A partial discharge real-time monitoring system and method for a substation switch cabinet

Abstract

The application relates to the technical field of substation monitoring, and discloses a partial discharge real-time monitoring system and method for a switch cabinet of a substation. The method captures original signal data sets containing time stamp sequences and spectrum distributions generated by partial discharge in the switch cabinet; processes the sets to segment discharge pulses and background noise components, calculates the rise time, energy integral value and frequency bandwidth of each discharge pulse, and forms a discharge characteristic parameter set; identifies potential discharge risk modes based on the parameter set, generates dynamic configuration instructions containing monitoring priority sequencing and parameter adaptive requirements; calls the instructions to modify the sampling interval and detection sensitivity of a monitoring unit in real time, executes an optimized signal acquisition process; integrates real-time data streams in the optimized acquisition process, updates the discharge characteristic parameter set, and iteratively generates new dynamic configuration instructions. The method can improve the comprehensiveness, accuracy and real-time performance of partial discharge monitoring, and adapts to the complex operating environment of the switch cabinet.

Description

Technical Field

[0001] This invention relates to the field of substation monitoring technology, specifically to a real-time monitoring system and method for partial discharge in substation switchgear. Background Technology

[0002] As a critical piece of equipment in the power system, the operating status of substation switchgear directly affects the stability and security of the power network. During long-term operation, partial discharge can easily occur inside the switchgear due to factors such as insulation aging, loose components, and changes in environmental humidity. This discharge phenomenon is often weak in the early stages, but it can gradually erode the insulation material. If it is not detected and dealt with in time, it may lead to insulation breakdown, equipment failure, or even large-scale power outages.

[0003] Currently, monitoring methods for partial discharge in switchgear have many limitations. Traditional offline detection methods require interrupting equipment operation and conducting periodic preventative tests, which not only affects the continuity of power supply but also makes it difficult to capture intermittent or sudden discharge signals. While online monitoring technology avoids downtime issues, existing monitoring systems mostly use a single sensor for signal acquisition, which is susceptible to signal distortion or missed detections due to the complex electromagnetic environment inside the switchgear.

[0004] Existing monitoring methods often employ fixed analysis parameters in the signal processing stage, failing to dynamically adjust based on the actual characteristics of the discharge signal. This results in low accuracy in identifying different types of discharge signals. Furthermore, in terms of discharge risk assessment, the lack of an effective prioritization mechanism makes it difficult to quickly locate high-risk discharge points, hindering maintenance personnel from taking timely and targeted measures and increasing the potential risk of equipment failure.

[0005] As power systems evolve towards intelligence and automation, higher demands are placed on the real-time performance, accuracy, and reliability of partial discharge monitoring in switchgear. How to achieve efficient acquisition, accurate analysis, and dynamic early warning of partial discharge signals has become a key technical challenge in the field of power equipment condition monitoring. Summary of the Invention

[0006] The purpose of this invention is to provide a real-time monitoring system and method for partial discharge in substation switchgear, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a method for real-time monitoring of partial discharge in substation switchgear, the method comprising:

[0008] A set of raw signal data generated by partial discharge inside the switch cabinet is captured by a distributed sensor array. The set of raw signal data includes a timestamp sequence and a spectral distribution.

[0009] The original signal data set is processed to separate the discharge pulses from the background noise components, and the rise time, energy integral value and frequency bandwidth of each discharge pulse are calculated to form a set of discharge characteristic parameters.

[0010] Based on the set of discharge characteristic parameters, potential discharge risk modes are identified, and dynamic configuration instructions, including monitoring priority ranking and parameter adaptation requirements, are generated.

[0011] The dynamic configuration command is invoked to modify the sampling interval and detection sensitivity of the monitoring unit in real time, and to execute the optimized signal acquisition process.

[0012] The system integrates and optimizes the real-time data stream during the acquisition process, updates the set of discharge characteristic parameters, and iteratively generates new dynamic configuration instructions.

[0013] Preferably, the original signal data set includes pulse waveform index number, interference type classification code, and sensor spatial mapping position; the discharge characteristic parameter set includes pulse energy accumulation, frequency offset characteristic value, and discharge event time distribution density; the dynamic configuration instructions include monitoring channel sorting weight, sampling parameter adjustment range, and control effective time window; and the optimized acquisition process includes signal reconstruction trigger state, noise suppression start time, and pulse integrity identification state.

[0014] Preferably, the acquisition of the original signal data set specifically involves: collecting partial discharge signals inside the switchgear using a distributed sensor array; extracting the electric field intensity change and time delay sequence before and after each discharge event; calculating the electric field offset distance and time delay period based on the electric field intensity difference and time sequence length for each event, and generating an offset and delay feature set; based on the electric field offset distance and time delay period in the offset and delay feature set, and combined with the amplitude change information of the discharge event at consecutive moments before and after the event, counting the number of pulse attenuations in the offset and delay stages of each discharge event, identifying the correlation distribution pattern between the number of attenuations and the delay period, and generating discharge event attenuation frequency features; calling the discharge event attenuation frequency features, determining discharge events whose attenuation frequency exceeds a preset discharge behavior critical threshold, associating and binding the corresponding event type with the location of the switchgear area, and generating the original signal data set.

[0015] Preferably, the generation operation of the discharge characteristic parameter set specifically involves: calling the switchgear area location identified in the original signal data set, obtaining three time data items for the corresponding location within the monitoring period: signal start time, first pulse trigger time, and tail pulse end time; calculating the first pulse response interval and tail pulse passage time difference respectively to generate a key passage time interval; based on the first pulse response interval and tail pulse passage time difference in the key passage time interval, calling the total number of pulses and pulse energy distribution within the monitoring period of the corresponding location, identifying the total length of the pulse queue and the queue density level to obtain pulse distribution density information; based on the pulse distribution density information, combined with the time interval distribution of the front, middle, and rear positions within the queue in the key passage time interval, calculating the pulse density offset characteristic value, identifying the position window of the density anomaly segment in the monitoring period, and generating a pulse density offset time period; calling the corresponding position point in the pulse density offset time period, evaluating the mapping relationship between the time window and the pulse release behavior segment within the monitoring period, filtering continuous segments with pulse release behavior offset, marking them as pulse behavior delay occurrence areas, and generating a discharge characteristic parameter set.

[0016] Preferably, the dynamic configuration instruction generation operation specifically involves: based on the monitoring channel number indicated in the discharge characteristic parameter set, extracting the duration of the tail pulse's continuous stagnation at the end of the signal segment and the release completion time point within two consecutive monitoring cycles of the channel; combining the time difference between the release completion time point and the signal end time point of the current cycle to obtain the channel tail segment release lag information; based on the channel tail segment release lag information, determining whether the release lag value exceeds the channel release completion benchmark threshold, filtering out channel numbers that have not completed release, and extracting the tail pulse stagnation time in the corresponding channel, sorting them according to the stagnation duration of the incomplete channels to generate a channel release stagnation priority sequence; calling the sorting information in the channel release stagnation priority sequence to configure additional monitoring time periods for the channels in sequence, adjusting the channel monitoring time window length within the total monitoring duration, recording the channel number and the corresponding adjusted monitoring duration, and generating a dynamic configuration instruction.

[0017] Preferably, the execution operation of the optimized acquisition process is as follows: calling the channel start configuration time value recorded in the dynamic configuration instruction, detecting the longitudinal distance distribution of the electric field in the pulse queue area of ​​the corresponding channel within the real-time monitoring period, identifying the continuous enhancement start time points of three adjacent pulses in the queue segment, and generating a queue enhancement start time sequence; based on the time points of the enhancement actions of the three adjacent pulses in the queue enhancement start time sequence, determining whether the enhancement time is earlier than the signal preparation switching time point within the channel monitoring time window; if the judgment condition is met, marking it as a synchronous enhancement state, associating the channel number with the status information, and executing the optimized acquisition process.

[0018] Preferably, the optimized acquisition process further includes additional operations: calling the marking period of the dynamic configuration instruction, filtering the trajectory sequence of the latter part of the tail pulse signal, comparing the pulse amplitude change trend with the tail duration, and if the amplitude continues to rise without reaching a stable state, calculating the time required to reach the predetermined release point and updating the tail control period to obtain the tail signal continuous regulation result; the tail signal continuous regulation result includes the remaining release path duration, the predicted tail pulse release completion period, and the recommended signal holding time window.

[0019] Preferably, the operation for obtaining the continuous control result of the tail segment signal specifically involves: calling the monitoring cycle number marked in the dynamic configuration instruction, filtering the trajectory data of the tail pulse in the later part of the signal under the corresponding cycle, extracting the continuous amplitude sequence and timestamp data from the start time of the tail segment to the end line of the pulse release, and generating the tail segment passage trajectory sequence; based on the amplitude sequence in the tail segment passage trajectory sequence, analyzing the amplitude change trend of the tail pulse within the duration of the tail segment signal, extracting the amplitude increase value of the tail segment of the sequence, and comparing it with the duration corresponding to the tail segment; if the amplitude continues to rise and has not entered the stable range, calculating the supplementary time required for the pulse to reach the predetermined release point, and generating the remaining release time of the tail segment; calling the supplementary time value required for the channel in the remaining release time of the tail segment, updating the real-time tail segment signal control cycle, correcting the originally configured tail segment signal end time point, resetting the control range of the channel tail segment, and obtaining the continuous control result of the tail segment signal.

[0020] Preferably, the present invention also includes a real-time partial discharge monitoring system for substation switchgear. The system is used to implement the real-time partial discharge monitoring method for substation switchgear described above. The system includes: a signal acquisition module that detects partial discharge behavior inside the switchgear, acquires the electric field offset position and time delay sequence of pulses, statistically analyzes the pulse attenuation frequency, and generates a raw signal data set; a discharge behavior analysis module that, based on the raw signal data set, identifies key pulse actions that cause signal interruption and binds these key pulse actions to location points, generating a set of discharge characteristic parameters; a signal delay calculation module that calls the discharge characteristic parameter set to obtain the signal start time, first pulse trigger time, and last pulse end time within the location point, calculates the time interval difference, and generates a dynamic configuration instruction; a monitoring time dynamic configuration module that, based on the dynamic configuration instruction, analyzes the completion status of channel release, calculates the stagnation duration of incomplete channels, reallocates the monitoring time window as needed, and executes an optimized acquisition process; and a signal execution module that calls the optimized acquisition process to detect and evaluate the longitudinal distance of the electric field and continuous enhancement points of the queued pulses, adjusts the signal execution time according to the pulse amplitude change trend and tail duration, and obtains the tail signal continuous control result.

[0021] Preferably, the system further includes a data interaction module: the output of the signal acquisition module is connected to the input of the discharge behavior analysis module to transmit the original signal data set; the output of the discharge behavior analysis module is connected to the input of the signal delay calculation module to transmit the discharge characteristic parameter set; the output of the signal delay calculation module is connected to the input of the monitoring time dynamic configuration module to transmit dynamic configuration instructions; the output of the monitoring time dynamic configuration module is connected to the input of the signal execution module to transmit parameters for optimizing the acquisition process; the output of the signal execution module is fed back to the data interaction module to update the original signal dataset and restart the signal acquisition module.

[0022] Compared with the prior art, the beneficial effects of the present invention are:

[0023] By using a distributed sensor array for signal acquisition, partial discharge signals inside the switchgear can be captured simultaneously from multiple locations, effectively expanding the monitoring range, reducing blind spots, and facilitating a comprehensive understanding of the discharge situation inside the equipment. Compared to a single sensor, the distributed arrangement reduces the impact of local electromagnetic interference on signal acquisition, improves the integrity and accuracy of the raw signal data, and provides a more reliable foundation for subsequent signal processing and analysis.

[0024] In the signal processing stage, by segmenting the discharge pulse from the background noise component and calculating characteristic parameters such as rise time, energy integral value, and frequency bandwidth, key information of the discharge signal can be extracted more accurately, reducing the impact of noise interference on the analysis results. This targeted characteristic parameter extraction method helps distinguish different types of discharge signals, enhances the ability to identify weak discharge signals, and enables the monitoring system to detect potential discharge phenomena earlier.

[0025] Based on a set of discharge characteristic parameters, potential discharge risk modes are identified, and dynamic configuration instructions are generated, enabling the monitoring system to adjust the sampling interval and detection sensitivity in real time according to the actual discharge situation. This adaptive adjustment mechanism can shorten the sampling interval and improve the temporal resolution of the data when the discharge signal is strong, while increasing the detection sensitivity when the signal is weak to avoid missed detections. By dynamically optimizing the signal acquisition process, the system can reasonably control the amount of data, reduce the computational load on the system, and improve the real-time performance of monitoring while ensuring monitoring effectiveness.

[0026] This method integrates and optimizes real-time data streams during the acquisition process, continuously updates the set of discharge characteristic parameters, and iteratively generates new dynamic configuration instructions, forming a closed-loop monitoring process. This iterative optimization mechanism enables the monitoring system to continuously adapt to changes in discharge signals and adjust monitoring strategies as the discharge state evolves, ensuring that the assessment and early warning of discharge risks remain highly accurate. Through this dynamic adjustment and iterative optimization, maintenance personnel can obtain the latest discharge risk information in a timely manner, quickly locate discharge points requiring attention based on priority, and facilitate the implementation of corresponding maintenance measures, reducing the possibility of equipment failure due to partial discharge.

[0027] This method does not require interrupting the normal operation of the switchgear and can achieve continuous monitoring while the equipment is energized, avoiding the impact on power supply caused by traditional offline monitoring. At the same time, the entire monitoring process is highly automated, reducing reliance on manual operation, minimizing errors caused by human factors, improving the efficiency and stability of monitoring work, and better meeting the intelligent requirements of modern power systems for equipment status monitoring. Attached Figure Description

[0028] Figure 1 This is a timing diagram of the real-time partial discharge monitoring method for substation switchgear described in this invention.

[0029] Figure 2 Flowcharts generated for data sets and instructions;

[0030] Figure 3 Flowchart generated for dynamically configured instructions;

[0031] Figure 4 To optimize the flowchart of the data acquisition process;

[0032] Figure 5 This is a flowchart for continuous modulation of the tail segment signal. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 This invention provides a method for real-time monitoring of partial discharge in substation switchgear, the method comprising:

[0035] A distributed sensor array deployed inside the switchgear continuously captures raw signal data sets generated by partial discharges. This raw signal data set includes timestamp sequences and spectral distribution information. The timestamp sequences record the occurrence time of each discharge event, while the spectral distribution covers the spectral characteristics from low to high frequencies. The sensor array is arranged in a distributed manner, covering different areas of the switchgear to ensure comprehensive signal coverage. The signal acquisition process begins with a fixed initial sampling frequency and sensitivity threshold.

[0036] When processing the original signal data set, a digital signal processor (DSP) is used to segment the discharge pulses and background noise components. The segmentation process is based on a preset threshold filtering algorithm: the background noise component is extracted using a low-pass filter with a bandwidth range of 0-100kHz, while the discharge pulses are identified by their signal strength exceeding the environmental noise baseline. After pulse segmentation, each discharge event is assigned an independent time window. For each segmented discharge pulse, the rise time, energy integral value, and frequency bandwidth are calculated. The rise time is defined as the time difference from the baseline to the pulse peak value. The energy integral value is calculated by integrating the pulse waveform in the time domain. The frequency bandwidth is measured by extracting the spectral envelope using a fast Fourier transform and then measuring the width of the main frequency band. The calculation results form a set of discharge characteristic parameters, which are stored in a buffer memory.

[0037] Based on the set of discharge characteristic parameters, a pattern recognition algorithm is used to identify potential discharge risk patterns. The pattern recognition process includes: comparing the rise time of multiple discharge events with a preset safety threshold to determine the hazard level of the discharge events; analyzing the cumulative distribution of energy integral values ​​to generate a risk probability matrix; and calculating the coefficient of variation of the frequency bandwidth to assess the consistency of discharge types. The identification results generate dynamic configuration instructions, including monitoring priority ranking and parameter adaptation requirements. The monitoring priority ranking assigns priority weight values ​​to monitoring units based on the risk probability matrix, while the parameter adaptation requirements define the adjustment range values ​​for the sampling interval and detection sensitivity.

[0038] When the dynamic configuration command is invoked, the command is sent to the control module of the monitoring unit. The module modifies the sampling interval and detection sensitivity in real time. The sampling interval value is dynamically shortened or extended according to priority weights, and the detection sensitivity value is adjusted by a gain amplifier. After modification, an optimized signal acquisition process is executed. The process uses the new sampling interval to start data stream acquisition, and simultaneously applies the adjusted sensitivity to filter interference signals. The real-time data stream from the optimized acquisition process is integrated: the newly acquired data is input into the signal processor to update the rise time, energy integral value, and frequency bandwidth in the discharge characteristic parameter set; after the update, the pattern recognition algorithm is re-run to generate a new dynamic configuration command. The entire process is executed iteratively, and data updates and command generation form a closed-loop feedback mechanism to achieve continuous monitoring and adaptive adjustment.

[0039] Example 1: See Figure 2 The acquisition of the raw signal data set is achieved through a distributed sensor array. The sensor array consists of multiple electric field strength sensors, which are spatially distributed and installed at different locations within the switchgear, including the high-voltage side, low-voltage side, busbar connections, and near the insulation support structure. Each sensor has an independent signal acquisition channel, capable of capturing real-time electric field changes caused by partial discharge. The arrangement of the sensor array ensures coverage of all areas within the switchgear where partial discharge may occur, avoiding monitoring blind spots. The sensors employ a high-frequency response design, with a sampling frequency set to 10MHz to meet the requirements for capturing rapid discharge pulses. The analog signals output by the sensors are amplified by a preamplifier and then converted into digital signals by a high-speed analog-to-digital converter, forming the raw signal data stream.

[0040] The raw signal data set includes pulse waveform index numbers, interference type classification codes, and sensor spatial mapping locations. Pulse waveform index numbers uniquely identify each captured discharge pulse, with the numbering rule based on an incremental time-sequence allocation. Each time a discharge event is triggered, the system assigns a unique index number, which is bound to the pulse's timestamp to form an index sequence. Interference type classification codes distinguish different types of background noise, using a pattern matching algorithm. This algorithm compares the spectrum of the captured noise signal with a pre-set noise feature library containing typical spectral templates for power frequency interference, electromagnetic interference, and random noise. The matching results generate corresponding classification codes in numeric identifier format, facilitating rapid noise identification and filtering by subsequent processing modules. Sensor spatial mapping locations record the physical coordinates of each sensor within the switch cabinet. The coordinate information is read from a pre-configured three-dimensional spatial location table, stored in non-volatile memory, ensuring that the sensor layout information is correctly loaded after a system restart.

[0041] The specific steps for acquiring the raw signal data set include: acquiring partial discharge signals inside the switchgear using a distributed sensor array. Each electric field strength sensor in the sensor array monitors the electric field changes at its location in real time. When the detected electric field strength exceeds a preset threshold, the signal acquisition process is triggered. The acquired signals include the change in electric field strength before and after the discharge event and a time delay sequence. The change in electric field strength is calculated by measuring the instantaneous difference in the sensor output voltage; this difference reflects the dynamic amplitude of the electric field strength change. The time delay sequence records the time interval from triggering to ending the discharge event. The timestamp is generated by a high-precision system clock with nanosecond-level synchronization accuracy, ensuring the accuracy of the timing data.

[0042] Based on the electric field intensity difference and time series length, the electric field offset distance and time delay period are calculated. The electric field offset distance characterizes the spatial propagation range of the discharge event. The calculation process utilizes the physical distance between sensors and the relative relationship between the electric field changes, estimating the positional offset of the discharge source through a spatial interpolation algorithm. The time delay period reflects the duration of the discharge event and is obtained by analyzing the difference between the start and end time points in the timestamp sequence. The calculation results are combined with amplitude change information before and after the discharge event. The amplitude change information is extracted from the signal waveform, and the peak sequence reflects the intensity evolution of the discharge pulse.

[0043] The number of pulse attenuation events during the offset and delay phases of each discharge event is statistically analyzed. The attenuation count is defined as the number of times the discharge pulse amplitude drops from its peak value to a specific percentage (e.g., 50%). The statistical process uses a sliding window algorithm to scan the amplitude sequence and identify points that meet the attenuation criteria. The correlation between the attenuation count and the time delay period is established through statistical analysis. The analysis process calculates the correlation between the attenuation count and the delay period, generating a distribution model describing their relationship. The model output includes attenuation frequency characteristics, which reflect the energy release pattern of the discharge event.

[0044] The system invokes the attenuation frequency characteristics of the discharge event to determine if the attenuation frequency exceeds a preset critical threshold for discharge behavior. This critical threshold, set based on historical data and equipment operating experience, distinguishes between normal and abnormal discharge behaviors. When the attenuation frequency exceeds the threshold, the system determines that the discharge event poses a potential risk and requires further analysis. The determination result is associated with the location of the switchgear area where the discharge event occurred. This association is achieved by querying a sensor spatial mapping table, combining the event type code with the location coordinates to form a complete set of raw signal data. The data set is stored in a structured format, containing timestamps, spectrum data, pulse indexes, interference codes, and location information, for subsequent processing modules to access.

[0045] The generation process of the discharge characteristic parameter set is based on key information from the original signal data set. The set includes pulse energy accumulation, frequency offset characteristic values, and discharge event temporal distribution density. Pulse energy accumulation is calculated by time-domain integration of the discharge pulse waveform, with the integration interval covering the entire pulse duration; the result reflects the total energy released by the discharge event. Frequency offset characteristic values ​​are extracted from the spectral analysis of the pulse signal. The analysis process uses Fast Fourier Transform to convert the time-domain signal into a frequency-domain representation, identifies the dominant frequency component, and calculates its offset from the reference frequency. The discharge event temporal distribution density statistically analyzes the number of discharge events occurring within a unit time window; the density value reflects the frequency of partial discharge activity.

[0046] The processing module integrates the above parameters into a discharge characteristic parameter set, which is stored in an array structure. Each parameter is associated with a corresponding pulse index number and sensor location information. The set data is transmitted to the pattern recognition module via a shared memory interface for subsequent risk pattern analysis and dynamic configuration command generation. The entire implementation process adopts a modular design, with each functional module communicating through standardized interfaces to ensure efficient data flow and real-time processing capabilities. During system operation, the original signal data set and the discharge characteristic parameter set are continuously updated, forming a closed-loop feedback mechanism to dynamically adapt to changes in partial discharge behavior within the switchgear.

[0047] Example 2: See Figure 3 The process of generating a set of discharge characteristic parameters begins by retrieving the switchgear area location information identified in the original signal data set. These location coordinates are read from a spatial mapping table loaded by the system from non-volatile storage units, with each location corresponding to a unique sensor identifier. For each specific location, the system retrieves three core time parameters for that point within the current monitoring period from the time-series database: the signal start time records the moment when the discharge behavior was first detected; the first pulse trigger time marks the precise time point when the amplitude of the first discharge pulse exceeds a preset threshold; and the tail pulse end time marks the timestamp when the last pulse signal falls back to the baseline level. These three time data are stored in a circular buffer with millisecond-level precision.

[0048] Based on the three time parameters mentioned above, the system performs difference calculations to generate critical time intervals for passage. The first pulse response interval is calculated by subtracting the signal start time from the first pulse trigger time, reflecting the response delay characteristics of the discharge event from initial detection to the first significant pulse. The tail pulse passage time difference is obtained by subtracting the average duration of all pulses within the monitoring period from the tail pulse end time, characterizing the degree of persistence of the pulse group at the end of the entire event. The calculation results are stored in the form of time intervals, including two dimensions: the start time difference and the end time difference.

[0049] The system further retrieves the total number of pulses and pulse energy distribution data for the corresponding location within the current monitoring period. The total number of pulses is accumulated by a high-speed counter during the event duration, and the value is updated to the register in real time. The pulse energy distribution is read from the output of the energy integrator, containing the amplitude integral sequence of each pulse. Using this data, the system identifies pulse queue characteristics: the total queue length is defined as the time span from the trigger time of the first pulse to the end time of the last pulse; the queue density level is obtained by dividing the total number of pulses by the total queue length to obtain the frequency value of pulse occurrence within a unit time window. These two parameters constitute a pulse distribution density information array, which is stored indexed by sensor location.

[0050] By combining pulse distribution density information with key time intervals for passage, the system analyzes the temporal distribution characteristics of the front, middle, and rear positions in the queue. Position selection is based on time proportions: the front position is set at 25% of the queue time, the middle position at 50%, and the rear position at 75%. The system extracts the pulse frequency within fixed time windows surrounding these three positions and calculates the density value for each window. The pulse density offset feature value is generated by calculating the standard deviation of the density values ​​from the three windows; this feature value quantifies the dispersion of the pulse sequence along the time axis. When the standard deviation exceeds a preset threshold, the system marks this time period as a density anomaly segment, and its start and end times are recorded in a structured array of pulse density offset time periods.

[0051] The system retrieves the coordinates of the locations recorded during the pulse density offset time period to establish a mapping relationship between the time window and the pulse release behavior segment. This mapping process is accomplished by querying a space-time association table, which stores the pulse activity identifier for each location point on the time axis. The system scans the time window corresponding to the density anomaly segment, detecting whether there are consecutive segments with pulse trigger intervals exceeding the normal range. These segments are identified as pulse behavior delay areas, with characteristic parameters including delay duration, location, and the number of pulses contained within. These parameters, along with the previously calculated critical passage time interval and pulse density offset characteristic values, constitute a set of discharge characteristic parameters, which is transmitted to the instruction generation module via the data bus.

[0052] The process of generating dynamic configuration instructions begins with parsing the set of discharge characteristic parameters. The system reads the list of monitoring channel numbers identified in the set, with each number corresponding to a physical sensor's connection channel. For each channel number, the system retrieves the tail pulse data from the historical database for the two most recent consecutive monitoring cycles. The data includes the duration of the tail pulse's sustained pause at the end of the signal segment, i.e., the duration during which the final pulse amplitude remains unchanged; and the release completion time, defined as the moment when the tail pulse amplitude finally stabilizes at the baseline level. The system calculates the time difference between the release completion time and the preset signal end time for the current cycle; this difference is called the tail release hysteresis value, with a negative value indicating that the release was not completed on time. The calculation results form a channel tail release hysteresis information dataset, with one hysteresis value record corresponding to each channel.

[0053] Based on the hysteresis information dataset, the system compares the hysteresis value of each channel with a preset release completion benchmark threshold. The benchmark threshold is pre-configured in the system parameter table according to the device type. When the hysteresis value exceeds the threshold, the system determines that the channel has an incomplete release state and adds its number to the list of channels to be processed. For each channel in the list, the system extracts the actual measured value of its tail pulse stagnation duration. Then, a heap sort algorithm is used to sort all incomplete channel numbers in ascending order of stagnation duration, generating a channel release stagnation priority sequence. The sorting result is stored in a priority queue data structure, with the first element corresponding to the channel with the shortest stagnation duration.

[0054] The system calls the priority sequence ranking result and allocates supplementary monitoring time to each channel in turn. The calculation rule for the supplementary duration is to multiply the stagnation time by a dynamic coefficient, which is automatically adjusted according to the current overall resource utilization rate of the system. Under the constraint of a constant total monitoring cycle, the system dynamically adjusts the monitoring time window of each channel through the time resource allocation module: first, it reduces the time window length of channels with no stagnation or less stagnation, and allocates the freed-up time resources to channels in the priority sequence. A mapping table is formed between the channel number and the reconfigured monitoring duration, and the table records the newly set start and end times. This mapping table is encapsulated as a dynamic configuration command package, with a timestamp and version identifier attached, and sent to the signal acquisition unit via the control bus.

[0055] Example 3: See Figure 4 The implementation begins by invoking the channel start configuration time value recorded in the dynamic configuration instruction. This time value is extracted from the instruction data packet, which is a structured data array containing a timestamp field and a channel identifier field. The system reads the instruction packet via the control bus, uses a depacketizing algorithm to parse the start configuration time value, and stores it in the real-time clock register. The depacketizing process involves byte alignment operations and data type conversion to ensure that the time value's accuracy is maintained at the microsecond level. The channel start configuration time value defines the start time of the monitoring cycle, serving as the time reference point for subsequent detection operations. During the real-time monitoring cycle, the system activates the data acquisition hardware for the corresponding channel to acquire signal data from the pulse queue area. The pulse queue area refers to the physical location within the sensor coverage space where discharge pulses occur continuously. The area boundary is defined based on a pre-configured switchgear coordinate system, with coordinate values ​​loaded from a spatial mapping table.

[0056] The detection operation focuses on the longitudinal distance distribution of the electric field within the pulse queue region. The longitudinal distance characterizes the spatial variation of the electric field in the vertical direction. The distribution detection process utilizes a multi-point electric field strength sensor array: sensors are mounted at specific intervals on vertical supports of the switchgear, and each sensor measures the instantaneous value of the electric field strength at its location. The longitudinal distance is calculated as a function of the height difference between sensors and the electric field strength gradient. The system captures readings from multiple sensors in real time and calculates the longitudinal distance sequence using the difference. These numerical sequences are formed in matrix form, with rows corresponding to time points and columns corresponding to sensor pair numbers. The fluctuations in the longitudinal distance distribution reflect the aggregation or dispersion of the pulse group in the vertical spatial dimension; the system analyzes this distribution to determine the directional trend of pulse activity.

[0057] The system identifies the starting time points of continuous enhancement for three adjacent pulses within a queue segment. Continuous enhancement is defined as the phenomenon where the pulse amplitude sequence exhibits a monotonically increasing characteristic and the slope exceeds a set threshold. A queue segment refers to an ordered sequence of pulses within a continuous time window. The system traverses the pulse queue region and identifies the pulse amplitude data stream: amplitude data is captured from the output of the high-speed analog-to-digital converter and stored in a circular buffer. For each pulse, the system calculates its amplitude change rate, and the rate of change outputs a slope value based on a time differential algorithm. The starting time point is determined as the precise moment when the amplitude begins to rise and first exceeds the rising slope threshold. The selection rule for the three adjacent pulses is based on time order: the system scans the pulse index list, taking the pulses at the current index, the next index, and the two indices below the next, forming a three-pulse group. The identification process generates a starting time point dataset for each group, with the dataset format being a timestamp array. This array is called the queue enhancement starting time sequence, and the sequence elements include three time values: the starting time of the first pulse. The start time of the second pulse The start time of the third pulse After the sequence is generated, it is stored in a shared memory area, and the index is mapped by the pulse group number.

[0058] Based on the queue enhancement start time sequence, the system performs a judgment to determine whether the synchronous enhancement state condition is met. The judgment is based on a comparison between the enhancement time point and the signal preparation switching time point within the channel monitoring time window. The signal preparation switching time point is loaded from a preset parameter table; this time point identifies the moment when the system plans to switch channels or adjust sampling at the end of the monitoring cycle. The judgment condition is defined as: all three enhancement start time points are strictly earlier than the signal preparation switching time point. In implementation, the time point comparison uses a formula to calculate the judgment logic. The formula is expressed as:

[0059]

[0060] in: The enhancement start time point for the first pulse in the pulse group is taken from the first element of the queue enhancement start time sequence, and the unit is microseconds; The enhancement start time point for the second pulse in the pulse group is taken from the second element of the queue enhancement start time sequence, and the unit is microseconds; The enhancement start time point of the third pulse in the pulse group is taken from the third element of the queue enhancement start time sequence, and the unit is microseconds; The switching time point is prepared for the signal within the channel monitoring time window. The value is read from the system global clock module and marks a specific moment before the end of the monitoring cycle. The unit is microseconds. This is a logical operator representing the "AND" operation, which requires all conditions to be met simultaneously.

[0061] The formula is implemented in a hardware comparator circuit: the comparator receives the time value input, performs parallel subtraction, and compares the result with zero; a negative result indicates that the time point is earlier than the comparison point. The formula outputs a Boolean value: if the result is true (i.e., all three time points are earlier than the switching time point), the condition is met; otherwise, it is not met. The judgment operation runs periodically in the microcontroller, with the calculation interval set to 10% of the monitoring time window to ensure that no additional delay is introduced. The output is directly fed into the status flag logic.

[0062] If the judgment condition is met, the system marks it as a synchronization enhancement state. A synchronization enhancement state is a binary state identifier indicating that the pulse group is time-synchronized with the system operation and possesses stable enhancement characteristics. The marking process includes creating a state record: the record format includes a status code (set to an integer value of 1 to indicate synchronization), a time series index, and a calculated timestamp. The system maintains a state table, the table structure of which uses a hash mapping storage method, and the index is generated using the channel number and pulse group number. The state records are updated in real time to non-volatile storage units, supporting fast querying. After marking, the system associates the channel number with the status information: the association mechanism uses the channel number as the key and the status information as the value, forming a key-value pair dataset. The association process is executed through the data binding module, which loads the channel identifier list and merges it with the status code to generate an association array. The array is transmitted to the optimization control unit via a communication protocol.

[0063] The optimized acquisition strategy dynamically adjusts the signal acquisition based on associated state information. The operations include: the system retrieving channel and pulse group details for the synchronous enhancement state; and modifying the sampling protocol based on the state-based parameter adjustment logic. Specific modifications include shortening the sampling interval to accommodate the rapid dynamics of the enhanced pulses, with the sampling interval value reduced by 20% from the base value inherited from the dynamic configuration instruction. Simultaneously, the detection sensitivity is fine-tuned: the sensitivity gain is increased by 15% to capture high-enhancement signals. The optimized parameter configuration is configured in the pulse acquisition hardware registers, and real-time updates are achieved through write operations. The signal acquisition process initiates a new data stream acquisition sequence: the acquisition unit reinitializes using the adjusted parameters and captures the real-time data stream of the pulse queue region. The captured data is preprocessed before output, including waveform reconstruction and noise suppression steps.

[0064] Example 4: See Figure 5 The optimized data acquisition process includes invoking the marked period of the dynamic configuration command. The marked period is a time period identifier assigned by the system to a specific monitoring task, embedded by the command generation module when creating the dynamic configuration command. The system parses this identifier from the command data packet; it is a 32-bit unsigned integer, with the first 16 bits representing the monitoring task batch number and the last 16 bits representing the period sequence number. The parsed marked period value is used to retrieve the corresponding monitoring data records, which are stored in a circular buffer and indexed chronologically. The retrieval process employs a binary search algorithm to quickly locate the target dataset.

[0065] Filtering the trailing trajectory sequence of the tail pulse signal is a crucial step in the implementation. The tail pulse signal trailing segment is defined as the time period from when the pulse amplitude reaches 90% of its peak value until it falls back to 10% of the baseline. The trailing trajectory sequence contains continuous sampling points of the pulse amplitude within this time period and their corresponding timestamps. The system extracts this data from the raw signal data stream, with sampling point intervals of 100 nanoseconds to ensure the complete capture of the pulse decay process. The extracted data is arranged in chronological order to form a trajectory sequence array. The array structure contains two fields: the amplitude field stores the normalized voltage value, ranging from 0 to 1; the timestamp field records the offset relative to the pulse trigger moment, in microseconds.

[0066] When comparing the pulse amplitude change trend with the tail segment duration, the system executes a trend analysis algorithm. This algorithm first smooths the amplitude sequence, using a moving average filter to eliminate high-frequency noise, with a window width of 10 sampling points. The smoothed sequence is then input into the change detection module, which calculates the amplitude difference between adjacent sampling points, forming a rate of change sequence. A positive rate of change indicates an increase in amplitude, while a negative value indicates a decrease. The system statistically analyzes the sign distribution of the rate of change throughout the entire tail segment duration. If the proportion of positive rate of change exceeds 70%, it is determined that the amplitude has maintained an upward trend and has not yet reached a stable state. The tail segment duration is directly obtained from the difference between the first and last timestamps of the trajectory sequence, and the value is stored in microseconds.

[0067] When the amplitude continues to rise without reaching a stable state, the system calculates the time required to reach the predetermined release point. The predetermined release point is the critical position where the amplitude reaches a stable state, defined as the position where the absolute value of the rate of change is less than 0.5% for five consecutive sampling points. The calculation process scans backward from the end of the trajectory sequence to identify the time point when this condition is first met. The required time is the time difference from the current moment to that time point. The system dynamically updates the tail control cycle based on the calculation results. The original control cycle is the preset signal acquisition duration, and the updated cycle is extended by the calculated supplementary time value. The extension operation is implemented by modifying the comparison register of the hardware timer to ensure precise control.

[0068] The generation of the tail-end signal continuous modulation result involves multiple data dimensions. The remaining release path duration records the remaining time from the current moment to the predetermined release point; this value is updated in real time until the pulse reaches a stable state. The tail pulse release completion prediction period is the release time range estimated based on the current rate of change, calculated as a confidence interval of ±20% of the point estimate. The recommended signal holding time window is the minimum time length recommended by the system for continued acquisition, determined by the upper limit of the prediction period plus a safety margin, which is fixed at 50 microseconds. These parameters are encapsulated into a structure and transmitted to the display unit and log system via the data bus. Table 1 shows a typical tail-end signal continuous modulation result record.

[0069] Table 1: A typical record of the results of continuous modulation of the tail segment signal.

[0070]

[0071] The system detected a discharge pulse at the B-phase busbar connection of a switchgear. The dynamic configuration command period was marked as 0x00030015, corresponding to the 21st monitoring period of the third batch of tasks. The trajectory sequence of the tail pulse signal contained 248 sampling points, with a total duration of 24.8 microseconds. Amplitude change analysis showed that the positive change rate accounted for 82% within the first 18 microseconds, indicating a continuous upward trend. The predetermined release point was identified at the 215th sampling point in the sequence, and the remaining release time was calculated to be 9.3 microseconds. The original control period of 50 microseconds was updated to 59.3 microseconds, and the hardware timer responded and adjusted immediately. The control results were displayed in real time on the operation and maintenance interface and recorded in the database.

[0072] The system continuously monitors the update process until the pulse reaches a stable state. Each time new sampled data arrives, the above parameters are recalculated. When the rate of change is detected to meet the stability condition, the system automatically terminates the extended control cycle and restores the default acquisition settings. The entire implementation process is fully automated, requiring no manual intervention. All operation records, including raw trajectory data, analysis results, and control command changes, are archived chronologically, supporting post-event auditing and fault analysis. Data storage uses a compressed format, generating a data file every hour, with the file name including the device number and date / time information.

[0073] The implementation process paid special attention to real-time requirements. The calculation of key timing parameters was completed in a dedicated digital signal processor, ensuring results were output within 10 microseconds. Control command updates employed a hardware interrupt mechanism, with latency controlled to within 2 microseconds. Data logging operations ran in low-priority threads to avoid impacting real-time processing performance. The system resource allocation scheme ensured that each channel received equal processing time during multi-channel parallel monitoring. Through this refined implementation method, precise control and dynamic adjustment of the tail pulse signal were achieved, adapting to various complex discharge conditions.

[0074] Example 5: The acquisition of the tail-end signal continuous modulation result begins with the monitoring period number marked in the dynamic configuration instruction. This number is stored in hexadecimal encoding format at the header of the instruction data packet, with a fixed length of 4 bytes. The system extracts this number through the instruction parser, and the index range is limited to the preset monitoring task library. After the number is parsed, the data positioning module is activated. The module establishes an index mapping relationship in the time series database, associating the number with the corresponding physical storage sector address. The storage sectors adopt a cyclic overlay mechanism, with each sector containing monitoring data records for 24 consecutive hours, arranged in millisecond-level timestamp order. After the address pointer locks the target sector, the system loads the complete waveform buffer data of the tail pulse signal in that sector.

[0075] The filtering operation targets the trajectory data of the tail pulse in the latter part of the signal under a specified monitoring period. The trajectory data includes two core pieces of information: an amplitude sequence and a timestamp sequence. The amplitude sequence records the normalized instantaneous voltage value of the pulse signal, with a quantization precision of 16 bits and a value range of 0 to 4095 corresponding to a 0-5V voltage range. The timestamp sequence stores the absolute time of each sampling point, with the time referenced to the system's master clock, and the clock source using a GPS-synchronized crystal oscillator circuit. The filtering process performs data window truncation: the window start point is the beginning of the tail segment, which is identified by the edge detection algorithm as the point where the pulse amplitude first exceeds the steady-state threshold of 85%; the window end point is the pulse release end line, defined as the beginning of a stable state where the amplitude fluctuation range enters ±1% for more than 50 microseconds. The truncation operation generates a continuous subset of amplitude sequences and a subset of timestamp sequences, with dynamically variable sequence lengths and a minimum recording unit of one sampling point. The two sequences maintain a strict synchronization index relationship and are merged to form the tail segment pass-through trajectory sequence. This sequence is stored in a high-speed buffer, which is allocated an independent memory channel to avoid access conflicts.

[0076] Based on the amplitude sequence of the tail segment trajectory sequence, the amplitude variation trend of the tail pulse within the tail segment signal duration is analyzed. The analysis process employs a triple processing flow: first, a five-point cubic smoothing algorithm is applied to eliminate random noise interference; second, the amplitude change rate of adjacent sampling points is calculated using a differential operator; finally, the least squares method is used to fit and output the trend baseline. The amplitude increase value specifically refers to the characteristic quantity of a specific interval at the tail end of the sequence: the system selects the last 20% of the data window and calculates the arithmetic mean of the change rates of all adjacent sampling points within this window. The duration value corresponding to the tail segment is directly derived from the difference between the first and last elements of the timestamp sequence, in microseconds. Stability judgment is performed using binary detection: the slope value of the trend baseline is processed by a sign function, with a positive sign indicating an upward trend; simultaneously, the standard deviation of the entire sequence amplitude is calculated, and if the standard deviation is greater than a preset floating threshold, it is determined that the signal has not entered the stable interval. The judgment logic runs on a coprocessor, completing one detection operation every microsecond.

[0077] When the condition that the amplitude continues to rise and has not entered the stable range is met, the required supplementary duration for the pulse to reach the predetermined release point is calculated. The predetermined release point is the target amplitude value pre-configured in the system parameter table, which is set to 98% of the steady-state fluctuation center line by default. The calculation requires three inputs: the current amplitude position (taking the last sample value of the sequence), the target amplitude position (preset value), and the current trend slope (fitted output value). The supplementary duration is obtained through a linear extrapolation formula: (target amplitude position - current amplitude position) / current trend slope. The calculation result is rounded to an integer multiple of the nearest sampling interval, which is fixed at 0.1 microseconds. The supplementary duration is stored as a 16-bit integer and encapsulated in a tail-end remaining release time structure. The structure also contains the timestamp of the calculation time and the channel status word.

[0078] The system retrieves the supplementary time value from the remaining release time interval of the tail segment to update the real-time tail segment signal control cycle. The update process involves deep interaction with the hardware timer: the system reads the end time register value of the current control cycle, adds the supplementary time value to this value, and writes it to the new end time register. Time value conversion uses hardware clock units, and the conversion factor is stored in read-only memory. The correction operation takes effect immediately, triggering the hardware comparator to update the comparison threshold. The control interval of the channel tail segment is reset: the start point of the new interval remains unchanged from the original start time, and the end point is updated to the corrected time point. The control interval parameters are written to the channel control logic unit via a dedicated bus. The unit responds to the configuration to generate a time slot allocation scheme, which includes a time slot start flag, a duration counter, and interrupt trigger conditions. The updated control parameter set constitutes the continuous control result of the tail segment signal. The result data packet format includes four fields: the original control interval start time, the new end time, the supplementary time value, and the update operation status code.

[0079] The control results are fed back to the signal conditioning unit in real time. This unit dynamically adjusts the gain and filtering parameters of the acquisition circuit according to the newly set control interval: the gain value increases proportionally with the remaining release time, with an increase of 0.5% per microsecond; the high-pass filter cutoff frequency is adjusted downwards based on the time margin, with an adjustment step of 10kHz every 50 microseconds. Simultaneously, a tail pulse monitoring thread is started: the thread priority is set to the second highest level, and the scan cycle is synchronized with the sampling rate. During each sampling, the deviation between the actual amplitude and the predicted value is monitored; if the deviation exceeds 5%, a recalculation process is triggered. The predicted tail pulse release period is generated based on continuous monitoring results: the lower limit of the period is 90% of the current remaining time, and the upper limit is 110%, with the output being a time interval descriptor. The recommended signal hold time window is set to 120% of the supplementary time value as the minimum, with the buffer capacity as the upper limit. All output parameters are written to the system status register group and transmitted to the central monitoring platform via the backplane bus.

[0080] Data integrity safeguards include: freezing the hardware timer state before the time management module performs a write operation; triggering verification and calculation after the update is completed; storing key parameters in triple backups on different storage devices; and generating an operation log for each configuration change. The log records include the original control values, calculated parameters, new set values, and hardware response status, with a fixed record length of 64 bytes. The system is designed with a fault-tolerant mechanism: when the compensation time exceeds the safety threshold (default 200 microseconds), it automatically switches to conservative mode and issues an abnormal signal. In a dual-powered switchgear monitoring example, the system successfully corrected the tail pulse truncation problem at the C-phase bushing connection, fully recording the release trajectory from 92% amplitude to the steady state. All processing response delays are strictly controlled within 3 microseconds, achieving accurate capture of transient discharge behavior. Maintenance tools can read the historical records of the tail signal's continuous control results through the diagnostic interface, supporting waveform playback and analysis functions.

[0081] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0082] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for real-time monitoring of partial discharge in substation switchgear, characterized in that, Includes the following operations: A set of raw signal data generated by partial discharge inside the switch cabinet is captured by a distributed sensor array. The set of raw signal data includes a timestamp sequence and a spectral distribution. The original signal data set is processed to separate the discharge pulses from the background noise components, and the rise time, energy integral value and frequency bandwidth of each discharge pulse are calculated to form a set of discharge characteristic parameters. Based on the set of discharge characteristic parameters, potential discharge risk modes are identified, and dynamic configuration instructions, including monitoring priority ranking and parameter adaptation requirements, are generated. The dynamic configuration command is invoked to modify the sampling interval and detection sensitivity of the monitoring unit in real time, and to execute the optimized signal acquisition process. The system integrates and optimizes the real-time data stream during the acquisition process, updates the set of discharge characteristic parameters, and iteratively generates new dynamic configuration instructions. The acquisition of the original signal data set is specifically as follows: the partial discharge signal inside the switch cabinet is collected by a distributed sensor array, the change in electric field intensity and the time delay sequence before and after each discharge event are extracted, and the electric field offset distance and time delay period are calculated based on the electric field intensity difference and time sequence length for each event, thereby generating an offset and delay feature set. Based on the electric field offset distance and time delay period of the offset and delay feature set, combined with the amplitude change information of the discharge event at consecutive moments before and after the occurrence, the number of pulse attenuation times of each discharge event in the offset and delay stages is counted, the correlation distribution pattern between the number of attenuation times and the delay period is identified, and the attenuation frequency characteristics of the discharge event are generated. The discharge event attenuation frequency characteristics are invoked to determine discharge events whose attenuation frequency exceeds a preset critical threshold for discharge behavior. The corresponding event type is then associated with the location of the switchgear area to generate an original signal data set.

2. The method for real-time monitoring of partial discharge in substation switchgear according to claim 1, characterized in that, The original signal data set includes pulse waveform index number, interference type classification code, and sensor spatial mapping position; the discharge characteristic parameter set includes pulse energy accumulation, frequency offset characteristic value, and discharge event time distribution density; the dynamic configuration instructions include monitoring channel sorting weight, sampling parameter adjustment range, and control effective time window; the optimized acquisition process includes signal reconstruction trigger state, noise suppression start time, and pulse integrity identification state.

3. The method for real-time monitoring of partial discharge in substation switchgear according to claim 1, characterized in that, The specific operation for generating the discharge characteristic parameter set is as follows: call the switch cabinet area location identified in the original signal data set, obtain the three time data of the corresponding location within the monitoring period: signal start time, first pulse trigger time, and tail pulse end time, calculate the first pulse response interval and tail pulse passage time difference respectively, and generate the passage key time interval. Based on the difference between the first pulse response interval and the last pulse passage time in the critical passage time interval, the total number of pulses and pulse energy distribution within the corresponding position monitoring period are called to identify the total length and density level of the pulse queue, and obtain pulse distribution density information. According to the pulse distribution density information, combined with the time interval distribution of the front, middle and rear positions of the queue within the critical passage time interval, the pulse density offset feature value is calculated, the position window of the density anomaly segment in the monitoring period is identified, and a pulse density offset time period is generated. The corresponding position point in the pulse density offset time period is called to evaluate the mapping relationship between the time window and the pulse release behavior segment within the monitoring period, and continuous segments with pulse release behavior offset are screened and marked as pulse behavior delay occurrence areas to generate a set of discharge characteristic parameters.

4. The method for real-time monitoring of partial discharge in substation switchgear according to claim 3, characterized in that, The specific operation for generating the dynamic configuration instruction is as follows: Based on the monitoring channel number indicated in the discharge characteristic parameter set, extract the duration of the continuous stagnation of the tail pulse at the end of the signal segment and the release completion time point of the channel within two consecutive monitoring cycles. Combine the time difference between the release completion time point and the signal end time point of the current cycle to obtain the channel tail segment release lag information. Based on the channel tail segment release lag information, determine whether the release lag value exceeds the channel release completion benchmark threshold, filter the channel numbers that have not completed release, and extract the tail pulse stagnation time in the corresponding channel. Sort the channels according to the stagnation duration of the incomplete channels to generate a channel release stagnation priority sequence. Call the sorting information in the channel release stagnation priority sequence to configure additional monitoring time periods for the channels in sequence. Adjust the channel monitoring time window length within the total monitoring time range, record the channel number and the corresponding adjusted monitoring duration, and generate the dynamic configuration instruction.

5. The method for real-time monitoring of partial discharge in substation switchgear according to claim 4, characterized in that, The specific execution operation of the optimized acquisition process is as follows: calling the channel start configuration time value recorded in the dynamic configuration instruction, detecting the longitudinal distance distribution of the electric field in the pulse queue area of ​​the corresponding channel within the real-time monitoring period, identifying the continuous enhancement start time point of three adjacent pulses in the queue segment, and generating the queue enhancement start time sequence; Based on the time points of the three adjacent pulse enhancement actions in the queue enhancement start time sequence, it is determined whether the enhancement time is earlier than the signal preparation switching time point within the channel monitoring time window. If the judgment condition is met, it is marked as synchronous enhancement state, and the channel number and status information are associated to execute the optimized acquisition process.

6. The method for real-time monitoring of partial discharge in substation switchgear according to claim 5, characterized in that, The optimized acquisition process also includes additional operations: calling the marking period of the dynamic configuration instruction, filtering the trajectory sequence of the latter part of the tail pulse signal, comparing the pulse amplitude change trend with the tail duration, and if the amplitude continues to rise without reaching a stable state, calculating the time required to reach the predetermined release point and updating the tail control period to obtain the tail signal continuous regulation result; the tail signal continuous regulation result includes the remaining release path duration, the predicted tail pulse release completion period, and the recommended signal holding time window.

7. The method for real-time monitoring of partial discharge in substation switchgear according to claim 6, characterized in that, The specific operation for obtaining the continuous control result of the tail segment signal is as follows: The monitoring cycle number marked in the dynamic configuration instruction is called, the trajectory data of the tail pulse in the later part of the signal under the corresponding cycle is filtered, and the continuous amplitude sequence and timestamp data from the start time of the tail segment to the end line of the pulse release are extracted to generate the tail segment passage trajectory sequence; Based on the amplitude sequence in the tail segment passage trajectory sequence, the amplitude change trend of the tail pulse within the tail segment signal duration is analyzed, the amplitude increase value of the tail segment is extracted, and compared with the corresponding duration of the tail segment. If the amplitude continues to rise and has not entered the stable range, the supplementary time required for the pulse to reach the predetermined release point is calculated, and the remaining release time of the tail segment is generated; The supplementary time value required for the channel in the remaining release time of the tail segment is called, the real-time tail segment signal control cycle is updated, the originally configured tail segment signal end time point is corrected, the control range of the channel tail segment is reset, and the continuous control result of the tail segment signal is obtained.

8. A real-time monitoring system for partial discharge in substation switchgear, characterized in that, The system is used to implement the real-time monitoring method for partial discharge in substation switchgear as described in any one of claims 1-7. The system includes: a signal acquisition module that detects partial discharge behavior inside the switchgear, acquires the electric field offset position and time delay sequence of pulses, calculates the attenuation frequency of pulses, and generates an original signal data set; a discharge behavior analysis module that, based on the original signal data set, identifies key pulse actions that cause signal interruption, binds the key pulse actions to location points, and generates a set of discharge characteristic parameters; a signal delay calculation module that calls the set of discharge characteristic parameters to obtain the signal start time, first pulse trigger time, and tail pulse end time within the location point, calculates the time interval difference, and generates a dynamic configuration instruction; a monitoring time dynamic configuration module that, based on the dynamic configuration instruction, analyzes the completion status of channel release, calculates the stagnation time of incomplete channels, reallocates the monitoring time window as needed, and executes an optimized acquisition process; and a signal execution module that calls the optimized acquisition process to detect and evaluate the longitudinal distance of the electric field and continuous enhancement points of the queued pulses, adjusts the signal execution time according to the pulse amplitude change trend and tail duration, and obtains the tail signal continuous control result.

9. The real-time partial discharge monitoring system for substation switchgear according to claim 8, characterized in that, The system also includes a data interaction module: the output of the signal acquisition module is connected to the input of the discharge behavior analysis module, transmitting the original signal data set; the output of the discharge behavior analysis module is connected to the input of the signal delay calculation module, transmitting the discharge characteristic parameter set; the output of the signal delay calculation module is connected to the input of the monitoring time dynamic configuration module, transmitting dynamic configuration instructions; the output of the monitoring time dynamic configuration module is connected to the input of the signal execution module, transmitting parameters for optimizing the acquisition process; the output of the signal execution module is fed back to the data interaction module, updating the original signal dataset and merging and restarting the signal acquisition module.

Images