On-load tap changer contact overheating fault diagnosis method and system

Abstract

The application discloses a kind of on-load tap-changer contact overheating fault diagnosis method and system, it is related to on-load tap-changer technical field, to solve the technical problem that existing on-load tap-changer contact overheating fault diagnosis method accuracy is poor, comprising the following steps: S1, historical data acquisition;Several groups of historical contact temperature curves of the contact to be diagnosed are collected, and each group of historical contact temperature curve corresponds to one kind of historical tap position change situation S2, characteristic temperature interval division;Each group of historical contact temperature curve is divided into multiple characteristic temperature intervals;S3, binary classifier training;For each kind of historical tap position change situation, based on the characteristic temperature interval under this kind of historical tap position change situation, binary classifier is trained, and binary classifier is used to determine whether the current characteristic temperature interval of the contact temperature curve to be diagnosed belongs to abnormal characteristic temperature interval;S4, curve to be diagnosed analysis.The application has the advantages of improving the accuracy of on-load tap-changer contact overheating fault diagnosis.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of on-load tap changer, more particularly, to a method and system for diagnosing overheat fault of on-load tap changer contact. BACKGROUND

[0002] As an important component of transformers in power systems, on-load tap changer plays a role in switching tap under load, thereby adjusting the output voltage of the transformer and ensuring the stable operation of the power system. However, in actual operation, on-load tap changer contacts are prone to overheat faults, which may cause contact burnout, insulation aging, and even power outage accidents, seriously affecting the reliability and safety of the power system.

[0003] Currently, there are mainly the following methods for diagnosing overheat faults of on-load tap changer contacts:

[0004] 1. Infrared temperature measurement method: detect the temperature distribution on the surface of the contact by using an infrared thermal imager to determine whether the contact is overheating. This method can intuitively show the temperature of the contact, can realize non-contact measurement, and is relatively simple to operate.

[0005] 2. Oil dissolved gas analysis method: analyze the composition and content of dissolved gases in transformer oil, and determine whether the contact has an overheat fault according to the change characteristics of the gases. This method can reflect the internal fault condition of the contact to some extent, and has good diagnostic effect on early latent faults.

[0006] 3. Electrical parameter monitoring method: monitor the electrical parameters such as current and voltage of the on-load tap changer, and determine the working state of the contact according to the changes of the parameters. This method can obtain electrical information in real time, and is convenient for dynamic monitoring of the operating state of the switch.

[0007] Although the above methods have played a certain role in diagnosing overheat faults of on-load tap changer contacts, there are still some shortcomings:

[0008] Limited accuracy: the infrared temperature measurement method is easily affected by environmental factors such as light and shielding, and the measurement results may have deviations; the oil dissolved gas analysis method can only indirectly reflect the fault condition of the contact, and cannot accurately determine the specific location and severity of the fault; the electrical parameter monitoring method may not be sensitive to some slight overheat faults, and may miss some cases.

[0009] In view of this, we propose a method and system for diagnosing overheat faults of on-load tap changer contacts. SUMMARY

[0010] The application aims to provide a load tap changer contact overheating fault diagnosis method and system to solve the technical problem of poor accuracy of existing load tap changer contact overheating fault diagnosis methods.

[0011] To solve the above technical problems, the application provides the following technical solutions: a load tap changer contact overheating fault diagnosis method, comprising the following steps:

[0012] S1, historical data collection;

[0013] Collecting several groups of historical contact temperature curves of the contact to be diagnosed, each group of historical contact temperature curves corresponding to a historical tap position change condition

[0014] S2, characteristic temperature interval division;

[0015] Dividing each group of historical contact temperature curves into multiple characteristic temperature intervals;

[0016] S3, binary classifier training;

[0017] For each historical tap position change condition, training a binary classifier based on the characteristic temperature intervals under the historical tap position change condition, the binary classifier being used to determine whether the current characteristic temperature interval of the contact temperature curve to be diagnosed belongs to an abnormal characteristic temperature interval;

[0018] S4, analysis of the curve to be diagnosed;

[0019] Collecting the contact temperature curve to be diagnosed under the tap position change condition to be diagnosed, dividing the contact temperature curve to be diagnosed, and obtaining multiple characteristic temperature intervals;

[0020] Inputting each characteristic temperature interval into the corresponding binary classifier, the binary classifier outputting the binary classifier being marked as an abnormal contact temperature curve, obtaining several contact temperature curves to be diagnosed being marked as abnormal contact temperature curves, and the contact temperature curve to be diagnosed being diagnosed as an abnormal contact temperature curve.

[0021] Preferably, the specific process of the binary classifier training comprises:

[0022] S31, curve classification;

[0023] Based on each characteristic temperature interval of the historical temperature curve of the contact to be diagnosed, obtaining the characteristic temperature intervals corresponding to several historical tap position change conditions, and based on the characteristic temperature intervals corresponding to the several historical tap position change conditions, classifying the historical contact temperature curve into an abnormal contact temperature curve and a normal contact temperature curve;

[0024] S32, dimension expansion;

[0025] The multi-dimensional extension is performed on each feature temperature interval to obtain a plurality of new feature temperature intervals;

[0026] S33, feature extraction;

[0027] Based on the normal contact temperature curve and the abnormal contact temperature curve, sample features of each new feature temperature interval are extracted;

[0028] S34, classifier training;

[0029] Based on the sample features, binary classifier training is performed to obtain a plurality of binary classifiers

[0030] Preferably, the step of dividing the historical contact temperature curve into the abnormal contact temperature curve and the normal contact temperature curve is specifically as follows:

[0031] S311, temperature rise coefficient calculation;

[0032] The feature temperature rise coefficient α of each feature temperature interval of the historical contact temperature curve is obtained, wherein, T start is the temperature at the start time of the feature temperature interval, T end is the temperature at the end time of the feature temperature interval, and Δt is the time length of the feature temperature interval;

[0033] S312, preliminary classification;

[0034] The feature temperature interval with the feature temperature rise coefficient greater than the first threshold value α1 is marked as the abnormal contact temperature curve, and the feature temperature interval with the feature temperature rise coefficient less than the second threshold value α2 is marked as the normal contact temperature curve, wherein α2<α1;

[0035] S313, pending processing;

[0036] The feature temperature interval with the feature temperature rise coefficient equal to the second threshold value α2 is marked as the pending contact temperature curve;

[0037] S314, persistence calculation;

[0038] The persistence β of each feature temperature interval of the historical contact temperature curve is calculated, and the calculation formula is wherein, ΔT i is the temperature change value of the i-th adjacent time in the feature temperature interval, and N is the number of temperature data points in the feature temperature interval minus 1;

[0039] S314, persistence classification;

[0040] The feature temperature interval with the persistence greater than the third threshold value β1 is marked as the normal contact temperature curve, and the feature temperature interval with the persistence less than the fourth threshold value β2 is marked as the abnormal contact temperature curve, wherein β2<β1.

[0041] Preferably, the step of extending each feature temperature interval in multiple dimensions specifically comprises:

[0042] S321, dimension division;

[0043] Each feature temperature interval is divided into n dimensions, each dimension corresponding to a binary classifier, and each dimension corresponding binary classifier is used to determine whether the corresponding dimension in the feature temperature interval is normal;

[0044] A dimension of a feature temperature interval is determined by the time interval from the corresponding time in the historical tapping gear change corresponding to each data point of the feature temperature interval to the corresponding time in the historical tapping gear change corresponding to the next data point;

[0045] S322, multi-dimensional extension;

[0046] The feature temperature interval corresponding to the historical tapping gear change is extended in multiple dimensions to obtain a plurality of new feature temperature intervals, which are determined by the corresponding dimensions of each feature temperature interval and the corresponding tapping gear change of each feature temperature interval, and each two-dimensional dimension corresponds to a binary classifier, wherein, n k represents the number of dimensions divided by the kth feature temperature interval, and m is the total number of feature temperature intervals.

[0047] Preferably, the S4 further comprises: obtaining the abnormal feature temperature interval corresponding historical tapping gear change through the output results of the plurality of binary classifiers, calculating the similarity S between the to-be-diagnosed tapping gear change and each historical tapping gear change, and the calculation formula is wherein, x i is the gear value of the to-be-diagnosed tapping gear change at the i th moment, y i is the gear value of the historical tapping gear change at the i th moment, and L is the total number of moments.

[0048] The contact overheating fault type and severity corresponding to the historical tapping gear change with the highest similarity to the to-be-diagnosed tapping gear change and the corresponding abnormal feature temperature interval are output.

[0049] A load-tapping switch contact overheating fault diagnosis system, comprising a server, the server comprising a memory, a processor, and executable instructions stored on the memory and executable on the processor.

[0050] Preferably, it further comprises:

[0051] The acquisition module is used for acquiring a plurality of groups of historical contact temperature curves of the to-be-diagnosed contact, and acquiring a to-be-diagnosed contact temperature curve under a to-be-diagnosed tapping gear change;

[0052] The feature temperature interval division module is configured to divide the historical contact temperature curve and the contact temperature curve to be diagnosed into a plurality of feature temperature intervals.

[0053] The classifier training module is configured to train a binary classifier based on the feature temperature intervals under each historical tap position change condition, to obtain a plurality of binary classifiers, and the binary classifier is configured to determine whether the current feature temperature interval of the contact temperature curve to be diagnosed belongs to an abnormal feature temperature interval.

[0054] The diagnosis module is configured to input each feature temperature interval of the contact temperature curve to be diagnosed into the plurality of binary classifiers, and the binary classifier outputting an abnormal feature temperature interval is marked as an abnormal contact temperature curve, and the abnormal contact temperature curve is the abnormal contact temperature curve.

[0055] Compared with the prior art, the present application has the following advantages:

[0056] 1. The present application can fully consider the influence of tap position change on contact temperature by collecting a plurality of historical contact temperature curves of the contact to be diagnosed and training a binary classifier for each historical tap position change condition, thereby more accurately determining whether the feature temperature interval of the contact temperature curve to be diagnosed is abnormal, and effectively improving the accuracy of fault diagnosis.

[0057] 2. In the dimension expansion step, each feature temperature interval is expanded in multiple dimensions to obtain a plurality of new feature temperature intervals. This multi-dimensional analysis method can more comprehensively reflect the working state and heating condition of the contact, overcoming the limitations of the prior art which only diagnoses from a single angle. By comprehensively considering information of different dimensions, abnormal feature temperature intervals can be more accurately identified, and the reliability of fault diagnosis can be improved.

[0058] 3. The present application obtains the corresponding historical tap position change condition of the abnormal feature temperature interval through the output results of the plurality of binary classifiers, and outputs the historical tap position change condition with the highest similarity to the tap position change condition to be diagnosed, the corresponding contact overheating fault type and severity of the abnormal feature temperature interval. This enables the operation and maintenance personnel to quickly and accurately understand the specific situation of the fault and take targeted maintenance measures in a timely manner, reducing the impact of the fault on the power system. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 The figure is a schematic diagram of the system framework of the present application. DETAILED DESCRIPTION

[0060] For the person skilled in the art to understand the technical scheme of the present application, the technical scheme of the present application will be further described in conjunction with the drawings of the specification.

[0061] In an embodiment, the present application provides a load tap changer contact overheating fault diagnosis method, comprising the following steps:

[0062] S1, historical data collection;

[0063] Collecting a plurality of sets of historical contact temperature curves of the contact to be diagnosed, each set of historical contact temperature curves corresponding to a historical tap position change condition

[0064] S2, characteristic temperature interval division;

[0065] Dividing each set of historical contact temperature curves into a plurality of characteristic temperature intervals;

[0066] S3, binary classifier training;

[0067] For each historical tap position change condition, a binary classifier is trained based on the characteristic temperature intervals under the historical tap position change condition, and the binary classifier is used to determine whether the current characteristic temperature interval of the contact temperature curve to be diagnosed belongs to an abnormal characteristic temperature interval;

[0068] S4, analysis of the curve to be diagnosed;

[0069] Collecting the contact temperature curve to be diagnosed under the tap position change condition to be diagnosed, dividing the contact temperature curve to be diagnosed, and obtaining a plurality of characteristic temperature intervals;

[0070] Each characteristic temperature interval is input into the corresponding binary classifier, and the binary classifier output is that the binary classifier outputting an abnormal characteristic temperature interval is marked as an abnormal contact temperature curve, a plurality of contact temperature curves to be diagnosed marked as abnormal contact temperature curves are obtained, and the contact temperature curve to be diagnosed of the abnormal contact temperature curve is diagnosed as an abnormal contact temperature curve.

[0071] In an embodiment of the present application, the specific process of binary classifier training comprises:

[0072] S31, curve classification;

[0073] Based on each characteristic temperature interval of the historical temperature curve to be diagnosed, a plurality of characteristic temperature intervals corresponding to the historical tap position change condition are obtained, and based on the plurality of characteristic temperature intervals corresponding to the historical tap position change condition, the historical contact temperature curve is divided into an abnormal contact temperature curve and a normal contact temperature curve;

[0074] S32, dimension expansion;

[0075] The characteristic temperature intervals are extended in multiple dimensions to obtain a plurality of new characteristic temperature intervals;

[0076] S33, feature extraction;

[0077] Based on the normal contact temperature curve and the abnormal contact temperature curve, sample features of the new characteristic temperature intervals are extracted;

[0078] S34, classifier training;

[0079] Based on the sample features, binary classifier training is performed to obtain a plurality of binary classifiers

[0080] In the embodiment of the application, the step of dividing the historical contact temperature curve into the abnormal contact temperature curve and the normal contact temperature curve is specifically as follows:

[0081] S311, temperature rise coefficient calculation;

[0082] The feature temperature rise coefficient α of each characteristic temperature interval of the historical contact temperature curve is obtained, Wherein, T start is the temperature at the start time of the characteristic temperature interval, T end is the temperature at the end time of the characteristic temperature interval, and Δt is the time length of the characteristic temperature interval;

[0083] S312, preliminary classification;

[0084] The characteristic temperature interval with the feature temperature rise coefficient greater than the first threshold α1 is marked as the abnormal contact temperature curve, and the characteristic temperature interval with the feature temperature rise coefficient less than the second threshold α2 is marked as the normal contact temperature curve, wherein α2<α1;

[0085] S313, pending processing;

[0086] The characteristic temperature interval with the feature temperature rise coefficient equal to the second threshold α2 is marked as the pending contact temperature curve;

[0087] S314, persistence calculation;

[0088] The persistence β of each characteristic temperature interval of the historical contact temperature curve is calculated, and the calculation formula is Wherein, ΔT i is the temperature change value of the i-th adjacent time in the characteristic temperature interval, and N is the number of temperature data points in the characteristic temperature interval minus 1;

[0089] S314, persistence classification;

[0090] The characteristic temperature interval with the persistence greater than the third threshold β1 is marked as the normal contact temperature curve, and the characteristic temperature interval with the persistence less than the fourth threshold β2 is marked as the abnormal contact temperature curve, wherein β2<β1.

[0091] In an embodiment of the present invention, the step of multidimensionally expanding each characteristic temperature range specifically includes:

[0092] S321, Dimensional Division;

[0093] Each feature temperature range is divided into m dimensions, and each dimension corresponds to a binary classifier. The binary classifier for each dimension is used to determine whether the corresponding dimension within the feature temperature range is normal.

[0094] One dimension of the characteristic temperature range is determined by the time interval between the corresponding moment in the historical tap position change of each data point in the corresponding characteristic temperature range and the corresponding moment in the historical tap position change of the next data point.

[0095] S322, Multi-dimensional Expansion;

[0096] The characteristic temperature ranges corresponding to historical tap position changes are expanded in multiple dimensions to obtain several new characteristic temperature ranges. Each new characteristic temperature range is determined by its corresponding dimension and the tap position changes. One two-dimensional dimension corresponds to one binary classifier. n k This represents the number of dimensions used to divide the k-th characteristic temperature interval, where m is the total number of characteristic temperature intervals.

[0097] In an embodiment of the present invention, S4 further includes: obtaining the historical tap position changes corresponding to the abnormal characteristic temperature range through the output results of multiple binary classifiers, and calculating the similarity S between the tap position change to be diagnosed and the historical tap position changes, using the following formula: Where, x i Let y be the gear value at time i, representing the gear position change to be diagnosed. i Let L be the gear value at time i, representing the historical gear shift changes, and L be the total number of time moments.

[0098] The system will output the historical tap position change data with the highest similarity to the tap position change data to be diagnosed, along with the corresponding abnormal temperature range and the type and severity of the contact overheating fault.

[0099] Example 2, as follows Figure 1 As shown, an on-load tap changer contact overheating fault diagnosis system includes a server, which includes a memory, a processor, and executable instructions stored in the memory and executable on the processor.

[0100] In the embodiment of the present application, the intelligent inspection device is further included, and the intelligent inspection device comprises a sensor, a sampling circuit and a microprocessor; the sensor is installed on the on-load tap changer contact, and the sensor is connected with the microprocessor through the sampling circuit; the sensor is used for collecting the current, temperature and tap position of the on-load tap changer, the sampling circuit is used for digitizing the current, temperature and tap position information, and the microprocessor sends the tap position, current, temperature and time information of the tap changer to the upper computer; the upper computer comprises a memory and a processor, the memory is used for storing the collected historical contact temperature curve, and the processor is used for dividing each historical contact temperature curve into a plurality of characteristic temperature intervals, training a plurality of binary classifiers, and the upper computer further comprises an on-load tap changer contact overheating fault diagnosis system.

[0101] In the embodiment of the present application, the intelligent inspection device is further included, and the intelligent inspection device comprises a sensor, a sampling circuit and a microprocessor; the sensor is installed on the on-load tap changer contact, and the sensor is connected with the microprocessor through the sampling circuit; the sensor is used for collecting the current, temperature and tap position of the on-load tap changer, the sampling circuit is used for digitizing the current, temperature and tap position information, and the microprocessor sends the tap position, current, temperature and time information of the tap changer to the upper computer; the upper computer comprises a memory and a processor, the memory is used for storing the collected historical contact temperature curve, and the processor is used for dividing each historical contact temperature curve into a plurality of characteristic temperature intervals, training a plurality of binary classifiers, and the upper computer further comprises an on-load tap changer contact overheating fault diagnosis system.

[0102] The acquisition module is used for acquiring a plurality of groups of historical contact temperature curves of the contact to be diagnosed and acquiring a contact temperature curve of the contact to be diagnosed under a tap position change condition to be diagnosed;

[0103] The characteristic temperature interval division module is used for dividing the historical contact temperature curve and the contact temperature curve to be diagnosed into a plurality of characteristic temperature intervals;

[0104] The classifier training module is used for training a binary classifier based on the characteristic temperature interval under each historical tap position change condition, so as to obtain a plurality of binary classifiers, and the binary classifier is used for judging whether a current characteristic temperature interval of the contact temperature curve to be diagnosed belongs to an abnormal characteristic temperature interval;

[0105] The diagnosis module is used for inputting each characteristic temperature interval of the contact temperature curve to be diagnosed into the plurality of binary classifiers, and the binary classifier output is that the binary classifier outputting an abnormal characteristic temperature interval is marked as an abnormal contact temperature curve, and the abnormal contact temperature curve is the abnormal contact temperature curve.

[0106] The embodiment of the present application discloses a preferred embodiment, but is not limited thereto, and a person skilled in the art can easily understand the spirit of the present application and make different inferences and changes according to the above-mentioned embodiment, as long as the spirit of the present application is not deviated, which is within the protection scope of the present application.

Claims

1. A method of diagnosing an overheat fault of a load tap changer contact, characterized in that, The method comprises the following steps: S1, historical data collection; S2, characteristic temperature interval division; S3, binary classifier training; S31, curve classification; S311, temperature rise coefficient calculation; S312, preliminary classification; S313, pending processing; S314, persistence calculation; S315, persistent classification; S32, dimension expansion; obtaining a feature temperature rise coefficient of each feature temperature interval of the historical contact temperature curve α , wherein, T start is a temperature at a start time of the feature temperature interval, T end is a temperature at an end time of the feature temperature interval, is a time length of the feature temperature interval; S33, feature extraction; characteristic temperature interval of the characteristic temperature curve of the first threshold value α 1 is marked as an abnormal contact temperature curve, and the characteristic temperature interval of the characteristic temperature curve of the second threshold value α 2 is marked as a normal contact temperature curve, wherein, α 2 < 1 α 1; S34, classifier training; Set the characteristic temperature rise coefficient equal to the second threshold. α The characteristic temperature range of 2 is marked as the contact temperature curve to be determined; S4, analysis of the curve to be diagnosed; Duration of each characteristic temperature interval of the calculated historical contact temperature curve β , the calculation formula is , wherein, is the temperature change value of the i adjacent moment in the characteristic temperature interval, N is the number of temperature data points in the characteristic temperature interval minus 1; The step of expanding each characteristic temperature interval in multiple dimensions comprises: characteristic temperature interval of the temperature curve of the contact is marked as a normal contact temperature curve, and the persistence is less than a fourth threshold β 1 is marked as a normal contact temperature curve, and the persistence is less than a fourth threshold β 2 is marked as an abnormal contact temperature curve, wherein, β 2 < 1 β 1; S321, dimension division; S322, multi-dimensional expansion; The server comprises a memory, a processor, and executable instructions stored on the memory and executable on the processor. The method further comprises: A collection module for collecting multiple groups of historical contact temperature curves of the contact to be diagnosed and collecting a contact temperature curve of the contact to be diagnosed under the tap position change condition to be diagnosed; A characteristic temperature interval division module for dividing the historical contact temperature curves and the contact temperature curve to be diagnosed into multiple characteristic temperature intervals; ​ ​ ​ 2. The method of claim 1, wherein the method further comprises: ​ ​ each feature temperature interval is divided into dimensions according to the dimensions, each dimension corresponds to a binary classifier, and each binary classifier corresponding to the dimension is used to determine whether the corresponding dimension in the feature temperature interval is normal; ​ ​ The characteristic temperature intervals corresponding to the historical tapping gear change situations are extended in multiple dimensions to obtain a plurality of new characteristic temperature intervals, the new characteristic temperature intervals are determined by the dimension corresponding to each characteristic temperature interval and the tapping gear change situation corresponding to each characteristic temperature interval, one two-dimension corresponds to one two-classifier, wherein, , represents the sum of the dimensions in all characteristic temperature intervals, n k represents the number of dimensions divided by the i-th characteristic temperature interval, k m is the total number of characteristic temperature intervals.​ 3. A method of diagnosing overheat fault of a load tap changer contact according to claim 2, characterized in that, S4 further includes: obtaining the historical tap position changes corresponding to the abnormal characteristic temperature range through the output results of multiple binary classifiers, and calculating the similarity between the tap position change to be diagnosed and the changes of each historical tap position. S The calculation formula is: ,in, x i The change in the gear position to be diagnosed is in the first... i The gear value at any given moment. y i The historical gear shift changes are shown in the first... i The gear value at any given moment. L This represents the total number of time points. ​ 4. A load tap changer contact overheating fault diagnostic system for use in the load tap changer contact overheating fault diagnostic method as claimed in claim 3, characterized by, ​ 5. A diagnostic system for overheat fault detection of a load tap changer contact according to claim 4, characterized in that, ​ ​ ​ The classifier training module is configured to train a binary classifier based on the feature temperature interval under each historical tapping gear change condition, to obtain a plurality of binary classifiers, and the binary classifier is configured to determine whether the current feature temperature interval of the to-be-diagnosed contact temperature curve belongs to an abnormal feature temperature interval. The diagnosis module is configured to input each feature temperature interval of the to-be-diagnosed contact temperature curve into the plurality of binary classifiers, and the binary classifier outputting an abnormal feature temperature interval is marked as an abnormal contact temperature curve.

Images