the second increase, and then add this value to the amount of gas in the oil GA1 to obtain the amount of gas in the oil at time t14. In addition, the amount of gas in the oil at times t15 and t16 is corrected by adding the increase from the amount of gas in the oil at time t14 to the corrected amount of gas in the oil.

In addition, when an inflection point based on the change trend of the amount of gas in the oil is extracted, the correction unit 134 may correct the amount of gas in the oil after the deaeration operation based on the amount of gas in the oil at the time of inspection before and after the inflection point. In this case, the correction unit 134 performs correction so that the continuity of the change trend is maintained based on the amount of gas in the oil obtained from multiple measurement results before and after the inflection point.

As a result, as shown in the graph on the left side of Figure 12, it is possible to obtain a change trend that maintains continuity, and based on this change trend, it is possible to make more appropriate predictions from time t16 onwards.

Figure 13 is a flowchart showing the flow of a series of processes in the condition diagnosis device 100D of the fifth embodiment. The example in Figure 13 differs from the process of steps S100 to S108 of the condition diagnosis device 100 of the first embodiment shown in Figure 5 in that the process of steps S220 to S224 has been added. Therefore, the following description will mainly focus on the above-mentioned differences, and descriptions of other processes will be omitted.

In the example of FIG. 13, after the processing of step S102, the degassing determination unit 132 compares the amount of gas in oil at two consecutive inspection points based on the history information stored in the inspection history information 122 (step S220), and determines whether the amount of gas in oil in the transformer at the later inspection has decreased by more than a threshold value since the previous inspection (step S222). If it is determined that the amount of gas in oil in the transformer at the later inspection has decreased by more than a threshold value since the previous inspection, the degassing determination unit 132 determines that degassing work has been performed. Then, the correction unit 134 corrects the amount of gas in oil after the later inspection of the two consecutive inspection points (step S224). If it is determined that the amount of gas in oil has not decreased by more than the threshold value after the processing of step S224 or in the processing of step S222, the processing of step S104 and subsequent steps is executed.

As described above, the fifth embodiment not only achieves the same effects as the first embodiment, but also makes it possible to more appropriately predict the future state of the transformer by making corrections that maintain the continuity of the change trend of the gas in the oil, even when deaeration work is performed for maintenance or the like.

Each of the first to fifth embodiments described above may be combined with part or all of the other embodiments. Therefore, the inspection history information 122 includes, for example, at least one of the following: the inspection date of the transformer, the analysis results of the gas contained in the transformer oil, the operating time of the transformer, the overload rate of the transformer, and information obtained from a sensor permanently or temporarily attached to the transformer. The condition diagnosis devices 100, 100A to 100D use the inspection history information 122 and the like to perform the various predictions and diagnoses described above. According to each of the above-mentioned embodiments, by diagnosing the condition of the transformer based on the analysis results of the gas in the transformer insulating oil, it is possible to more appropriately provide at least one of the future inspection timing, failure timing, or lifespan.

Although several embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the gist of the invention. These embodiments and their modifications are within the scope of the invention and its equivalents as set forth in the claims, as well as the scope and gist of the invention.

100... condition diagnosis device, 110... inspection result acquisition unit, 120... storage unit, 130... pre-processing unit, 140... prediction unit, 150... condition diagnosis unit, 160... output unit, 170... correction value generation unit, 180... sensing data acquisition unit, 190... condition estimation unit

Claims (11)

a pre-processing unit that performs a predetermined pre-processing on the history information based on history information regarding past inspections of the transformer and equipment information of the transformer;
a prediction unit that predicts a change trend of the amount of gas in the transformer based on the information processed by the pre-processing unit;
a condition diagnosis unit that diagnoses at least one of a future inspection timing, a future failure timing, and a future lifespan of the transformer based on the change trend of the amount of gas in the oil of the transformer predicted by the prediction unit;
a correction value generating unit that generates a correction value based on a prediction error between a predicted value of the change trend of the amount of gas in the oil of the transformer by the prediction unit and an actual measured value of the change trend of the amount of gas in the oil based on an inspection result of the transformer ,
The condition diagnosis unit corrects the change trend of the amount of gas in the oil of the transformer predicted by the prediction unit based on the correction value generated by the correction value generation unit, and diagnoses at least one of a future inspection timing, a failure timing, or a lifespan of the transformer based on the corrected change trend.
Condition diagnostic device. The history information includes at least one of the following: an inspection date of the transformer, an analysis result of gas contained in the oil of the transformer, an operating time of the transformer, an overload rate of the transformer, and information acquired from a sensor permanently or temporarily attached to the transformer;
The equipment information includes a manufacturer type and a voltage capacity of the transformer.
The condition diagnosis device according to claim 1 . The prediction unit predicts a change trend of the amount of gas in the oil of the transformer using a plurality of prediction methods,
The condition diagnosis unit diagnoses at least one of a future inspection timing, a future failure timing, and a future lifespan of the transformer based on each prediction result predicted by the plurality of prediction methods.
The condition diagnosis device according to claim 1 or 2. At least one of the plurality of prediction methods includes a prediction method based on one or both of a physical aspect and a chemical aspect of the gas in oil of the transformer;
The condition diagnosis device according to claim 3. a sensing data acquisition unit that acquires data detected by one or more sensors provided in the transformer;
A state estimation unit that estimates a state of the transformer based on the sensing data acquired by the sensing data acquisition unit,
The state diagnosis unit diagnoses a state of the transformer based on a change trend of the amount of gas in the oil of the transformer predicted by the prediction unit and an estimation result estimated by the state estimation unit.
The condition diagnosis device according to any one of claims 1 to 4 . The sensing data includes at least one of a sound output from the transformer, a vibration of the transformer, or a temperature of the transformer.
The condition diagnosis device according to claim 5 . The pre-processing unit corrects or removes the amount of gas in the oil that is determined to be inappropriate based on the trend of changes in the amount of gas in the oil of the transformer from the past to the present.
The condition diagnosis device according to any one of claims 1 to 6 . The pre-processing unit corrects or removes the amount of gas in the oil as an outlier when the amount of gas in the oil of the transformer included in the history information is not within a predetermined range.
The condition diagnosis device according to claim 7 . The pre-processing unit compares the amount of gas in the oil of the transformer at two consecutive inspection points, and if the amount of gas in the oil of the transformer at the later inspection point has decreased by a threshold value or more from the previous inspection point, determines that degassing work of the transformer has been performed, and corrects the amount of gas in the oil of the transformer at the later inspection point based on the change trend of the amount of gas in the oil up to the previous inspection point.
9. A condition diagnosis device according to claim 7 or 8 . The prediction unit predicts a change trend of the amount of gas in the oil of the transformer by switching a prediction method of the amount of gas in the oil of the transformer or by combining a plurality of prediction methods based on one or both of a voltage class and a load condition of the transformer to be diagnosed.
The condition diagnosis device according to any one of claims 1 to 9 . The condition diagnostic device is
performing predetermined pre-processing on historical information regarding past inspections of the transformer and equipment information of the transformer;
predicting a change trend of the amount of gas in the transformer based on the pre-processed information;
Diagnosing at least one of a future inspection timing, a future failure timing, and a future lifespan of the transformer based on the predicted change trend of the amount of gas in the oil of the transformer;
generating a correction value based on a prediction error between a predicted value of the change trend of the amount of gas dissolved in oil of the transformer and an actual measured value of the change trend of the amount of gas dissolved in oil based on an inspection result of the transformer;
correcting a change trend of the amount of gas in the oil of the transformer based on the generated correction value, and diagnosing at least one of a future inspection timing, a future failure timing, and a lifespan of the transformer based on the corrected change trend.
How to diagnose the condition.

Images