Transformer partial discharge mode identification method and system based on multi-source data fusion

Abstract

The invention discloses a transformer partial discharge mode identification method and system based on multi-source data fusion, and relates to the technical field of transformer maintenance, the transformer partial discharge mode identification system comprises an identification execution module, a multi-source data fusion module and a discharge identification module, the identification execution module is used for collecting ultrahigh frequency electromagnetic wave, ultrasonic wave and high frequency pulse current partial discharge signals; and the multi-source data fusion module is used for carrying out unified time reference and intensity normalization processing on a detection period and a sampling resolution ratio, carrying out dynamic adjustment on two output parameter characteristics according to a subsequent detection result, carrying out multi-source data fusion processing, and carrying out multi-source data fusion processing on the two output parameter characteristics. The discharge identification module is used for comprehensively judging a partial discharge mode according to the processed data, and the identification execution module comprises a UHF detection unit, an AE detection unit, an HFCT detection unit, a gas component detection unit and an ultrahigh frequency electromagnetic wave analysis module.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer maintenance, and in particular to a transformer partial discharge pattern recognition method and system based on multi-source data fusion. Background Art

[0002] In addition to electrical measurement integrated into the transformer itself, transformer partial discharge detection methods also include external sensor methods such as ultra-high frequency electromagnetic wave (UHF) detection, ultrasonic (AE) detection, and high-frequency pulse current (HFCT) detection. These methods, combined with dissolved gas analysis for auxiliary judgment, can be used to detect common discharge types such as corona discharge, metal tip discharge, surface discharge, gap discharge, and oil discharge.

[0003] Existing technologies often use a single discharge detection method for one or more discharge types. This single basis is susceptible to noise interference, resulting in a certain misjudgment rate. Directly using data fusion processing, the lack of a unified acquisition and modeling architecture can easily cause data distortion during normalization, as each sensor has a significantly different detection cycle and sampling resolution when detecting different discharge modes. Forcibly unifying the time base and intensity normalization processing can cause the detection cycle and sampling resolution to deviate from normal acquisition requirements. Therefore, it is necessary to design a transformer partial discharge pattern recognition method and system based on unified modeling and multi-source data fusion. Summary of the Invention

[0004] The object of the present invention is to provide a transformer partial discharge pattern recognition method and system based on multi-source data fusion to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a transformer partial discharge pattern recognition system based on multi-source data fusion, comprising an identification execution module, a multi-source data fusion module, and a discharge recognition module. The identification execution module is used to collect ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals, and perform multi-dimensional detection in combination with the gas content in the oil. The multi-source data fusion module is used to unify the time reference and intensity normalization processing of the detection period and sampling resolution, dynamically adjust the two output parameter characteristics according to subsequent detection results, and perform multi-source data fusion processing. The discharge recognition module is used to comprehensively determine the partial discharge pattern based on the processed data.

[0006] According to the above technical solution, the identification execution module includes a UHF detection unit, an AE detection unit, an HFCT detection unit, a gas composition detection unit, an ultra-high frequency electromagnetic wave analysis module, an ultrasonic analysis module, a high-frequency pulse current analysis module, and a gas composition analysis module. The UHF detection unit is electrically connected to the ultra-high frequency electromagnetic wave analysis module, the AE detection unit is electrically connected to the ultrasonic analysis module, the HFCT detection unit is electrically connected to the high-frequency pulse current analysis module, and the gas composition detection unit is electrically connected to the gas composition analysis module. The UHF detection unit, the AE detection unit, and the HFCT detection unit are respectively used to collect ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals. The gas composition detection unit is used to detect the gas components precipitated in the insulating oil. The ultra-high frequency electromagnetic wave analysis module, the ultrasonic analysis module, and the high-frequency pulse current analysis module are respectively used to analyze partial discharge signals. The gas composition analysis module is used to analyze the components of various gases. The multi-source data fusion module includes a detection period adjustment module, a sampling resolution adjustment module, a data alignment processing module, a discharge type corresponding module, and a timing module. The detection period adjustment module and the sampling resolution adjustment module are both electrically connected to the UHF detection unit, the AE detection unit, and the HFCT detection unit. The data alignment processing module and the discharge type corresponding module are both electrically connected to the UHF detection unit, the AE detection unit, and the HFCT detection unit. The timing module is used to count the duration after a certain discharge form occurs. The detection period adjustment module and the sampling resolution adjustment module are respectively used to adjust the detection period and sampling resolution of each detection unit. The data alignment processing module is used to normalize the sampling period and sampling resolution of each detection unit. The discharge identification module includes a gas-assisted judgment module and a discharge mode judgment module. The timing module is electrically connected to the data alignment processing module. The gas-assisted judgment module is electrically connected to the discharge mode judgment module. The gas-assisted judgment module is used to analyze the changes in the gas composition precipitated from the transformer insulating oil to assist in judging the discharge mode. The discharge mode judgment module is used to judge the discharge mode with the highest probability.

[0007] The transformer partial discharge pattern recognition method based on multi-source data fusion includes the following steps: S0. Place UHF detection units on the oil valve and inspection hole on the side wall of the transformer's main oil tank. Place AE detection units in the middle section of the tank wall near the high-voltage bushing outlet and at the same height as the winding. Install HFCT detection units on the grounding loop. Install a gas composition detection unit on the inner wall of the transformer's main oil tank. Clearly identify which detection method corresponds to each discharge mode. S1. Each detection unit operates and detects partial discharge signals. Each detection unit outputs a detection result at an initial detection period and sampling resolution, and uses a corresponding analysis module to detect whether partial discharge occurs. S2. When a detection unit detects a certain type of discharge, the detection units of other types that are closest to the current detection unit are instructed to adjust their detection cycles and sampling resolutions so that the detection cycles and sampling resolutions of the detection units that need to be adjusted are consistent with those of the current detection unit. S3. As time goes on, based on whether the same partial discharge signal continues to be detected, dynamically adjusting the detection period and sampling resolution of the detection unit that needs to be adjusted; S4. When the detection period and sampling resolution of each detection unit are not uniform, the detection data is processed by the data alignment processing module to normalize the detection period and sampling resolution; S5. Determine the partial discharge pattern based on the analysis results of the gas components released from the transformer oil.

[0008] According to the above technical solution, in S0, it is clarified which detection method corresponds to each discharge mode: S0-1, gap discharge will generate high-frequency electromagnetic pulses, local breakdown leading to acoustic emission, and pulse charge, which can be detected by UHF, AE, and HFCT. Among them, UHF has the highest sensitivity and is used as the main judgment method, while AE and HFCT are used as auxiliary judgment methods; S0-2, metal tip discharge and corona discharge will generate high-frequency electromagnetic radiation and pulse current signals, which can be effectively detected by UHF and HFCT, with HFCT as the dominant judgment method and UHF as the auxiliary judgment method; S0-3: When surface discharge develops along the surface of the insulation, it will generate mechanical shock waves and pulse currents. Both AE and HFCT can detect them. UHF can also capture them when they are strong. AE is used as the main judgment method, and HFCT and UHF are used as auxiliary judgment methods. S0-4. Discharge in oil will produce obvious bubble oscillation and acoustic wave impact. AE is more sensitive and is used as the dominant judgment method, while UHF is used as an auxiliary judgment method.

[0009] According to the above technical solution, in S2, the detection period and sampling resolution are unified as follows: S2-1, each detection unit detects the discharge mode corresponding to the dominant judgment mode by default, and uses its default detection cycle and sampling resolution Output the test results, The number of detection units is as follows: when a detection unit detects a certain pattern of partial discharge signal that it is responsible for, the detection unit corresponding to the partial discharge signal of this pattern, which is used as an auxiliary judgment mode, needs to be adjusted; S2-2, Order No. The detection unit detects the partial discharge signal, and its default detection cycle is , the default sampling resolution is , then the initial probability of occurrence of partial discharge signal of this mode is , at this time, you need to The detection cycle and sampling resolution of each detection unit are adjusted. Before adjustment, the default detection cycle and sampling resolution are and , so that the adjusted detection cycle , the adjusted sampling resolution , making the and The detection cycle and sampling resolution of each detection unit are adjusted to be consistent.

[0010] According to the above technical solution, in S3, the dynamic adjustment is specifically as follows: S3-1, Order No. The detection unit maintains the detection cycle and sampling resolution continued time period, if During the time period When the first detection unit detects the same pattern of partial discharge signal again, observe the Whether the detection units detect the same pattern of partial discharge signals at the same time, if detected, the probability of the occurrence of partial discharge signals of this pattern is increased, and the adjusted probability is ,in For the When the first detection unit is used as an auxiliary judgment method, The probability increment brought by the detection of the same pattern of partial discharge signal by the detection unit is If the detection units do not detect the same pattern of partial discharge signals at the same time, ; S3-2, if During the time period When the first detection unit does not detect the partial discharge signal of the same mode, the The detection cycle and sampling resolution of each detection unit are determined by and to its initial value and Gradually recover, so that the distance The time elapsed after the time period ends is , then the detection cycle at this time and sampling resolution The calculation formulas are: hour, ,when hour, ,when hour, ,when hour, ,in 、 is the time conversion factor until and Adjust to and No further adjustments will be made afterwards.

[0011] According to the above technical solution, in S4, the normalization process of the detection period and the sampling resolution is specifically as follows: S4-1. First, At the end of the time period, and The first sampling time of the two detection units is aligned. At this time, the two detection units sample at the same time. The next sampling only collects the data that meets the and The detection data of the lowest common multiple of the time will not be collected; other data will not be collected; S4-2, the and The maximum and minimum values ​​of the detection data of each detection unit are mapped to The dimensional difference of sampling resolution between different detection units is eliminated within the range of , so that the output signals of the two can be compared and fused at a unified scale.

[0012] According to the above technical solution, in S5, the mode of partial discharge is determined as follows: before each detection unit detects a partial discharge phenomenon, the gas content detected by the gas component detection unit is respectively ,in The number of gas types involved in partial discharge. When a certain pattern of partial discharge is detected, if this discharge pattern will lead to changes, and , actually detected The probability of this partial discharge mode occurring is ,in Is the gas increment influence coefficient, if not detected Incremental changes , combined with S3-1 The final probability is calculated when hour, is the probability judgment threshold, which determines that this discharge mode is established and needs to be repaired.

[0013] Compared with the existing technology, the beneficial effects achieved by the present invention are: the present invention collects ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals, combines them with the gas content in the oil for multi-dimensional detection, and introduces an attention mechanism. When a certain detection method detects a partial discharge signal, the detection cycle and sampling resolution of other corresponding detection methods are adjusted, and a unified time base and intensity normalization processing are performed. The two output parameter characteristics are dynamically adjusted according to subsequent detection results, so that the detection cycle and sampling resolution can be optimized within a suitable time period without affecting the normal acquisition work of various detection methods as much as possible. The combined judgment of multiple detection methods can effectively reduce the misjudgment rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings: Figure 1 It is a schematic diagram of the overall module structure of the present invention. DETAILED DESCRIPTION

[0015] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0016] See also Figure 1 The present invention provides a technical solution: a transformer partial discharge pattern recognition system based on multi-source data fusion, including an identification execution module, a multi-source data fusion module, and a discharge recognition module. The identification execution module is used to collect ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals, and perform multi-dimensional detection in combination with the gas content in the oil. The multi-source data fusion module is used to perform unified time base and intensity normalization processing on the detection period and sampling resolution, dynamically adjust the two output parameter characteristics according to subsequent detection results, and perform multi-source data fusion processing. The discharge recognition module is used to comprehensively determine the partial discharge pattern based on the processed data; The three discharge modes overlap with each other, and the discharge mode matching each detection unit is determined based on the detection sensitivity and accuracy obtained from a large number of experiments.

[0017] The identification execution module includes a UHF detection unit, an AE detection unit, an HFCT detection unit, a gas composition detection unit, an ultra-high frequency electromagnetic wave analysis module, an ultrasonic analysis module, a high-frequency pulse current analysis module, and a gas composition analysis module. The UHF detection unit is electrically connected to the ultra-high frequency electromagnetic wave analysis module, the AE detection unit is electrically connected to the ultrasonic analysis module, the HFCT detection unit is electrically connected to the high-frequency pulse current analysis module, and the gas composition detection unit is electrically connected to the gas composition analysis module. The UHF detection unit, the AE detection unit, and the HFCT detection unit are respectively used to collect ultra-high frequency electromagnetic wave, ultrasonic wave, and high-frequency pulse current partial discharge signals. The gas composition detection unit is used to detect the gas components precipitated in the insulating oil. The ultra-high frequency electromagnetic wave analysis module, the ultrasonic analysis module, and the high-frequency pulse current analysis module are respectively used to analyze partial discharge signals. The gas composition analysis module is used to analyze the components of various gases. The multi-source data fusion module includes a detection cycle adjustment module, a sampling resolution adjustment module, a data alignment processing module, a discharge type corresponding module, and a timing module. The detection cycle adjustment module and the sampling resolution adjustment module are electrically connected to the UHF detection unit, the AE detection unit, and the HFCT detection unit. The data alignment processing module and the discharge type corresponding module are electrically connected to the UHF detection unit, the AE detection unit, and the HFCT detection unit. The timing module is used to count the duration after a certain discharge form occurs. The detection cycle adjustment module and the sampling resolution adjustment module are used to adjust the detection cycle and sampling resolution of each detection unit, respectively. The data alignment processing module is used to normalize the sampling cycle and sampling resolution of each detection unit. The discharge identification module includes a gas-assisted judgment module and a discharge mode judgment module. The timing module is electrically connected to the data alignment processing module, and the gas-assisted judgment module is electrically connected to the discharge mode judgment module. The gas-assisted judgment module is used to analyze the changes in the gas composition of the transformer insulating oil to assist in judging the discharge mode. The discharge mode judgment module is used to determine the discharge mode with the highest probability. The transformer partial discharge pattern recognition method based on multi-source data fusion includes the following steps: S0. Place UHF detection units on the oil valve and inspection hole on the side wall of the transformer's main oil tank. Place AE detection units in the middle section of the tank wall near the high-voltage bushing outlet and at the same height as the winding. Install HFCT detection units on the grounding loop. Install a gas composition detection unit on the inner wall of the transformer's main oil tank. Clearly identify which detection method corresponds to each discharge mode. S1. Each detection unit operates and detects partial discharge signals. Each detection unit outputs a detection result at an initial detection period and sampling resolution, and uses a corresponding analysis module to detect whether partial discharge occurs. S2. When a detection unit detects a certain type of discharge, the detection units of other types that are closest to the current detection unit are instructed to adjust their detection cycles and sampling resolutions so that the detection cycles and sampling resolutions of the detection units that need to be adjusted are consistent with those of the current detection unit. S3. As time goes on, based on whether the same partial discharge signal continues to be detected, dynamically adjusting the detection period and sampling resolution of the detection unit that needs to be adjusted; S4. When the detection period and sampling resolution of each detection unit are not uniform, the detection data is processed by the data alignment processing module to normalize the detection period and sampling resolution; S5. Determine the partial discharge pattern based on the analysis results of the gas components released from the transformer oil; In S0, it is clear which detection method corresponds to each discharge mode: S0-1, gap discharge will generate high-frequency electromagnetic pulses, local breakdown leading to acoustic emission, and pulse charge, which can be detected by UHF, AE, and HFCT. Among them, UHF has the highest sensitivity and is used as the main judgment method, while AE and HFCT are used as auxiliary judgment methods; S0-2, metal tip discharge and corona discharge will generate high-frequency electromagnetic radiation and pulse current signals, which can be effectively detected by UHF and HFCT, with HFCT as the dominant judgment method and UHF as the auxiliary judgment method; S0-3: When surface discharge develops along the surface of the insulation, it will generate mechanical shock waves and pulse currents. Both AE and HFCT can detect them. UHF can also capture them when they are strong. AE is used as the main judgment method, and HFCT and UHF are used as auxiliary judgment methods. S0-4, discharge in oil will produce obvious bubble oscillation and acoustic shock, AE is more sensitive as the dominant judgment method, and UHF is used as an auxiliary judgment method; In S2, the detection period and sampling resolution are unified as follows: S2-1, each detection unit detects the discharge mode corresponding to the dominant judgment mode by default, and uses its default detection cycle and sampling resolution Output the test results, The number of detection units is as follows: when a detection unit detects a certain pattern of partial discharge signal that it is responsible for, the detection unit corresponding to the partial discharge signal of this pattern, which is used as an auxiliary judgment mode, needs to be adjusted; S2-2, Order No. The detection unit detects the partial discharge signal, and its default detection cycle is , the default sampling resolution is , then the initial probability of occurrence of partial discharge signal of this mode is , at this time, you need to The detection cycle and sampling resolution of each detection unit are adjusted. Before adjustment, the default detection cycle and sampling resolution are and , so that the adjusted detection cycle , the adjusted sampling resolution , making the and The detection cycle and sampling resolution of each detection unit are adjusted to be consistent; In S3, dynamic adjustment is as follows: S3-1, Order No. The detection unit maintains the detection cycle and sampling resolution continued time period, if During the time period When the first detection unit detects the same pattern of partial discharge signal again, observe the Whether the detection units detect the same pattern of partial discharge signals at the same time, if detected, the probability of the occurrence of partial discharge signals of this pattern is increased, and the adjusted probability is ,in For the When the first detection unit is used as an auxiliary judgment method, The probability increment brought by the detection of the same pattern of partial discharge signal by the detection unit is If the detection units do not detect the same pattern of partial discharge signals at the same time, ; S3-2, if During the time period When the first detection unit does not detect the partial discharge signal of the same mode, the The detection cycle and sampling resolution of each detection unit are determined by and to its initial value and Gradually recover, so that the distance The time elapsed after the time period ends is , then the detection cycle at this time and sampling resolution The calculation formulas are: hour, ,when hour, ,when hour, ,when hour, ,in 、 is the time conversion factor until and Adjust to and No further adjustments will be made afterwards; In S4, the normalization process of the detection period and sampling resolution is specifically as follows: S4-1. First, At the end of the time period, and The first sampling time of the two detection units is aligned. At this time, the two detection units sample at the same time. The next sampling only collects the data that meets the and The detection data of the lowest common multiple of the time will not be collected; other data will not be collected; S4-2, the and The maximum and minimum values ​​of the detection data of each detection unit are mapped to Within the range of , the dimensional difference of sampling resolution between different detection units is eliminated, so that the output signals of the two can be compared and fused at a unified scale; In S5, the mode of partial discharge is determined as follows: before each detection unit detects a partial discharge phenomenon, the gas content detected by the gas component detection unit is respectively ,in The number of gas types involved in partial discharge. When a certain pattern of partial discharge is detected, if this discharge pattern will lead to changes, and , actually detected The probability of this partial discharge mode occurring is ,in is the gas increment influence coefficient. If no Incremental changes , combined with S3-1 The final probability is calculated when hour, is the probability judgment threshold, which determines that this discharge mode is established and needs to be repaired.

[0018] The present invention collects partial discharge signals of ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse currents, and performs multi-dimensional detection in combination with the gas content in oil. It also introduces an attention mechanism. When a certain detection method detects a partial discharge signal, it adjusts the detection cycle and sampling resolution of other corresponding detection methods, performs a unified time reference and intensity normalization process, and dynamically adjusts the two output parameter characteristics according to subsequent detection results. This allows the detection cycle and sampling resolution to be optimized within a suitable time period without affecting the normal acquisition work of various detection methods as much as possible. The combined judgment of multiple detection methods can effectively reduce the misjudgment rate.

[0019] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0020] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art will be able to modify the technical solutions described in the aforementioned embodiments or substitute equivalents for some of the technical features. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A transformer partial discharge pattern recognition system based on multi-source data fusion, characterized by: It includes an identification execution module, a multi-source data fusion module, and a discharge identification module. The identification execution module is used to collect ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals, and perform multi-dimensional detection in combination with the gas content in the oil. The multi-source data fusion module is used to unify the time basis and intensity normalization processing of the detection period and sampling resolution, dynamically adjust the two output parameter characteristics according to subsequent detection results, and perform multi-source data fusion processing. The discharge identification module is used to comprehensively determine the partial discharge mode based on the processed data.

2. The transformer partial discharge pattern recognition system based on multi-source data fusion according to claim 1 is characterized in that: The identification execution module includes a UHF detection unit, an AE detection unit, an HFCT detection unit, a gas composition detection unit, an ultra-high frequency electromagnetic wave analysis module, an ultrasonic analysis module, a high-frequency pulse current analysis module, and a gas composition analysis module. The UHF detection unit, the AE detection unit, and the HFCT detection unit are respectively used to collect ultra-high frequency electromagnetic waves, ultrasonic waves, and high-frequency pulse current partial discharge signals. The gas composition detection unit is used to detect the gas components precipitated in the insulating oil. The ultra-high frequency electromagnetic wave analysis module, the ultrasonic analysis module, and the high-frequency pulse current analysis module are respectively used to analyze partial discharge signals. The gas composition analysis module is used to analyze the components of various gases. The multi-source data fusion module includes a detection cycle adjustment module, a sampling resolution adjustment module, a data alignment processing module, a discharge type corresponding module, and a timing module. The timing module is used to count the duration after a certain discharge form occurs. The detection cycle adjustment module and the sampling resolution adjustment module are used to adjust the detection cycle and sampling resolution of each detection unit respectively. The data alignment processing module is used to normalize the sampling cycle and sampling resolution of each detection unit. The discharge identification module includes a gas-assisted judgment module and a discharge mode judgment module. The gas-assisted judgment module is used to analyze the changes in gas composition precipitated from the transformer insulating oil to assist in judging the discharge mode. The discharge mode judgment module is used to judge the discharge mode with the highest probability.

3. A transformer partial discharge pattern recognition method based on multi-source data fusion is characterized by: The following steps are involved: S0. Place a UHF detection unit on the oil valve and inspection hole on the side wall of the transformer's main oil tank, place an AE detection unit in the middle section of the tank wall near the high-voltage bushing lead-out side and at the same height as the winding, install an HFCT detection unit on the grounding loop, and install a gas composition detection unit on the inner wall of the transformer's main oil tank; S1. Each detection unit operates and detects partial discharge signals. Each detection unit outputs a detection result at an initial detection period and sampling resolution, and uses a corresponding analysis module to detect whether partial discharge occurs. S2. When a detection unit detects a certain type of discharge, the detection units of other types that are closest to the current detection unit are instructed to adjust their detection cycles and sampling resolutions so that the detection cycles and sampling resolutions of the detection units that need to be adjusted are consistent with those of the current detection unit. S3. As time goes on, based on whether the same partial discharge signal continues to be detected, dynamically adjusting the detection period and sampling resolution of the detection unit that needs to be adjusted; S4. When the detection period and sampling resolution of each detection unit are not uniform, the detection data is processed by the data alignment processing module to normalize the detection period and sampling resolution; S5. Determine the partial discharge pattern based on the analysis results of the gas components released from the transformer oil.

4. The transformer partial discharge pattern recognition method based on multi-source data fusion according to claim 3 is characterized in that: In S0, it is clarified which detection method corresponds to each discharge mode: S0-1, gap discharge UHF is used as the main judgment method, AE and HFCT are used as auxiliary judgment methods; S0-2, metal tip discharge and corona discharge HFCT as the main judgment method, UHF as the auxiliary judgment method; S0-3, surface discharge AE is used as the main judgment method, HFCT and UHF are used as auxiliary judgment methods; S0-4, oil discharge AE is used as the main judgment method, and UHF is used as the auxiliary judgment method.

5. The transformer partial discharge pattern recognition method based on multi-source data fusion according to claim 4 is characterized in that: In S2, the detection period and sampling resolution are unified as follows: S2-1, each detection unit detects the discharge mode corresponding to the dominant judgment mode by default, and uses its default detection cycle and sampling resolution Output the test results, is the number of detection units; S2-2, Order No. The detection unit detects the partial discharge signal, and its default detection cycle is , the default sampling resolution is , then the initial probability of occurrence of partial discharge signal of this mode is , at this time, you need to The detection cycle and sampling resolution of each detection unit are adjusted. Before adjustment, the default detection cycle and sampling resolution are and , so that the adjusted detection cycle , the adjusted sampling resolution .

6. The transformer partial discharge pattern recognition method based on multi-source data fusion according to claim 5 is characterized in that: In S3, the dynamic adjustment is specifically as follows: S3-1, Order No. The detection unit maintains the detection cycle and sampling resolution continued time period, if During the time period When the first detection unit detects the same pattern of partial discharge signal again, observe the Whether the detection units detect the same pattern of partial discharge signals at the same time, if detected, the probability of the occurrence of partial discharge signals of this pattern is increased, and the adjusted probability is ,in For the When the first detection unit is used as an auxiliary judgment method, The probability increment brought by the detection of the same pattern of partial discharge signal by the detection unit is If the detection units do not detect the same pattern of partial discharge signals at the same time, ; S3-2, if During the time period When the first detection unit does not detect the partial discharge signal of the same mode, the The detection cycle and sampling resolution of each detection unit are determined by and to its initial value and Gradually recover, so that the distance The time elapsed after the time period ends is , then the detection cycle at this time and sampling resolution The calculation formulas are: hour, ,when hour, ,when hour, ,when hour, ,in 、 is the time conversion factor.

7. The transformer partial discharge pattern recognition method based on multi-source data fusion according to claim 6 is characterized in that: In S4, the normalization process of the detection period and the sampling resolution is specifically as follows: S4-1. First, At the end of the time period, and The first sampling time of the two detection units is aligned. At this time, the two detection units sample at the same time. The next sampling only collects the data that meets the and The detection data of the lowest common multiple of the time will not be collected; other data will not be collected; S4-2, the and The maximum and minimum values ​​of the detection data of each detection unit are mapped to The dimensional difference of sampling resolution between different detection units is eliminated within the range of .

8. The transformer partial discharge pattern recognition method based on multi-source data fusion according to claim 7 is characterized in that: In the above S5, the mode of partial discharge is determined as follows: before each detection unit detects a partial discharge phenomenon, the gas content detected by the gas component detection unit is respectively ,in The number of gas types involved in partial discharge. When a certain pattern of partial discharge is detected, if this discharge pattern will lead to changes, and , actually detected The probability of this partial discharge mode occurring is ,in is the gas increment influence coefficient. If no Incremental changes , combined with S3-1 The final probability is calculated when hour, is the probability judgment threshold, which determines that this discharge mode is established and needs to be repaired.

Images