Method and system for correcting on-line measurement of partial discharge of medium voltage switchgear

Abstract

The application discloses a medium-voltage switch cabinet partial discharge online measurement correction method and system, relates to the technical field of discharge measurement, and comprises the following steps: loading a partial discharge log library to dynamically associate and adjust a partial discharge sensing double-layer structure of the medium-voltage switch cabinet, and obtaining a partial discharge sensing optimization structure; performing partial discharge online monitoring and self-adaptive gain adjustment on the medium-voltage switch cabinet according to the partial discharge sensing optimization structure to obtain a partial discharge first thermal map; performing real partial discharge correction on the partial discharge first thermal map under power frequency phase association analysis to obtain a partial discharge second thermal map; performing coupling interference correction on the partial discharge second thermal map under multi-field joint to obtain a partial discharge third thermal map; performing working condition influence dynamic bias correction on the partial discharge third thermal map to obtain a partial discharge fourth thermal map, and generating a partial discharge early warning instruction. The application solves the technical problems that the existing partial discharge online measurement result is prone to interference and lacks authenticity, and achieves the technical effects of improving the accuracy of partial discharge online measurement and the reliability of early warning.

Description

Technical Field

[0001] This invention relates to the field of discharge measurement technology, specifically to a method and system for online measurement and correction of partial discharge in medium-voltage switchgear. Background Technology

[0002] During the operation of medium-voltage switchgear, partial discharge signals are weak in amplitude and complex in pulse characteristics. They are also susceptible to various factors such as power frequency voltage phase, cabinet structure coupling, multi-physics field interference, and changes in operating conditions. The signals acquired during online monitoring often contain a large number of non-partial discharge interference components, leading to deviations in measurement results. Furthermore, fixed sensor deployment methods and single processing techniques are difficult to adapt to signal variations under different operating conditions, easily misinterpreting interference signals as partial discharges or weakening the true characteristics of partial discharges, making it difficult to guarantee the authenticity and accuracy of online partial discharge measurement results. Summary of the Invention

[0003] This application provides a method and system for correcting online partial discharge measurement in medium-voltage switchgear, which addresses the technical problems of existing online partial discharge measurement results being susceptible to interference and lacking authenticity and accuracy.

[0004] In view of the above problems, this application provides a method and system for online measurement and correction of partial discharge in medium-voltage switchgear.

[0005] The first aspect of this application provides a method for online measurement and correction of partial discharge in medium-voltage switchgear, the method comprising:

[0006] A partial discharge log library of the medium-voltage switchgear is loaded, and the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear is dynamically correlated and adjusted according to the partial discharge log library to obtain an optimized partial discharge sensing architecture. Based on the optimized partial discharge sensing architecture, online partial discharge monitoring and adaptive gain adjustment are performed on the medium-voltage switchgear to obtain a first partial discharge heatmap. Based on the real-time voltage sequence of the medium-voltage switchgear, the first partial discharge heatmap is corrected for actual partial discharge under power frequency phase correlation analysis to obtain a second partial discharge heatmap. Based on the cabinet characteristic dataset of the medium-voltage switchgear, the second partial discharge heatmap is corrected for coupling interference under multi-field joint conditions to obtain a third partial discharge heatmap. Based on the real-time operating condition data stream of the medium-voltage switchgear, the third partial discharge heatmap is dynamically biased to correct for operating condition influences to obtain a fourth partial discharge heatmap. Based on the fourth partial discharge heatmap, a partial discharge early warning command is generated.

[0007] A second aspect of this application provides an online measurement and correction system for partial discharge in medium-voltage switchgear, the system comprising:

[0008] The system includes the following modules: a dynamic correlation adjustment module for loading the partial discharge log library of the medium-voltage switchgear and dynamically adjusting the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear based on the partial discharge log library to obtain an optimized partial discharge sensing architecture; a monitoring and adjustment module for performing online partial discharge monitoring and adaptive gain adjustment of the medium-voltage switchgear based on the optimized partial discharge sensing architecture to obtain a first partial discharge heatmap; a real partial discharge correction module for performing real partial discharge correction on the first partial discharge heatmap based on the real-time voltage sequence of the medium-voltage switchgear under power frequency phase correlation analysis to obtain a second partial discharge heatmap; a coupling interference correction module for performing multi-field joint coupling interference correction on the second partial discharge heatmap based on the cabinet characteristic dataset of the medium-voltage switchgear to obtain a third partial discharge heatmap; a dynamic bias correction module for performing dynamic bias correction on the third partial discharge heatmap based on the real-time operating condition data stream of the medium-voltage switchgear to obtain a fourth partial discharge heatmap; and an instruction generation module for generating a partial discharge early warning instruction based on the fourth partial discharge heatmap.

[0009] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0010] This application loads a partial discharge log library of a medium-voltage switchgear and dynamically adjusts the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear based on the partial discharge log library to obtain an optimized partial discharge sensing architecture. Based on the optimized partial discharge sensing architecture, the application performs online partial discharge monitoring and adaptive gain adjustment of the medium-voltage switchgear to obtain a first partial discharge heatmap. Based on the real-time voltage sequence of the medium-voltage switchgear, the application performs real partial discharge correction under power frequency phase correlation analysis on the first partial discharge heatmap to obtain a second partial discharge heatmap. Based on the cabinet characteristic dataset of the medium-voltage switchgear, the application performs coupling interference correction under multi-field joint conditions on the second partial discharge heatmap to obtain a third partial discharge heatmap. Based on the real-time operating condition data stream of the medium-voltage switchgear, the application performs dynamic bias correction of the operating condition influence on the third partial discharge heatmap to obtain a fourth partial discharge heatmap. Based on the fourth partial discharge heatmap, a partial discharge early warning command is generated. This invention addresses the technical problems of existing partial discharge online measurement results being susceptible to interference and lacking authenticity and accuracy. By performing multi-stage correction processing on partial discharge monitoring results, it achieves the technical effect of improving the accuracy of online partial discharge measurement and the reliability of early warning. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A schematic diagram of the online measurement and correction method for partial discharge in medium-voltage switchgear provided in this application embodiment;

[0013] Figure 2 A schematic diagram of the structure of the online measurement and correction system for partial discharge of medium-voltage switchgear provided in this application embodiment.

[0014] Explanation of reference numerals in the attached figures: Dynamic correlation adjustment module 11, monitoring and adjustment module 12, real partial discharge correction module 13, coupling interference correction module 14, dynamic bias correction module 15, instruction generation module 16. Detailed Implementation

[0015] This application provides a method and system for correcting online partial discharge measurements in medium-voltage switchgear. It addresses the technical problems of existing online partial discharge measurement results being susceptible to interference and lacking authenticity and accuracy by performing multi-stage correction processing on the partial discharge monitoring results, thereby improving the accuracy of online partial discharge measurements and the reliability of early warning systems.

[0016] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0017] It should be noted that any variation of the terms "comprising" and "having" is intended to cover non-exclusive inclusion, for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such processes, methods, products, or devices.

[0018] Example 1, as Figure 1 As shown, this application provides a method for online measurement and correction of partial discharge in medium-voltage switchgear, the method comprising:

[0019] Step S100: Load the partial discharge log library of the medium-voltage switchgear, and dynamically adjust the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear according to the partial discharge log library to obtain the optimized architecture of the partial discharge sensing.

[0020] In this embodiment of the application, the partial discharge log library formed during the long-term operation of the medium-voltage switchgear is first loaded. The partial discharge log library is used to store the time information, amplitude characteristics, spatial distribution characteristics and corresponding sensor response information of historical partial discharge events.

[0021] Based on the partial discharge log database, when dynamically adjusting the dual-layer architecture of partial discharge sensing in a medium-voltage switchgear, the system first categorizes historical partial discharge locations by their location characteristics, forming multiple sets of partial discharge events corresponding to these locations. Then, based on each event set, the system evaluates the partial discharge sensitivity of the corresponding historical partial discharge locations by assessing the response intensity and stability, obtaining the location-specific partial discharge sensitivity distribution for each location. Next, based on this distribution, the system makes adjustments to the spatial relationships of the partial discharge sensors in the partial discharge sensing array layer, forming the first partial discharge sensing adjustment strategy. Following this, the first adjustment strategy is used to perform dynamic mapping adjustment compensation on the sensing gain adjustment layer, resulting in the second partial discharge sensing adjustment strategy. Finally, the second adjustment strategy is used to perform multi-layer collaborative adjustment of the partial discharge sensing array layer and the sensing gain adjustment layer, generating an optimized partial discharge sensing architecture that matches the partial discharge characteristics of the medium-voltage switchgear.

[0022] Furthermore, in the method provided in the application embodiment, the dynamic correlation adjustment of the partial discharge sensing dual-layer architecture of the medium-voltage switchgear based on the partial discharge log library to obtain the optimized partial discharge sensing architecture further includes:

[0023] The partial discharge log database is used to classify location characteristics and obtain multiple partial discharge event sets corresponding to multiple historical partial discharge locations. The partial discharge sensitivity of the multiple historical partial discharge locations is evaluated based on the multiple partial discharge event sets to obtain a location partial discharge sensitivity characteristic distribution. Based on the location partial discharge sensitivity characteristic distribution, correlation adjustment decisions are made for the partial discharge sensor array layer to obtain a first partial discharge sensor adjustment strategy. Based on the first partial discharge sensor adjustment strategy, dynamic mapping adjustment compensation of the sensor gain adjustment layer is executed to obtain a second partial discharge sensor adjustment strategy. Based on the second partial discharge sensor adjustment strategy, multi-layer collaborative adjustment is performed on the dual-layer partial discharge sensor architecture to generate the optimized partial discharge sensor architecture.

[0024] In this embodiment, the partial discharge log database is first structured and parsed. Each record in the log database is organized into fields such as historical partial discharge location, acquisition time, partial discharge sensor number, discharge pulse amplitude, discharge pulse arrival time, pulse count, and event energy. The historical partial discharge location is then used as the unique index field for grouping. Subsequently, location characteristic classification is performed. In this process, all records in the partial discharge log database are traversed. Records with the same historical partial discharge location are sequentially written into the same data group. Within each data group, records are sorted in ascending order by acquisition time. The acquisition time difference between adjacent records is calculated. If the acquisition time difference is less than a preset time threshold, they are considered the same partial discharge event and merged into one partial discharge event segment. If the acquisition time difference is greater than the preset time threshold, they are considered different partial discharge events. This results in a partial discharge event set containing multiple partial discharge event segments for each historical partial discharge location, ultimately yielding multiple partial discharge event sets corresponding to multiple historical partial discharge locations.

[0025] Next, partial discharge sensitivity evaluation is performed on multiple historical partial discharge locations. In this process, for any set of partial discharge events corresponding to a historical partial discharge location, the number of partial discharge event segments is calculated within a fixed statistical window to obtain the location event frequency. The average amplitude of all discharge pulses at that location is calculated to obtain the location amplitude mean, and the maximum value is taken as the location amplitude peak. The standard deviation of the amplitude sequence is calculated to obtain the location amplitude fluctuation, and the event energy fields are accumulated to obtain the location energy accumulation. Subsequently, linear normalization is performed on the location event frequency, location amplitude mean, location amplitude peak, location amplitude fluctuation, and location energy accumulation. The normalization formula is that the normalized value equals the location index minus the minimum value across all locations, divided by the difference between the maximum and minimum values ​​across all locations. Then, a weighted sum is performed according to fixed weights to obtain the partial discharge sensitivity score for that historical partial discharge location. The weight for normalizing the location event frequency is set to 0.30, the weight for normalizing the location amplitude mean is set to 0.25, the weight for normalizing the location amplitude peak is set to 0.20, the weight for normalizing the location amplitude fluctuation is set to 0.15, and the weight for normalizing the location energy accumulation is set to 0.10. The sum of the above weights is 1. The above calculation is repeated for all historical partial discharge locations to obtain the partial discharge sensitivity score corresponding to each historical partial discharge location. The score is then mapped and output according to the historical partial discharge location number to form a location partial discharge sensitivity characteristic distribution.

[0026] Subsequently, correlation adjustment decisions are made for the partial discharge sensor array layer. In this process, the coverage relationship between historical partial discharge locations and partial discharge sensors is first established. For the partial discharge event set of each historical partial discharge location, the number of times each partial discharge sensor appears in the event set at that location is counted, and the average discharge pulse amplitude of the sensor is calculated. Subsequently, based on the distribution of partial discharge sensitivity characteristics at different locations, a weighting rule is applied to the coverage relationship statistics. When the partial discharge sensitivity score of a historical partial discharge location is in the top 30% of all location scores, the two partial discharge sensors with the highest coverage frequency and the highest average discharge pulse amplitude at that location are designated as high-weight sensors and assigned a weighting coefficient of 0.6. The remaining coverage sensors are designated as low-weight sensors and assigned a weighting coefficient of 0.4. Furthermore, the weighting coefficients of all sensors within the same historical partial discharge location are normalized so that the sum of the weighting coefficients is 1. Then, the sensor weighting coefficients obtained at each historical partial discharge location are summarized according to the partial discharge sensor number. The arithmetic mean of the weighting coefficients obtained by the same partial discharge sensor at different historical partial discharge locations is taken to obtain the final sensor weighting coefficient. The final sensor weighting coefficient is used as the fusion parameter of the partial discharge sensor array layer to form the first partial discharge sensing adjustment strategy.

[0027] After obtaining the first partial discharge (PD) sensing adjustment strategy, dynamic mapping adjustment compensation of the sensing gain adjustment layer is executed. In this process, a one-to-one correspondence is first established between the PD sensors and the sensing gain regulators, ensuring that each PD sensor corresponds to only one gain adjustment channel. Then, the final sensor weight coefficient in the first PD sensing adjustment strategy is converted into a target gain coefficient. The conversion relationship is set so that the target gain coefficient equals the reference gain multiplied by the final sensor weight coefficient, and minimum and maximum limits are set for the target gain coefficient to avoid excessively small or large gains. Then, in each gain update cycle, the average amplitude of the discharge pulse currently output by each PD sensor is collected as the current average amplitude. The current average amplitude is compared with the median of the preset target amplitude range, and the deviation is calculated as the target amplitude median minus the current average amplitude. This deviation is then multiplied by a scaling factor to obtain the gain compensation amount. Finally, the target gain coefficient and the gain compensation amount are added to obtain the compensated gain setting value, which is then written into the channel control parameters of the sensing gain adjustment layer to form the second PD sensing adjustment strategy.

[0028] Finally, multi-layer collaborative adjustment is performed on the partial discharge (PD) sensing dual-layer architecture. In this process, the first PD sensing adjustment strategy is applied to the PD sensing array layer, causing the signals acquired by the PD sensors to be weighted according to the final sensor weight coefficients during the fusion calculation. Then, the second PD sensing adjustment strategy is applied to the sensing gain adjustment layer, causing each gain regulator to adjust its gain according to the compensated gain setpoint. Next, within a unified verification cycle, the average amplitude of the discharge pulses output by each PD sensor is statistically analyzed. The difference in average amplitude between the high-weight sensor group and the low-weight sensor group is calculated and compared with a preset discrimination threshold. When the average amplitude difference does not reach the preset discrimination threshold, the compensated gain setpoint corresponding to the high-weight sensor group is increased by a fixed adjustment step while the compensated gain setpoint corresponding to the low-weight sensor group is simultaneously decreased until the average amplitude difference reaches the preset discrimination threshold. When the fusion parameters of the PD sensing array layer and the channel control parameters of the sensing gain adjustment layer remain stable and unchanged for multiple consecutive verification cycles, the configuration parameters of the PD sensing dual-layer architecture are solidified, generating the optimized PD sensing architecture.

[0029] Furthermore, the method provided in the application embodiments also includes:

[0030] The partial discharge sensing dual-layer architecture includes a partial discharge sensing array layer and a sensing gain adjustment layer. The partial discharge sensing array layer includes multiple partial discharge sensors deployed in the medium-voltage switchgear. The sensing gain adjustment layer includes multiple gain regulators. The partial discharge sensing array layer and the sensing gain adjustment layer are bidirectionally connected through data mapping and feedback control relationships.

[0031] In this embodiment, the partial discharge sensing dual-layer architecture is constructed in a layered collaborative structure. The partial discharge sensing array layer serves as the front-end sensing layer, which includes multiple partial discharge sensors deployed in different functional units and key insulation parts of the medium-voltage switchgear. Each partial discharge sensor corresponds to a different spatial position and detection direction, and is used to synchronously collect pulse signals generated by partial discharge and form multi-channel sensing data.

[0032] The sensor gain adjustment layer, acting as a signal conditioning layer, contains multiple gain regulators corresponding one-to-one with each partial discharge sensor channel. These regulators independently configure the gain of the signals output from each sensor channel to compensate for differences in signal amplitude under different sensing locations, sensor sensitivities, and operating conditions. A clear channel correspondence and parameter association are established between the partial discharge sensor array layer and the sensor gain adjustment layer through data mapping. This ensures that the output data of each partial discharge sensor is accurately mapped to its corresponding gain regulator. Simultaneously, feedback control transmits the signal state, amplitude changes, and stability information after gain adjustment back to dynamically correct the gain parameters and sensing response. This enables bidirectional communication and collaborative operation between the partial discharge sensor array layer and the sensor gain adjustment layer.

[0033] Step S200: Perform online partial discharge monitoring and adaptive gain adjustment on the medium-voltage switchgear according to the partial discharge sensing optimization architecture to obtain the first partial discharge thermal map.

[0034] In this embodiment, when performing online partial discharge monitoring of medium-voltage switchgear based on the partial discharge sensing optimization architecture, all partial discharge sensors in the partial discharge sensing array layer are first activated simultaneously. Each partial discharge sensor continuously samples the electrical signals generated near its location according to a pre-set fixed sampling frequency. When partial discharge occurs inside the medium-voltage switchgear, the partial discharge sensor captures the instantaneously changing pulse signal and converts the pulse signal into a voltage signal output. At the same time, the installation position number corresponding to the partial discharge sensor and the current sampling time information are added to the output signal, so that each acquired signal can be clearly associated with a specific sensing location and time point.

[0035] After the partial discharge sensor completes signal acquisition and output, the acquired signal immediately enters the sensor gain adjustment layer. Each partial discharge sensor corresponds to a gain regulator. The gain regulator continuously monitors the signal amplitude change of its corresponding channel and calculates the maximum and minimum amplitude of the output signal of that channel within a set time period. When the overall signal amplitude is detected to be too low and it is difficult to distinguish pulse characteristics, the gain regulator gradually increases the amplification factor to increase the signal amplitude. When the signal amplitude is detected to be too high and close to the upper limit of the system's allowable range, the gain regulator gradually decreases the amplification factor to bring the signal amplitude back to the normal range. By repeatedly adjusting the amplification factor, the signal amplitude output by each channel is stabilized in a suitable range that is neither too small nor too large, thereby completing the adaptive gain adjustment of the partial discharge signal.

[0036] After the signal amplitude stabilizes, the signal of each channel is checked point by point. When the signal voltage exceeds the preset judgment threshold, the signal segment is determined to be a partial discharge pulse. The specific time of the pulse occurrence, the maximum amplitude of the pulse, and the duration of the pulse are recorded. At the same time, multiple pulses appearing in the same channel within a certain time range are counted to obtain the number and intensity of partial discharge pulses in that channel within that time range.

[0037] After obtaining the partial discharge pulse information corresponding to each partial discharge sensor channel, the sensor channel is matched with a specific area inside the medium-voltage switchgear according to the installation position of each partial discharge sensor in the medium-voltage switchgear. The inside of the medium-voltage switchgear is divided into multiple fixed areas, each area corresponding to one or more partial discharge sensors. When the partial discharge sensor in a certain area detects more or stronger partial discharge pulses, it is considered that the partial discharge activity in that area is more obvious. The partial discharge intensity value of that area is obtained by accumulating the pulse amplitude and pulse number in that area.

[0038] After calculating the partial discharge intensity values ​​for each region, the partial discharge intensity values ​​for each region of the medium-voltage switchgear are mapped to different color shades in ascending order. Regions with lower partial discharge intensity are displayed in light colors, and regions with higher partial discharge intensity are displayed in dark colors. The colors of each region are arranged and combined according to the actual structural position of the medium-voltage switchgear, and finally, a partial discharge first thermal map that can intuitively reflect the partial discharge distribution inside the medium-voltage switchgear is generated.

[0039] Step S300: Based on the real-time voltage sequence of the medium-voltage switchgear, perform real partial discharge correction under power frequency phase correlation analysis on the first partial discharge heat map to obtain the second partial discharge heat map.

[0040] In this embodiment, when performing power frequency phase correlation analysis to correct the partial discharge (PD) in the first PD heatmap based on the real-time voltage sequence of the medium-voltage switchgear, the real-time voltage sequence is first processed to achieve power frequency phase synchronization, establishing a voltage cycle phase reference model to characterize the phase change of the voltage cycle. Then, each PD pulse signal contained in the first PD heatmap is phase-mapped with the voltage cycle phase reference model to extract the distribution of PD pulses at different power frequency phases, forming power frequency phase distribution features. Next, the repeatability of the PD pulse signals and their corresponding power frequency phase distribution features is identified to obtain a correlation repetition characteristic distribution reflecting the repetitive occurrence of PD pulses within the power frequency cycle. Simultaneously, the consistency of the PD pulse signals and their corresponding power frequency phase distribution features is identified to obtain a correlation consistency characteristic distribution reflecting the stability of the relationship between PD pulses and the power frequency phase. Finally, based on the correlation repetition characteristic distribution and the correlation consistency characteristic distribution, pseudo-PD signals in the first PD heatmap are suppressed, thereby generating the second PD heatmap corrected by power frequency phase correlation analysis.

[0041] Furthermore, in the method provided in the application embodiment, the method further includes performing real partial discharge correction under power frequency phase correlation analysis on the first partial discharge heat map based on the real-time voltage sequence of the medium-voltage switchgear to obtain the second partial discharge heat map, and also includes:

[0042] The real-time voltage sequence is subjected to power frequency phase synchronization processing to establish a voltage cycle phase reference model; the partial discharge pulse signals in the first partial discharge heatmap are phase mapped with the voltage cycle phase reference model to obtain the power frequency phase distribution characteristics; the partial discharge pulse signals and the power frequency phase distribution characteristics are subjected to repeatability identification to obtain the associated repeatability characteristic distribution; the partial discharge pulse signals and the power frequency phase distribution characteristics are subjected to consistency identification to obtain the associated consistency characteristic distribution; the pseudo partial discharge signal is suppressed in the first partial discharge heatmap according to the associated repeatability characteristic distribution and the associated consistency characteristic distribution to generate the second partial discharge heatmap.

[0043] In this embodiment, when performing power frequency phase synchronization processing on the real-time voltage sequence and establishing a voltage cycle phase reference model, the real-time voltage sequence of the medium-voltage switchgear is first continuously sampled at a fixed sampling interval to obtain a voltage sampling point sequence arranged in ascending order of time. Each voltage sampling point includes the sampling time and voltage value. Then, the sign change of the voltage values ​​between adjacent points is determined point by point in the voltage sampling point sequence. When a situation occurs where the voltage value of the preceding point is less than or equal to zero and the voltage value of the following point is greater than zero, it is recorded as a positive zero-crossing point. The sampling point range between every two adjacent positive zero-crossing points is defined as a power frequency cycle. Next, the number of sampling points within each power frequency cycle is calculated as the cycle length, and the first sampling point within that power frequency cycle is defined as the phase start point. Then, the phase value of any sampling point within that power frequency cycle is calculated. The phase value of a sampling point is equal to the index of the sampling point within the cycle divided by the total number of sampling points within the cycle, multiplied by 360 degrees, thus ensuring that the phase at the cycle start point is 0 degrees and the phase at the cycle end point is close to 360 degrees. Finally, the sampling time and phase value of each sampling point are stored accordingly to form a voltage cycle phase reference model.

[0044] When performing phase mapping between each partial discharge pulse signal within the first partial discharge heatmap and the voltage cycle phase reference model, the occurrence time of each partial discharge pulse signal recorded during the generation of the first partial discharge heatmap is first read. A sampling time with the same occurrence time is then searched in the voltage cycle phase reference model. When the occurrence time falls between two sampling times, the sampling time closer to the occurrence time is selected. The phase value corresponding to this sampling time is then extracted as the power frequency phase of the partial discharge pulse signal. Next, the range from 0 degrees to 360 degrees is divided into phase intervals of fixed width, for example, 36 phase intervals, each 10 degrees wide. Each partial discharge pulse signal is then categorized according to the phase interval into which its power frequency phase falls. The number of partial discharge pulse signals is then counted within each phase interval, and the amplitude or energy of the partial discharge pulse signals falling within that phase interval is summed to obtain the statistical values ​​of the number of pulses and pulse energy for each phase interval, thus forming the power frequency phase distribution characteristics.

[0045] Subsequently, when performing repeatability identification on the distribution characteristics of each partial discharge pulse signal and each power frequency phase, the continuously acquired partial discharge pulse signals were first grouped according to the power frequency cycle number. The power frequency cycle number was determined by the order of positive zero crossover points, and each power frequency cycle number corresponded to a power frequency cycle time period. Then, within each power frequency cycle number, the number of pulses in each phase interval within that cycle was re-counted according to the aforementioned phase interval division, resulting in a phase interval pulse count vector for that power frequency cycle. Next, the phase interval pulse count vectors of multiple adjacent power frequency cycles were compared interval by interval, and the repetition count for each phase interval was calculated. The repetition count was calculated as the number of cycles in which the number of pulses in that phase interval was greater than or equal to 1 in N consecutive power frequency cycles. Then, the repetition count for each phase interval was divided by N to obtain the repeatability ratio value. The larger the repeatability ratio value, the more likely that partial discharge pulses are to repeatedly occur in different power frequency cycles in that phase interval. Finally, the repeatability ratio values ​​of all phase intervals were arranged in phase interval order to obtain the associated repeatability characteristic distribution.

[0046] Next, when identifying the consistency of the distribution characteristics of each partial discharge pulse signal and each power frequency phase, the entire power frequency cycle is first divided into multiple continuous time periods in chronological order. Each time period contains M power frequency cycles, and the number of pulses in each phase interval is summarized and counted within each time period to obtain the pulse number vector of the phase interval for that time period. Then, the phase concentration is calculated for each time period, and the phase interval with the largest number of pulses in that time period is designated as the main phase interval. The proportion of the sum of the pulses in the main phase interval and its several adjacent phase intervals to the total number of pulses in all phase intervals of that time period is calculated as the concentration ratio value for that time period. Then, the main phase intervals of adjacent time periods are compared, and the offset of the main phase interval between different time periods is counted. The offset is represented by the difference in phase interval index. Then, the phase interval range with a high concentration ratio value and a small main phase interval offset is determined as the phase range with strong consistency. A consistency score is calculated for each phase interval. This consistency score is calculated based on the change in the number of pulses in that phase interval within each time period. The smaller the change, the higher the consistency score. Finally, the consistency scores of all phase intervals are arranged in phase interval order to obtain the distribution of correlation consistency characteristics.

[0047] Finally, when suppressing pseudo-partial discharge signals in the first partial discharge heatmap based on the correlation repetition characteristic distribution and correlation consistency characteristic distribution, a retention criterion is first established for each phase interval. The retention criterion is determined by both the repetition ratio and the consistency score. That is, when the repetition ratio and consistency score of a certain phase interval are both below a preset threshold, the phase interval is marked as a pseudo-partial discharge contribution interval. Subsequently, each partial discharge pulse signal in the first partial discharge heatmap is checked for its power frequency phase interval. If the phase interval is a pseudo-partial discharge contribution interval, the contribution value of the partial discharge pulse signal in the heatmap intensity accumulation is set to zero, and the pulse is not counted in the pulse number accumulation and pulse energy accumulation at the corresponding spatial location. If the phase interval is not a pseudo-partial discharge contribution interval, the contribution value of the partial discharge pulse signal is retained and continues to be included in the accumulation. Then, the heatmap intensity value is recalculated for each spatial location, and the accumulated pulse energy and accumulated pulse number of the retained partial discharge pulse signals in that spatial location are directly accumulated to obtain the spatial intensity value. Finally, the recalculated spatial intensity values ​​are visualized and mapped according to the original spatial layout to generate a second partial discharge heatmap after pseudo partial discharge signal suppression.

[0048] Step S400: Based on the cabinet characteristic dataset of the medium-voltage switchgear, perform multi-field joint coupling interference correction on the second partial discharge thermal map to obtain the third partial discharge thermal map.

[0049] In this embodiment, when performing multi-field coupled interference correction on the second partial discharge thermal map based on the cabinet characteristic dataset of the medium-voltage switchgear, a multi-physics cabinet coupling model is first constructed based on the cabinet characteristic dataset to characterize the interaction of multiple physical fields within the medium-voltage switchgear. Then, based on the multi-physics cabinet coupling model, the partial discharge process is inverted on the partial discharge spatial distribution reflected in the second partial discharge thermal map, yielding multiple corresponding partial discharge inversion process data. Next, coupling interference detection is performed on each partial discharge inversion process data to identify and determine each partial discharge coupling interference component introduced by the multi-physics coupling effect during the partial discharge process. Finally, the second partial discharge thermal map is mapped and corrected based on each partial discharge coupling interference component, thereby generating the third partial discharge thermal map after multi-physics coupled interference correction.

[0050] Furthermore, in the method provided in the application embodiment, the method of performing multi-field joint coupling interference correction on the second partial discharge thermal map based on the cabinet characteristic dataset of the medium-voltage switchgear to obtain the third partial discharge thermal map further includes:

[0051] Based on the cabinet characteristic dataset, a multiphysics cabinet coupling model is constructed; based on the multiphysics cabinet coupling model, the partial discharge process is inverted on the second partial discharge heat map to obtain multiple partial discharge inversion process data; coupling interference detection is performed on each partial discharge inversion process data to determine each partial discharge coupling interference component; based on each partial discharge coupling interference component, the second partial discharge heat map is mapped and corrected to generate the third partial discharge heat map.

[0052] In this embodiment, when constructing a multiphysics-based cabinet coupling model based on the cabinet characteristic dataset, the cabinet structural characteristic data is first read. The dimensional parameters and installation position parameters of the medium-voltage switchgear's shell, partitions, conductor components, and insulation components are mapped to the same spatial coordinate system to determine the geometric boundaries of each structural component within the cabinet. Then, internal spatial distribution data is read to determine the spatial relationships of different areas within the cabinet, such as the busbar compartment, cable compartment, and circuit breaker compartment. Next, cabinet material characteristic data is read, binding the physical parameters such as dielectric constant and conductivity of each structural component to its spatial location. Simultaneously, sensor deployment data is read to determine the installation coordinates of each partial discharge sensor within the cabinet and establish the correspondence between sensor positions and cabinet spatial positions. After completing the above data integration, a multiphysics-based cabinet coupling model describing the propagation path, propagation distance, and material attenuation characteristics of partial discharge signals within the cabinet is formed.

[0053] Next, when inverting the partial discharge process based on the second partial discharge heatmap using the multiphysics cabinet coupling model, the second partial discharge heatmap is first represented as a set of intensities composed of multiple spatial locations, with each spatial location corresponding to a partial discharge intensity value. Then, based on the cabinet structural characteristics and internal spatial distribution data, multiple candidate discharge locations are selected in insulation gaps, conductor connections, and structural edge regions. For any candidate discharge location, it is assumed that a unit intensity discharge occurs at that location, with the unit intensity discharge value set to 1. The response results generated by this unit intensity discharge at each spatial location are calculated under the constraints of the multiphysics cabinet coupling model. During the calculation, the straight-line distance between the candidate discharge location and each spatial location is first calculated. Then, based on the material type traversed between them, the corresponding attenuation coefficient is determined; this attenuation coefficient is given by the cabinet material characteristics data. Next, the response value is calculated based on the distance and the attenuation coefficient. The unit intensity is divided by the square of the distance and then multiplied by the attenuation coefficient; that is, the response value equals 1 divided by the square of the distance and then multiplied by the attenuation coefficient. When a partition or conductor obstructs a candidate discharge location from a given spatial location, the calculated result is further multiplied by a preset reflection or obstruction correction coefficient. By repeating the distance calculation, attenuation calculation, and correction calculation steps for all spatial locations, a set of response results corresponding to the unit intensity discharge at each spatial location is obtained; this set of response results is the unit intensity response result. Next, the discharge intensity is calculated based on this response result. The actual intensity value at each spatial location in the second partial discharge thermogram is multiplied one by one with the corresponding unit intensity response result, and the sum is used as the numerator. The squares of the unit intensity response results at each spatial location are then summed to obtain the denominator. The discharge intensity equals the numerator divided by the denominator, and the discharge intensity is limited to a non-negative value. Subsequently, the calculated discharge intensity is substituted into the aforementioned response calculation process, that is, the unit intensity is replaced with the calculated discharge intensity, and the distance calculation, attenuation calculation, and correction calculation steps are re-executed to obtain the model output intensity result corresponding to each spatial location under the conditions of the candidate discharge location and the calculated discharge intensity. The model's output intensity result is then compared position-by-position with the actual intensity values ​​at each spatial location in the second partial discharge thermogram. The residual value is obtained by calculating and summing the squared differences between the two values. Finally, adjacent locations around the candidate discharge location are selected, and the above process of calculating the unit intensity response, discharge intensity, and residual is repeated. The location with the smallest residual value is selected as the final discharge location. This final discharge location, its corresponding discharge intensity, and the model's output intensity result are output as a partial discharge inversion process data. By repeating the above inversion process for different candidate discharge locations, multiple partial discharge inversion process data are obtained.

[0054] Subsequently, coupled interferometry detection was performed on the data from each partial discharge inversion process. In this process, firstly, multi-field coupled partial discharge interference records generated during the operation of the medium-voltage switchgear were collected, and a coupled interferometry detection history space was constructed to characterize the distribution of interference features. Then, multiple coupled interferometry detection models for identifying different interference features were trained based on the coupled interferometry detection history space, and these models were integrated and fused to form a unified coupled interferometry detection channel. Finally, the data from each partial discharge inversion process was input into the coupled interferometry detection channel to identify and separate the interference components included in the inversion process, thereby obtaining the corresponding coupled interferometry components of each partial discharge.

[0055] Finally, the partial discharge (PD) second heatmap is mapped and corrected based on each PD coupling interference component. In this process, firstly, the spatial location of the PD coupling interference component corresponding to each PD inversion process data is calibrated, and a one-to-one correspondence is established between the PD coupling interference component and its corresponding spatial location in the PD second heatmap, ensuring that each spatial location is associated with a corresponding coupling interference intensity value. Then, the PD coupling interference components from different PD inversion process data are aggregated. When multiple PD inversion process data detect PD coupling interference components at the same spatial location, the intensity values ​​of the PD coupling interference components corresponding to that spatial location are accumulated to form a comprehensive PD coupling interference intensity distribution for each spatial location. Next, a mapping correction process is performed on the PD intensity values ​​in the PD second heatmap. Specifically, for each spatial location in the PD second heatmap, the original PD intensity value corresponding to that spatial location is read, and the comprehensive PD coupling interference intensity value corresponding to that spatial location is subtracted from the original PD intensity value to obtain the corrected PD intensity value. When the corrected PD intensity value is less than zero, it is limited to zero to ensure the physical rationality of the correction result. By repeating the above subtraction and correction process for all spatial locations in the second partial discharge heatmap, a set of partial discharge intensity distribution data after coupling interference reduction is obtained. Finally, following the same spatial layout and display rules as the second partial discharge heatmap, the corrected partial discharge intensity distribution data is visualized to generate a third partial discharge heatmap, which characterizes the spatial distribution of partial discharge inside the medium-voltage switchgear after multi-physics field coupling interference correction.

[0056] Furthermore, in the method provided in the application embodiments, coupling interferometry detection is performed on each partial discharge inversion process data to determine each partial discharge coupling interferometry component, and the method further includes:

[0057] Multi-field coupled partial discharge interference records of the medium-voltage switchgear are collected to obtain the coupled interference detection history space; multiple coupled interference detection models are trained based on the coupled interference detection history space; the multiple coupled interference detection models are integrated and fused for training to obtain the coupled interference detection channel; the data of each partial discharge inversion process are input into the coupled interference detection channel to obtain the coupled interference components of each partial discharge.

[0058] In this embodiment, when collecting multi-field coupled partial discharge interference records of a medium-voltage switchgear, the partial discharge pulse data of each partial discharge sensor channel are continuously collected during the operation of the switchgear, and the real-time voltage sequence, switch operation status, load level, and environmental parameters at the same moment are recorded simultaneously. The collected signals are then aligned by time and mapped to the cabinet spatial location according to the sensor installation position, so that each record includes a spatial location identifier, a time identifier, and the corresponding signal segment. Next, basic quantities characterizing the interference features are calculated for each signal segment, including the average pulse amplitude, peak pulse amplitude, number of pulses, cumulative pulse energy, the synchronous occurrence ratio of adjacent sensor channels, spatial diffusion range, and the change amplitude within adjacent time windows. These basic quantities are then combined into a feature vector. Historical data are then labeled, and multiple collection results are compared at the same spatial location. When the signal exhibits abnormal diffusion with non-single-source attenuation in space and fluctuates synchronously with changes in operation status, load level, or environmental parameters, the corresponding record is labeled as an interference sample, and the rest are labeled as non-interference samples. Finally, all historical records with feature vectors and annotation results are collected in spatial and temporal order to obtain the coupled interferometric detection historical space. This coupled interferometric detection historical space consists of a set of samples, each of which contains spatial location identifiers, feature vectors, and interferometric annotations.

[0059] Next, when training multiple coupled interferometry detection models based on the coupled interferometry detection historical space, the historical space is first partitioned into training and validation sets proportionally, and the same feature processing method is applied to both sets. Then, normalization is performed on each feature dimension in the training set by subtracting its mean from the training set's mean and dividing by its standard deviation, ensuring that features of different dimensions can participate in the calculation at the same scale. Next, multiple coupled interferometry detection models are constructed, each using the same model architecture: a feature vector as input, a weighted summation structure as computation, and a probability output structure as output. The weighted summation structure multiplies each feature dimension by its corresponding weight and sums the results to obtain a discriminant value, while the probability output structure converts the discriminant value into an interferometric probability between 0 and 1. During training, different feature subsets are selected as inputs for each coupled interferometry detection model. For example, the first coupled interferometry detection model uses only amplitude and energy-related features, the second model uses only spatial diffusion-related features, and the third model uses only operating condition synchronization fluctuation-related features, so that different coupled interferometry detection models focus on identifying different types of interference behaviors. Then, the parameters of each coupled interferometry detection model are solved, that is, the output interference probability is calculated for each sample on the training set, and the difference between the output interference probability and the sample label is taken as the error. The total error on the training set is gradually reduced by repeatedly adjusting the weights. Adjustment stops when the decrease in total error is less than a preset threshold after several rounds of adjustment. Finally, the recognition accuracy and false alarm rate of each coupled interferometry detection model are calculated on the validation set, and multiple coupled interferometry detection models that meet the preset indicators are retained, resulting in multiple coupled interferometry detection models that can be used for subsequent fusion.

[0060] Subsequently, when integrating and training multiple coupled interferometric detection models, the output of each model was first uniformly defined, ensuring that all outputs corresponded to the interference probability at the same spatial location. Then, the false alarm rate and false negative rate of each coupled interferometric detection model were calculated on the validation set, and a fusion weight was assigned to each model. The fusion weights satisfied that the sum of the fusion weights of all models was 1, with models having lower false alarm and false negative rates corresponding to higher fusion weights. Next, a coupled interferometric detection channel was constructed, with a parallel input and single output structure. The parallel input structure simultaneously inputs feature vectors at the same spatial location into multiple coupled interferometric detection models, each outputting its own interference probability. The single output structure weighted and summed the interference probabilities according to the fusion weights to obtain the fused interference probability. Then, a decision threshold was set for the fused interference probability on the validation set. When the fused interference probability was greater than or equal to the decision threshold, the interference was declared true; when the fused interference probability was less than the decision threshold, the interference was declared false. The decision threshold was adjusted to ensure that the false alarm and false negative rates on the validation set met preset requirements. Ultimately, a coupled interference detection channel is formed that can output the fused interference probability and interference determination result from the input feature vector.

[0061] Finally, when inputting the partial discharge (PD) inversion process data into the coupled interferometric detection channel, each PD inversion process data is first expanded according to its spatial location, ensuring that the PD inversion process data at each spatial location includes the corresponding inversion intensity value and the intensity difference with adjacent spatial locations. Then, a feature vector consistent with the historical space of the coupled interferometric detection is calculated for each spatial location. The feature vector consists of the inversion intensity value at that spatial location, the intensity difference with adjacent spatial locations, the spatial diffusion range index, and the fluctuation index changing with the time window, and is normalized using the same normalization parameters as in the training phase. Next, the feature vector of each spatial location is input into the coupled interferometric detection channel. Multiple coupled interferometric detection models output interference probabilities, which are then weighted and summed to obtain the fused interference probability. Simultaneously, the interference determination result is obtained based on the determination threshold. When the interference determination result is true, the PD coupled interference component at that spatial location is determined as the product of the fused interference probability and the inversion intensity value at that spatial location, so that a higher degree of interference corresponds to a larger PD coupled interference component. When the interference determination result is false, the PD coupled interference component at that spatial location is determined to be 0. By repeating the above input, calculation, and output steps for all spatial locations in the partial discharge inversion process data, the partial discharge coupling interference components covering each spatial location are obtained.

[0062] Furthermore, the method provided in the application embodiments also includes:

[0063] The cabinet characteristic dataset includes cabinet structural characteristic data, cabinet material characteristic data, internal space distribution data, and sensor deployment data of the medium-voltage switchgear.

[0064] In this embodiment, the cabinet characteristic dataset includes cabinet structural characteristic data, cabinet material characteristic data, internal space distribution data, and sensor deployment data for the medium-voltage switchgear. The cabinet structural characteristic data characterizes the structural form and geometric relationships of the medium-voltage switchgear's outer shell, partitions, and internal components; the cabinet material characteristic data characterizes the material properties of each structural component; the internal space distribution data characterizes the spatial division and positional relationships of the functional areas within the medium-voltage switchgear; and the sensor deployment data characterizes the installation positions and corresponding relationships of partial discharge sensors within the medium-voltage switchgear.

[0065] Step S500: Based on the real-time operating condition data stream of the medium-voltage switchgear, perform dynamic bias correction on the operating condition impact of the third partial discharge thermal map to obtain the fourth partial discharge thermal map.

[0066] In this embodiment, when performing dynamic bias correction on the third partial discharge heatmap based on the real-time operating condition data stream of the medium-voltage switchgear, the real-time operating condition data stream is first decomposed to extract the operating status data stream, environmental operating condition data stream, and load operating condition data stream. Then, the influence of the operating status data stream on the partial discharge intensity in the third partial discharge heatmap as a function of switch operation is analyzed, forming an operating-related bias characteristic distribution. Simultaneously, the relationship between the partial discharge intensity in the third partial discharge heatmap and environmental factors such as temperature and humidity is analyzed based on the environmental operating condition data stream, forming an environmental-related bias characteristic distribution. Finally, the offset relationship between the partial discharge intensity in the third partial discharge heatmap and load level changes is analyzed based on the load operating condition data stream, forming a load-related bias characteristic distribution. Then, the operating-related bias characteristic distribution, the environmental-related bias characteristic distribution, and the load-related bias characteristic distribution are jointly modeled to construct a multi-condition bias coupling correction model. Finally, the third partial discharge heatmap is collaboratively corrected based on the multi-condition bias coupling correction model, resulting in a fourth partial discharge heatmap after dynamic bias correction based on operating condition influence.

[0067] Furthermore, in the method provided in the application embodiment, the method of dynamically biasing the partial discharge third thermal map based on the real-time operating condition data stream of the medium-voltage switchgear to obtain the partial discharge fourth thermal map further includes:

[0068] Based on the real-time operating condition data stream, extract the operation status data stream, environmental operating condition data stream, and load operating condition data stream; analyze the operation-related bias influence of the third partial discharge heatmap based on the operation status data stream to obtain the operation-related bias characteristic distribution; analyze the environmental-related bias influence of the third partial discharge heatmap based on the environmental operating condition data stream to obtain the environmental-related bias characteristic distribution; analyze the load-related bias influence of the third partial discharge heatmap based on the load operating condition data stream to obtain the load-related bias characteristic distribution; construct a multi-operating condition bias coupling correction model based on the operation-related bias characteristic distribution, the environmental-related bias characteristic distribution, and the load-related bias characteristic distribution; perform collaborative correction on the third partial discharge heatmap based on the multi-operating condition bias coupling correction model to obtain the fourth partial discharge heatmap.

[0069] In this embodiment, when extracting the operation status data stream, environmental condition data stream, and load condition data stream from the real-time operating condition data stream, the real-time operating condition data stream is first sorted by timestamp and aligned to a unified time axis. Then, it is extracted according to the data field type. Fields containing circuit breaker commands, mechanism action signals, and operating mode switching identifiers are aggregated into the operation status data stream; fields containing temperature, humidity, and their time-varying values ​​are aggregated into the environmental condition data stream; and fields containing three-phase current, load rate, and their time-varying values ​​are aggregated into the load condition data stream. For fields with missing sampling points, linear interpolation of adjacent sampling points is used to fill in the missing points, ensuring that the three types of data streams have a corresponding relationship on the same time axis, thereby obtaining the operation status data stream, environmental condition data stream, and load condition data stream.

[0070] Next, the operation-related bias impact of the third partial discharge heatmap is analyzed based on the operation status data stream. In this process, the occurrence time of each switch opening / closing, mechanism action, and operating mode switch is first identified from the operation status data stream, and a pre-operation time window and a post-operation time window are constructed centered on each operation moment. Then, the corresponding third partial discharge heatmap data within the pre-operation and post-operation time windows are extracted respectively. For each spatial location, the difference between the average partial discharge intensity within the post-operation time window and the average partial discharge intensity within the pre-operation time window is calculated, and this difference is used as the operation bias amount for that spatial location under that operation. The above calculation process is then repeated for similar operation events, and the operation bias amount for the same spatial location is arithmetically averaged to eliminate the influence of occasional fluctuations, ultimately forming an operation-related bias characteristic distribution reflecting the degree to which the partial discharge intensity at each spatial location is affected by the operation behavior.

[0071] When analyzing the environmental correlation bias impact of the third heatmap of partial discharge based on the environmental condition data stream, the temperature and humidity sequences are first extracted from the data stream and then segmented into multiple temperature and humidity intervals according to preset intervals. Then, for each temperature interval, the third heatmap data of the corresponding time period is selected, and the average partial discharge intensity for each spatial location within that time period is calculated. Similarly, for each humidity interval, the average partial discharge intensity for the corresponding spatial location is calculated. Next, using the average partial discharge intensity of each spatial location over the entire time period as a benchmark, the offset of the average partial discharge intensity of each spatial location relative to the benchmark value is calculated for each temperature and humidity interval. Then, the temperature and humidity offsets are superimposed according to a preset ratio to obtain the comprehensive environmental bias of each spatial location under changing environmental conditions, and the results are output in spatial order to form an environmental correlation bias characteristic distribution.

[0072] When analyzing the load-related bias impact of the third heatmap of partial discharge based on the load condition data stream, the three-phase current values ​​are first extracted from the load condition data stream, and the corresponding load rate sequence is calculated. Then, the load rate sequence is divided into multiple load condition intervals. Next, for each load condition interval, the third heatmap data of partial discharge within the corresponding time period is selected, and the average partial discharge intensity within that interval is calculated for each spatial location. Then, using the average partial discharge intensity at each spatial location over the entire time period as a benchmark, the difference between the average partial discharge intensity at each spatial location in each load condition interval and the benchmark value is calculated to obtain the load bias. Afterwards, the relationship between the load bias and the load rate is fitted, and the load sensitivity coefficients corresponding to each spatial location are extracted. These load sensitivity coefficients are then arranged according to spatial location to form a load-related bias characteristic distribution.

[0073] Subsequently, a multi-condition bias coupling correction model is constructed based on the distributions of operation-related bias characteristics, environmental-related bias characteristics, and load-related bias characteristics. In this process, firstly, the distributions of operation-related bias characteristics, environmental-related bias characteristics, and load-related bias characteristics are subjected to unified spatial mapping and scale normalization, enabling bias information from different sources to form a comparable set of condition bias feature vectors under the same spatial coordinate system and numerical scale. Then, based on the set of condition bias feature vectors, the relative influence of different condition biases at various spatial locations is analyzed and weights are assigned, establishing a multi-condition bias attention relationship reflecting the strength of influence among operation conditions, environmental conditions, and load conditions. Afterwards, according to the multi-condition bias attention relationship, the set of condition bias feature vectors is weighted and coupled, giving a higher proportion to condition biases that have a more significant impact on partial discharge results in the correction, thereby generating a multi-condition bias coupling correction model that can comprehensively characterize the coupling effects of multiple condition factors.

[0074] Finally, the third partial discharge heatmap is collaboratively corrected based on the multi-condition bias coupling correction model. In this process, the original partial discharge intensity value is first read for each time point and spatial location in the third partial discharge heatmap, along with the corresponding operating status data stream, environmental condition data stream, and load condition data stream values. These values ​​are then substituted into the multi-condition bias coupling correction model to calculate the bias correction amount for that time point and spatial location. Next, the bias correction amount is subtracted from the original partial discharge intensity value to obtain the corrected partial discharge intensity value, which is then limited to a value not less than zero. Finally, the above collaborative correction process is repeated for all time points and all spatial locations, and the result is visualized using the same spatial mapping method as the third partial discharge heatmap, generating a fourth partial discharge heatmap that reflects the spatial distribution of partial discharge in the medium-voltage switchgear after dynamic bias correction under actual operating conditions.

[0075] Furthermore, the method provided in the application embodiments, which constructs a multi-condition bias coupling correction model based on the operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution, further includes:

[0076] The operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution are subjected to unified spatial mapping and scale normalization to obtain a set of operating condition bias feature vectors; attention is allocated according to the set of operating condition bias feature vectors to establish multi-operating condition bias attention relationships; attention-driven coupling is performed on the set of operating condition bias feature vectors according to the multi-operating condition bias attention relationships to generate the multi-operating condition bias coupling correction model.

[0077] In this embodiment, the operation-related bias characteristic distribution, environment-related bias characteristic distribution, and load-related bias characteristic distribution are first subjected to unified spatial mapping processing. The spatial division result adopted by the partial discharge third heat map is used as the sole spatial reference, and the interior of the medium-voltage switchgear is divided into several fixed spatial locations, each spatial location corresponding to a grid cell in the heat map. Subsequently, the bias values ​​in the operation-related bias characteristic distribution, environment-related bias characteristic distribution, and load-related bias characteristic distribution are read and mapped to the corresponding spatial locations. When the spatial resolution of a certain bias characteristic distribution is higher than the heat map resolution, the arithmetic mean of multiple bias values ​​falling within the same spatial location is taken as the bias value of that spatial location. When the spatial resolution is lower than the heat map resolution, the bias value is copied and mapped to all spatial locations it covers, thereby ensuring that each spatial location simultaneously possesses operation-related bias values, environment-related bias values, and load-related bias values.

[0078] After completing the unified spatial mapping, the operation-related bias values, environment-related bias values, and load-related bias values ​​are scaled and normalized. The maximum and minimum values ​​of the same type of bias are calculated across all spatial locations, and linear normalization is performed on the bias value at each spatial location to ensure that the normalized bias value falls within the range of 0 to 1, thereby eliminating differences in numerical magnitude and units between different bias sources. Subsequently, at each spatial location, the normalized operation-related bias values, environment-related bias values, and load-related bias values ​​are combined into a three-dimensional array in a fixed order to fully describe the bias state of that spatial location under the current operating conditions. The three-dimensional arrays corresponding to all spatial locations together constitute the set of operating condition bias feature vectors.

[0079] Then, attention is allocated based on the set of operating condition bias feature vectors. During this process, within the historical operating data range, the fluctuation amplitude of partial discharge intensity caused by changes in operating-related bias, environmental-related bias, and load-related bias at each spatial location over time is statistically analyzed. This fluctuation amplitude is obtained by calculating the average absolute value of the difference in partial discharge intensity before and after the bias change. Subsequently, the fluctuation amplitudes corresponding to the three types of biases are compared. A larger fluctuation amplitude indicates a more significant impact on the partial discharge measurement results, and a higher weight value is assigned to this bias; a smaller fluctuation amplitude is assigned a lower weight value. Next, the three weight values ​​are normalized so that the sum of the three weight values ​​equals 1, thus obtaining the multi-condition bias attention relationship used to characterize the degree of influence of operating conditions, environmental conditions, and load conditions on the partial discharge results.

[0080] After establishing the multi-condition bias attention relationship, the bias feature vector set is weighted and calculated based on this relationship. Specifically, for each spatial location, the operation-related bias value, environment-related bias value, and load-related bias value at that location are multiplied by their corresponding weight values ​​to obtain three weighted bias components. These three weighted bias components are then summed to obtain the bias correction amount corresponding to that spatial location under the current operating conditions. This bias correction amount is used to quantify the overall offset caused by the superposition of multiple operating conditions on the partial discharge measurement results.

[0081] Finally, the calculation rules for calculating the offset correction for each spatial location are unified and solidified, so that it can receive the operation-related offset value, environment-related offset value, and load-related offset value at any time point and output the corresponding offset correction, thereby forming a multi-condition offset coupling correction model.

[0082] Step S600: Generate a partial discharge early warning command based on the fourth partial discharge thermal map.

[0083] In this embodiment of the application, when generating a partial discharge early warning command based on the fourth partial discharge thermal map, the fourth partial discharge thermal map is first used as the direct basis for determining the partial discharge state. The corrected partial discharge intensity value corresponding to each spatial position in the fourth partial discharge thermal map is read. The corrected partial discharge intensity value is used to characterize the actual level of partial discharge activity after eliminating the effects of power frequency phase interference, multi-physics field coupling interference and dynamic bias of the operating condition.

[0084] Subsequently, the partial discharge intensity values ​​at each spatial location were continuously collected and statistically analyzed over time. The variation of partial discharge intensity at the same spatial location within a continuous time window was analyzed, and the maximum and average values ​​of the partial discharge intensity at that spatial location within the time window were calculated to reflect the persistence of partial discharge activity. Next, the maximum or average value of the partial discharge intensity was compared with a pre-set warning threshold. When the partial discharge intensity value at any spatial location exceeds the corresponding warning threshold, it is determined that there is a risk of abnormal partial discharge at that spatial location.

[0085] Finally, after detecting the risk of partial discharge anomaly, a partial discharge early warning command is generated to indicate that there is a partial discharge anomaly in the medium-voltage switchgear. The partial discharge early warning command includes at least the spatial location identifier that triggered the warning and the corresponding time information, and is used to issue a partial discharge early warning prompt.

[0086] In summary, the embodiments of this application have at least the following technical effects:

[0087] This application loads a partial discharge log library of a medium-voltage switchgear and dynamically adjusts the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear based on the partial discharge log library to obtain an optimized partial discharge sensing architecture. Based on the optimized partial discharge sensing architecture, the application performs online partial discharge monitoring and adaptive gain adjustment of the medium-voltage switchgear to obtain a first partial discharge heatmap. Based on the real-time voltage sequence of the medium-voltage switchgear, the application performs real partial discharge correction under power frequency phase correlation analysis on the first partial discharge heatmap to obtain a second partial discharge heatmap. Based on the cabinet characteristic dataset of the medium-voltage switchgear, the application performs coupling interference correction under multi-field joint conditions on the second partial discharge heatmap to obtain a third partial discharge heatmap. Based on the real-time operating condition data stream of the medium-voltage switchgear, the application performs dynamic bias correction of the operating condition influence on the third partial discharge heatmap to obtain a fourth partial discharge heatmap. Based on the fourth partial discharge heatmap, a partial discharge early warning command is generated. This invention addresses the technical problems of existing partial discharge online measurement results being susceptible to interference and lacking authenticity and accuracy. By performing multi-stage correction processing on partial discharge monitoring results, it achieves the technical effect of improving the accuracy of online partial discharge measurement and the reliability of early warning.

[0088] Example 2, based on the same inventive concept as the online measurement and correction method for partial discharge in medium-voltage switchgear in the previous examples, such as... Figure 2As shown, this application provides an online measurement and correction system for partial discharge in medium-voltage switchgear. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0089] The dynamic correlation adjustment module 11 is used to load the partial discharge log library of the medium-voltage switchgear and dynamically correlate and adjust the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear according to the partial discharge log library to obtain the optimized architecture of the partial discharge sensing; the monitoring and adjustment module 12 is used to perform online partial discharge monitoring and adaptive gain adjustment of the medium-voltage switchgear according to the optimized architecture of the partial discharge sensing to obtain the first partial discharge heat map; the real partial discharge correction module 13 is used to perform power frequency phase correction on the first partial discharge heat map according to the real-time voltage sequence of the medium-voltage switchgear. The system employs a correlation analysis method to correct for real partial discharge (PD) and obtain a second PD heatmap. A coupling interference correction module 14 performs multi-field coupling interference correction on the second PD heatmap based on the cabinet characteristic dataset of the medium-voltage switchgear, obtaining a third PD heatmap. A dynamic bias correction module 15 performs dynamic bias correction on the third PD heatmap based on the real-time operating condition data stream of the medium-voltage switchgear, obtaining a fourth PD heatmap. An instruction generation module 16 generates a PD warning instruction based on the fourth PD heatmap.

[0090] Furthermore, the system is also used to implement the following functions:

[0091] The partial discharge log database is used to classify location characteristics and obtain multiple partial discharge event sets corresponding to multiple historical partial discharge locations. The partial discharge sensitivity of the multiple historical partial discharge locations is evaluated based on the multiple partial discharge event sets to obtain a location partial discharge sensitivity characteristic distribution. Based on the location partial discharge sensitivity characteristic distribution, correlation adjustment decisions are made for the partial discharge sensor array layer to obtain a first partial discharge sensor adjustment strategy. Based on the first partial discharge sensor adjustment strategy, dynamic mapping adjustment compensation of the sensor gain adjustment layer is executed to obtain a second partial discharge sensor adjustment strategy. Based on the second partial discharge sensor adjustment strategy, multi-layer collaborative adjustment is performed on the dual-layer partial discharge sensor architecture to generate the optimized partial discharge sensor architecture.

[0092] Furthermore, the system is also used to implement the following functions:

[0093] The real-time voltage sequence is subjected to power frequency phase synchronization processing to establish a voltage cycle phase reference model; the partial discharge pulse signals in the first partial discharge heatmap are phase mapped with the voltage cycle phase reference model to obtain the power frequency phase distribution characteristics; the partial discharge pulse signals and the power frequency phase distribution characteristics are subjected to repeatability identification to obtain the associated repeatability characteristic distribution; the partial discharge pulse signals and the power frequency phase distribution characteristics are subjected to consistency identification to obtain the associated consistency characteristic distribution; the pseudo partial discharge signal is suppressed in the first partial discharge heatmap according to the associated repeatability characteristic distribution and the associated consistency characteristic distribution to generate the second partial discharge heatmap.

[0094] Furthermore, the system is also used to implement the following functions:

[0095] Based on the cabinet characteristic dataset, a multiphysics cabinet coupling model is constructed; based on the multiphysics cabinet coupling model, the partial discharge process is inverted on the second partial discharge heat map to obtain multiple partial discharge inversion process data; coupling interference detection is performed on each partial discharge inversion process data to determine each partial discharge coupling interference component; based on each partial discharge coupling interference component, the second partial discharge heat map is mapped and corrected to generate the third partial discharge heat map.

[0096] Furthermore, the system is also used to implement the following functions:

[0097] Multi-field coupled partial discharge interference records of the medium-voltage switchgear are collected to obtain the coupled interference detection history space; multiple coupled interference detection models are trained based on the coupled interference detection history space; the multiple coupled interference detection models are integrated and fused for training to obtain the coupled interference detection channel; the data of each partial discharge inversion process are input into the coupled interference detection channel to obtain the coupled interference components of each partial discharge.

[0098] Furthermore, the system is also used to implement the following functions:

[0099] Based on the real-time operating condition data stream, extract the operation status data stream, environmental operating condition data stream, and load operating condition data stream; analyze the operation-related bias influence of the third partial discharge heatmap based on the operation status data stream to obtain the operation-related bias characteristic distribution; analyze the environmental-related bias influence of the third partial discharge heatmap based on the environmental operating condition data stream to obtain the environmental-related bias characteristic distribution; analyze the load-related bias influence of the third partial discharge heatmap based on the load operating condition data stream to obtain the load-related bias characteristic distribution; construct a multi-operating condition bias coupling correction model based on the operation-related bias characteristic distribution, the environmental-related bias characteristic distribution, and the load-related bias characteristic distribution; perform collaborative correction on the third partial discharge heatmap based on the multi-operating condition bias coupling correction model to obtain the fourth partial discharge heatmap.

[0100] Furthermore, the system is also used to implement the following functions:

[0101] The operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution are subjected to unified spatial mapping and scale normalization to obtain a set of operating condition bias feature vectors; attention is allocated according to the set of operating condition bias feature vectors to establish multi-operating condition bias attention relationships; attention-driven coupling is performed on the set of operating condition bias feature vectors according to the multi-operating condition bias attention relationships to generate the multi-operating condition bias coupling correction model.

[0102] Furthermore, the system is also used to implement the following functions:

[0103] The partial discharge sensing dual-layer architecture includes a partial discharge sensing array layer and a sensing gain adjustment layer. The partial discharge sensing array layer includes multiple partial discharge sensors deployed in the medium-voltage switchgear. The sensing gain adjustment layer includes multiple gain regulators. The partial discharge sensing array layer and the sensing gain adjustment layer are bidirectionally connected through data mapping and feedback control relationships.

[0104] Furthermore, the system is also used to implement the following functions:

[0105] The cabinet characteristic dataset includes cabinet structural characteristic data, cabinet material characteristic data, internal space distribution data, and sensor deployment data of the medium-voltage switchgear.

[0106] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0107] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for online measurement and correction of partial discharge in medium-voltage switchgear, characterized in that, The method includes: The partial discharge log library of the medium-voltage switchgear is loaded, and the partial discharge sensing dual-layer architecture of the medium-voltage switchgear is dynamically correlated and adjusted according to the partial discharge log library to obtain the optimized partial discharge sensing architecture. This includes: classifying the location characteristics according to the partial discharge log library, obtaining multiple partial discharge event sets corresponding to multiple historical partial discharge locations, evaluating the partial discharge sensitivity of the multiple historical partial discharge locations according to the multiple partial discharge event sets, obtaining the location partial discharge sensitivity characteristic distribution, making correlation adjustment decisions for the partial discharge sensing array layer according to the location partial discharge sensitivity characteristic distribution, obtaining a first partial discharge sensing adjustment strategy, executing dynamic mapping adjustment compensation of the sensing gain adjustment layer according to the first partial discharge sensing adjustment strategy, obtaining a second partial discharge sensing adjustment strategy, and performing multi-layer collaborative adjustment of the partial discharge sensing dual-layer architecture according to the second partial discharge sensing adjustment strategy to generate the optimized partial discharge sensing architecture. Based on the partial discharge sensing optimization architecture, the medium-voltage switchgear is subjected to online partial discharge monitoring and adaptive gain adjustment to obtain the first partial discharge thermal map. The partial discharge first heat map is corrected for actual partial discharge under power frequency phase correlation analysis based on the real-time voltage sequence of the medium-voltage switchgear to obtain the partial discharge second heat map; Based on the cabinet characteristic dataset of the medium-voltage switchgear, the partial discharge second thermal map is subjected to multi-field joint coupling interference correction to obtain the partial discharge third thermal map; The partial discharge third heat map is dynamically biased and corrected based on the real-time operating condition data stream of the medium-voltage switchgear to obtain the partial discharge fourth heat map. Based on the fourth partial discharge heat map, a partial discharge early warning command is generated.

2. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 1, characterized in that, Based on the real-time voltage sequence of the medium-voltage switchgear, the first partial discharge heatmap is corrected for actual partial discharge under power frequency phase correlation analysis to obtain the second partial discharge heatmap, including: The real-time voltage sequence is subjected to power frequency phase synchronization processing to establish a voltage period phase reference model; Phase mapping is performed between each partial discharge pulse signal in the first partial discharge heat map and the voltage period phase reference model to obtain the distribution characteristics of each power frequency phase. Repeatability identification is performed on each partial discharge pulse signal and each power frequency phase distribution feature to obtain the associated repeatability characteristic distribution; Consistency identification is performed on the distribution characteristics of each partial discharge pulse signal and each power frequency phase to obtain the associated consistency characteristic distribution; Based on the correlation repetitive characteristic distribution and the correlation consistent characteristic distribution, the pseudo partial discharge signal is suppressed in the first partial discharge heatmap to generate the second partial discharge heatmap.

3. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 1, characterized in that, Based on the cabinet characteristic dataset of the medium-voltage switchgear, the second partial discharge thermal map is subjected to multi-field joint coupling interference correction to obtain the third partial discharge thermal map, including: Based on the cabinet characteristic dataset, a multiphysics cabinet coupling model is constructed; Based on the multiphysics cabinet coupling model, the partial discharge process is inverted using the second partial discharge thermal map to obtain multiple partial discharge inversion process data. Coupled interferometry detection was performed on the data from each partial discharge inversion process to determine the coupled interferometry components of each partial discharge. The partial discharge second thermal map is mapped and corrected based on the partial discharge coupling interference components to generate the partial discharge third thermal map.

4. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 3, characterized in that, Coupled interferometry detection was performed on the data from each partial discharge inversion process to determine the coupled interferometry components of each partial discharge, including: Collect multi-field coupled partial discharge interference records of the medium-voltage switchgear to obtain the historical space of coupled interference detection; Multiple coupled interferometry detection models are trained based on the coupled interferometry detection history space; The multiple coupled interferometric detection models are integrated and fused for training to obtain the coupled interferometric detection channel; The data from each partial discharge inversion process are input into the coupled interference detection channel to obtain the coupled interference components of each partial discharge.

5. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 1, characterized in that, Based on the real-time operating condition data stream of the medium-voltage switchgear, the third partial discharge heat map is dynamically biased to correct for operating condition influence, and a fourth partial discharge heat map is obtained, including: Based on the real-time operating condition data stream, extract the operation status data stream, environmental operating condition data stream, and load operating condition data stream; Based on the operation state data stream, the operation-related bias influence of the third partial discharge heatmap is analyzed to obtain the operation-related bias characteristic distribution. Based on the environmental operating condition data stream, the environmental correlation bias influence of the third partial discharge heatmap is analyzed to obtain the distribution of environmental correlation bias characteristics. Based on the load condition data stream, the load-related bias influence of the third partial discharge heatmap is analyzed to obtain the load-related bias characteristic distribution. Based on the operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution, a multi-condition bias coupling correction model is constructed. The third partial discharge thermal map is collaboratively corrected based on the multi-condition bias coupling correction model to obtain the fourth partial discharge thermal map.

6. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 5, characterized in that, Based on the operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution, a multi-condition bias coupling correction model is constructed, including: The operation-related bias characteristic distribution, the environment-related bias characteristic distribution, and the load-related bias characteristic distribution are subjected to unified spatial mapping and scale normalization to obtain a set of operating condition bias feature vectors. Attention is allocated based on the set of working condition bias feature vectors to establish multi-working-condition bias attention relationships; Based on the multi-condition bias attention relationship, the set of condition bias feature vectors is coupled with attention to generate the multi-condition bias coupling correction model.

7. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 1, characterized in that, The partial discharge sensing dual-layer architecture includes a partial discharge sensing array layer and a sensing gain adjustment layer. The partial discharge sensing array layer includes multiple partial discharge sensors deployed in the medium-voltage switchgear. The sensing gain adjustment layer includes multiple gain regulators. The partial discharge sensing array layer and the sensing gain adjustment layer are bidirectionally connected through data mapping and feedback control relationships.

8. The online measurement and correction method for partial discharge in medium-voltage switchgear as described in claim 1, characterized in that, The cabinet characteristic dataset includes cabinet structural characteristic data, cabinet material characteristic data, internal space distribution data, and sensor deployment data of the medium-voltage switchgear.

9. A partial discharge online measurement and correction system for medium-voltage switchgear, characterized in that, The system is used to execute the online measurement and correction method for partial discharge in medium-voltage switchgear as described in any one of claims 1-8, and the system includes: The dynamic correlation adjustment module is used to load the partial discharge log library of the medium-voltage switchgear and dynamically correlate and adjust the dual-layer architecture of the partial discharge sensing of the medium-voltage switchgear according to the partial discharge log library to obtain the optimized architecture of the partial discharge sensing. The monitoring and adjustment module is used to perform online partial discharge monitoring and adaptive gain adjustment on the medium-voltage switchgear according to the partial discharge sensing optimization architecture, and to obtain the first partial discharge thermal map. The real partial discharge correction module is used to perform real partial discharge correction on the first partial discharge heat map based on the real voltage sequence of the medium voltage switchgear under power frequency phase correlation analysis, and obtain the second partial discharge heat map. The coupling interference correction module is used to perform multi-field joint coupling interference correction on the partial discharge second thermal map based on the cabinet characteristic dataset of the medium-voltage switchgear, and obtain the partial discharge third thermal map. The dynamic bias correction module is used to perform dynamic bias correction on the third partial discharge thermal map based on the real-time operating condition data stream of the medium-voltage switchgear, and obtain the fourth partial discharge thermal map. The instruction generation module is used to generate a partial discharge early warning instruction based on the fourth partial discharge thermal map.

Images