A high-voltage switch cabinet working state intelligent prediction method and system

Abstract

The present application relates to the technical field of high-voltage switch cabinet online monitoring, and particularly relates to a high-voltage switch cabinet working state intelligent prediction method and system. The method comprises: taking each phase circuit path in the same physical compartment as an independent monitoring unit, respectively collecting data to form an action sequence, a temperature rise sequence and a partial discharge sequence; generating an action compartment candidate pair, a temperature rise compartment candidate pair and a partial discharge compartment candidate pair from the three sequences; pairing the action temperature rise candidate pair and the temperature rise partial discharge candidate pair; calculating a source exclusion quantity to determine attribution credibility; screening candidate links to determine link closure; comparing the link closure with a cloud baseline to select the baseline with the smallest deviation as a prediction result. The present application can improve the accuracy and reliability of high-voltage switch cabinet working state prediction.

Description

Technical Field

[0001] This invention relates to the field of online monitoring technology for high-voltage switchgear, specifically to an intelligent prediction method and system for the operating status of high-voltage switchgear. Background Technology

[0002] High-voltage switchgear typically consists of compartments such as circuit breaker compartment, busbar compartment, cable compartment, contact box, and instrument compartment. These compartments are connected by circuit breaker contacts, busbars, bushings, cable terminals, and insulating partitions, forming a primary circuit path. After a single closing operation, the opening distance, overtravel, closing bounce, opening and closing speed, and three-phase synchronicity affect the contact pressure and contact area. During subsequent load operation, if the contact pressure is insufficient or the contact area is too small, the contacts, busbar connection points, or cable terminals on the same phase path are prone to concentrated temperature rise, slow temperature drop, or interphase overheating. Sustained thermal anomalies may also alter the local insulation environment near the contact box, cable terminals, bushings, or insulating partitions, enhancing the partial discharge response.

[0003] Currently, the main technical solutions for predicting the operating status of high-voltage switchgear include: online monitoring schemes based on partial discharge sensors, status assessment schemes based on multi-source data fusion, and assessment schemes based on cloud-edge collaborative deep learning. While existing technologies can achieve a certain level of monitoring and assessment, the following problems still exist: First, solutions based on partial discharge sensors rely solely on the partial discharge signal itself for judgment. Since there are many metal partitions inside the switch cabinet, the partial discharge signal is easily reflected, attenuated, or crosstalked, making it difficult to determine whether the partial discharge enhancement comes from the current phase compartment, which can easily lead to misjudgment of the source.

[0004] Second, although the multi-source data fusion scheme integrates multi-source information such as action data, temperature data and partial discharge data, it usually treats the multi-source data as parallel inputs and does not determine whether the action response, temperature rise response and partial discharge response form a succession relationship along the primary loop path of the same phase. This can easily lead to the splicing of anomalies from different phases and different compartments into the same risk event, resulting in false alarms.

[0005] Third, in cloud-edge collaborative solutions, the cloud model relies on sample distribution and training results for identification. If the edge side does not form a judgment on the ownership of the phase compartment and the closure of the state chain, the cloud identification results are difficult to explain whether the partial discharge enhancement is supported by the deviation of the previous opening and closing and the subsequent current-carrying temperature rise retention. The prediction results lack interpretable physical basis. Summary of the Invention

[0006] This invention provides an intelligent prediction method and system for the operating status of high-voltage switchgear to solve existing problems.

[0007] The intelligent prediction method for the operating status of high-voltage switchgear of the present invention adopts the following technical solution: One embodiment of the present invention provides an intelligent prediction method for the operating status of high-voltage switchgear, the method comprising the following steps: Each phase circuit path in the same physical compartment is treated as an independent monitoring unit, and its action response, temperature rise response and partial discharge response data are collected separately. The action sequence, temperature rise sequence and partial discharge sequence are formed by aligning them according to the time window. Based on the association between each response and the candidate compartments and phase paths, candidate pairs of action compartments, candidate pairs of temperature rise compartments, and candidate pairs of partial discharge compartments are generated from the three sequences, respectively. From the candidate pairs, the action compartment candidate pairs that are in the same phase and adjacent in time are paired with the temperature rise compartment candidate pairs to form action temperature rise candidate succession pairs, and the temperature rise compartment candidate pairs are paired with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs. For each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. The source rejection quantity is calculated comprehensively, and the source rejection quantities of each candidate compartment under the same response are ranked. The attribution credibility is determined based on the smallest and second smallest difference. Candidate pairs that meet the requirements of attribution credibility are selected to form candidate links. The action temperature rise bearing fracture amount corresponding to the action temperature rise candidate receiving pair and the temperature rise partial discharge bearing fracture amount corresponding to the temperature rise partial discharge candidate receiving pair are determined. The link closure is determined based on the attribution credibility of the initial action and the two bearing fracture amounts. By comparing the link closure with the cloud baseline, removing links with insufficient closure, calculating the deviation from various baselines, and selecting the baseline with the smallest deviation as the prediction result.

[0008] Furthermore, the loop paths of each phase within the same physical compartment are treated as independent monitoring units, and their action response, temperature rise response, and partial discharge response data are collected separately. These data are then aligned according to time windows to form action sequences, temperature rise sequences, and partial discharge sequences, specifically including: Treat the circuit paths of each phase within the same physical compartment as independent monitoring units; For each independent monitoring unit, the opening and closing action data, mechanism stroke curve, opening distance, overtravel, closing bounce, opening and closing speed and three-phase synchronicity data of each phase are collected to form an action sequence. The action sequence is used to record the position deviation of each phase contact and mechanism during one opening and closing process. Collect load current, contact temperature, busbar connection point temperature and cable terminal temperature data for each phase to form a temperature rise sequence. The temperature rise sequence is used to record the temperature rise retention of contacts, busbar connection points or cable terminals during the load stage after closing. Data on UHF partial discharge, TEV partial discharge, ultrasonic partial discharge, and pulsed current partial discharge of each phase are collected to form a partial discharge sequence. The partial discharge sequence is used to record the response intensity, arrival order, and response duration of each partial discharge sensor within the same evaluation period. Align the action sequence, temperature rise sequence, and partial discharge sequence according to the switch cabinet number, phase number, compartment number, sensor number, and time window.

[0009] Furthermore, based on the association between each response and the candidate compartments and phase paths, candidate pairs of action compartments, candidate pairs of temperature rise compartments, and candidate pairs of partial discharge compartments are generated from the three sequences, specifically including: Based on the location of the acquisition sensors, the structural relationship of the compartments, and the loop connection relationship corresponding to each response, determine the phase to which each response belongs and the candidate compartment; From the action sequence, each action response is associated with its corresponding phase and candidate compartment to generate action compartment candidate pairs, where the candidate compartments of the action response include the circuit breaker compartment and the contact box; From the temperature rise sequence, each temperature rise response is associated with its corresponding phase and candidate compartment to generate candidate temperature rise compartment pairs. The candidate compartments for the temperature rise response include contact boxes, busbar connection points, and cable termination areas. From the partial discharge sequence, each partial discharge response is associated with its corresponding phase and candidate compartment to generate partial discharge compartment candidate pairs. The candidate compartments of the partial discharge response include cable compartments, bus bushings, contact boxes, adjacent insulation areas, and insulation partition areas.

[0010] Furthermore, from the candidate pairs, the action compartment candidate pairs that are of the same phase and adjacent in time are paired with the temperature rise compartment candidate pairs to form action-temperature rise candidate succession pairs, and the temperature rise compartment candidate pairs are paired with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs, specifically including: From the candidate pairs of action compartments and the candidate pairs of temperature rise compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate action temperature rise candidate succession pairs. Among them, the action temperature rise candidate succession pairs are used to characterize whether there is a succession relationship between the deviation of opening and closing and the subsequent current-carrying temperature rise retention. From the candidate pairs of temperature rise compartments and the candidate pairs of partial discharge compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate candidate pairs of temperature rise partial discharge. The candidate pairs of temperature rise partial discharge are used to characterize whether there is a succession relationship between the current-carrying temperature rise retention and the subsequent partial discharge enhancement.

[0011] Furthermore, for each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. Their source repulsion values ​​are calculated comprehensively, and the source repulsion values ​​of each candidate compartment under the same response are ranked. The attribution reliability is determined based on the smallest and second smallest difference, specifically including: For each candidate response compartment, candidate difference items are extracted relative to the candidate compartment. These candidate difference items include the degree of loop path disconnection, the degree of response timing anomaly, the adjacent deviation of temperature rise points, the deviation of partial discharge propagation attenuation, and the deviation of sensor installation direction. The sum of the candidate differences for the same candidate compartment is divided by the number of candidate differences to obtain the source rejection of the response relative to the candidate compartment. The source rejection factors of each candidate compartment under the same response are sorted from smallest to largest. The candidate compartment corresponding to the smallest source rejection factor is taken as the predicted source of the response, and the candidate compartment corresponding to the second smallest source rejection factor is taken as the competing source of the response. The attribution confidence of the compartment response is obtained by subtracting the source repulsion of the predicted source from the source repulsion of the competing source and then dividing by the source repulsion of the competing source.

[0012] Furthermore, the degree of loop path disconnection is determined based on the length of the connecting path between the candidate compartment and the response source in the primary loop topology table. The value is 0 when there is a direct connection, and 1 when there is no connecting path. When there are intermediate nodes, the degree of loop path disconnection is determined based on the number of intermediate nodes, and the degree of loop path disconnection is positively correlated with the number of intermediate nodes. The degree of response timing anomaly is determined based on the deviation between the response timestamp and the expected time window; The adjacent deviation of the temperature rise point is determined based on the length of the heat conduction path between the temperature sensor and the candidate compartment. The deviation of partial discharge propagation attenuation is determined by the difference between the theoretical amplitude calculated by the signal propagation attenuation model and the actual amplitude; The sensor installation orientation deviation is determined based on the angle between the sensor orientation and the candidate compartment direction.

[0013] Furthermore, candidate pairs that meet the attribution reliability requirements are selected to form candidate links, and the fracture amount of the action temperature rise candidate receiving pair and the fracture amount of the temperature rise partial discharge candidate receiving pair are determined, specifically including: From the candidate pairs whose attribution credibility meets the preset requirements, select the action compartment candidate pairs, temperature rise compartment candidate pairs and partial discharge compartment candidate pairs on the connection path of the same phase or adjacent compartments, and form candidate links along the primary loop path in the order of the action, temperature rise and partial discharge stages. For the candidate temperature rise receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate temperature rise receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of the temperature rise receiving. For the candidate partial discharge receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate partial discharge receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of partial discharge receiving. Among them, the actual degree of fracture of each item in the candidate temperature rise of the action includes the degree of discontinuity of the phase path, the degree of non-stagnation of temperature rise after load correction, and the degree of non-acceptance of the time window. The actual degree of fracture in the partial discharge candidate pair includes the degree of insufficient insulation proximity, the degree of abnormal response timing of partial discharge occurring before temperature rise, the degree of path misalignment of the partial discharge path in other phase compartments, and the degree of interference window not being eliminated.

[0014] Furthermore, the link closure is determined based on the reliability of the attribution of the initial action and the amount of breakage in the two segments, specifically including: The reliability of the initial action is used as the link support quantity, and the sum of the reliability of the initial action, the temperature rise failure rate, and the temperature rise partial discharge failure rate is used as the total link deviation quantity. The link closure is obtained by dividing the link support quantity by the total link deviation quantity.

[0015] Furthermore, the link closure is compared with the cloud baseline. After eliminating links with insufficient closure, the deviation from various baselines is calculated, and the baseline with the smallest deviation is selected as the prediction result. Specifically, this includes: Compare the link closure with a preset closure threshold and eliminate candidate links whose link closure is lower than the preset threshold; The remaining candidate links are compared with the cloud-based normal heating baseline, contact heating baseline, and insulation discharge baseline, respectively. The deviation between each candidate link and each type of baseline is calculated. The deviation includes the degree of deviation of the phase path, the degree of deviation of the stage sequence, and the degree of deviation of the response intensity evolution. The baseline type with the smallest deviation is selected as the prediction result for the candidate link; If the baseline with the smallest deviation is the contact heating baseline, then output "Contact heating attention"; if it is the insulation discharge baseline and the link closure meets the requirements, then output "Insulation degradation warning"; if no candidate link passes the closure threshold screening, then output "Source mismatch verification".

[0016] This invention proposes an intelligent prediction system for the operating status of high-voltage switchgear, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the intelligent prediction method for the operating status of high-voltage switchgear as described above.

[0017] The beneficial effects of the technical solution of the present invention are: In this embodiment of the invention, by first constructing candidate pairs of response compartments and calculating the source rejection quantity and attribution reliability, the action, temperature rise, and partial discharge response are locked to specific phase compartment sources, effectively reducing the risk of source mismatch caused by signal crosstalk under the metal enclosure structure. On this basis, by pairing to generate candidate acceptance pairs of action temperature rise and candidate acceptance pairs of temperature rise partial discharge, and calculating the two-segment breakage quantity and link closure, the physical link judgment is realized as to whether the deviation of opening and closing is accepted by the subsequent current-carrying temperature rise and further transmitted as the risk of insulation partial discharge, so that the state prediction has an interpretable physical basis. Finally, the deviation comparison between the link closure result on the edge side and the baseline of the same type of link in the cloud is performed. This not only makes up for the lack of single-cabinet samples by using historical samples from multiple cabinets, but also retains the real-time physical judgment capability of the edge side, avoiding the misjudgment of directly outputting insulation fault prediction based solely on a single partial discharge enhancement, and significantly improving the accuracy and reliability of high-voltage switchgear operating state prediction. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0019] Figure 1 A flowchart illustrating an intelligent prediction method for the operating status of a high-voltage switchgear according to an embodiment of the present invention; Figure 2 This is a structural diagram of an intelligent prediction system for the working status of a high-voltage switchgear, provided in one embodiment of the present invention. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the intelligent prediction method for the operating status of a high-voltage switchgear proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0022] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent prediction method for the working status of high-voltage switchgear provided by the present invention.

[0023] This invention provides an intelligent prediction method and system for the operating status of high-voltage switchgear. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of an intelligent prediction method for the operating status of a high-voltage switchgear according to an embodiment of the present invention. The method includes the following steps: S101. Treat the circuit paths of each phase in the same physical compartment as independent monitoring units, collect their action response, temperature rise response and partial discharge response data respectively, and align them according to the time window to form action sequence, temperature rise sequence and partial discharge sequence.

[0024] In this embodiment, the loop paths of each phase within the same physical compartment are treated as independent monitoring units, and their action response, temperature rise response, and partial discharge response data are collected separately. These data are then aligned according to time windows to form action sequences, temperature rise sequences, and partial discharge sequences, specifically including: Treat the circuit paths of each phase within the same physical compartment as independent monitoring units; For each independent monitoring unit, the opening and closing action data, mechanism stroke curve, opening distance, overtravel, closing bounce, opening and closing speed and three-phase synchronicity data of each phase are collected to form an action sequence. The action sequence is used to record the position deviation of each phase contact and mechanism during one opening and closing process. Collect load current, contact temperature, busbar connection point temperature and cable terminal temperature data for each phase to form a temperature rise sequence. The temperature rise sequence is used to record the temperature rise retention of contacts, busbar connection points or cable terminals during the load stage after closing. Data on UHF partial discharge, TEV partial discharge, ultrasonic partial discharge, and pulsed current partial discharge of each phase are collected to form a partial discharge sequence. The partial discharge sequence is used to record the response intensity, arrival order, and response duration of each partial discharge sensor within the same evaluation period. Align the action sequence, temperature rise sequence, and partial discharge sequence according to the switch cabinet number, phase number, compartment number, sensor number, and time window.

[0025] In one specific embodiment, the first In a single operational evaluation cycle, each phase compartment of the high-voltage switchgear is continuously monitored. Here, "each phase compartment" refers to treating the primary circuit paths of phases A, B, and C as independent monitoring and tracking units within the same physical compartment (such as a circuit breaker compartment or cable compartment). Although the contacts or busbars of phases A, B, and C may be located in the same physical space, they are forcibly distinguished as three independent tracking objects during the data acquisition phase. Responses from different phases within the same compartment are not mixed into the same monitoring unit. This provides a physical coordinate basis for subsequent judgments regarding whether actions, temperature rises, and partial discharges are transmitted along the same phase path.

[0026] For each individual monitoring unit, the following data were collected: Circuit breaker opening and closing action data, including mechanism stroke curve, opening distance, overtravel, closing bounce, opening and closing speed, and three-phase synchronicity data; load current data; temperature data, including contact temperature, busbar connection point temperature, and cable terminal temperature; partial discharge data, including UHF partial discharge, TEV partial discharge, ultrasonic partial discharge, and pulse current partial discharge data; installation and configuration data, including sensor installation location, compartment structure relationship, phase number, and primary circuit connection relationship.

[0027] The aforementioned data is collected by various sensors deployed in the switchgear compartments, including travel sensors, current transformers, thermocouples or infrared temperature sensors, ultra-high frequency partial discharge sensors, transient ground voltage partial discharge sensors, ultrasonic sensors, and pulse current sensors. Each sensor continuously collects data at a preset sampling frequency, and the collected data is timestamped.

[0028] After data collection, alignment is performed according to five dimensions: switch cabinet number, phase number, compartment number, sensor number, and time window. Specifically, the raw data collected by each sensor are grouped into the same time window according to their respective switch cabinet, phase, compartment, and sensor identification, so that the action data, temperature data, and partial discharge data within the same time window have comparable time and physical location references.

[0029] After alignment, the following data sequence is formed: Action window sequence (i.e., action sequence): records the position deviation of each phase contact and mechanism during a single opening and closing operation. For example, in a closing operation, if the overtravel of a phase contact is lower than the standard value and the closing bounce time exceeds the threshold, the closing event is recorded as the action deviation event of that phase within that time window.

[0030] Current-carrying temperature rise window sequence (i.e., temperature rise sequence): records the temperature rise retention behavior of contacts, busbar connection points, or cable terminals during the load-bearing stage after closing. For example, within a preset time window after the load current stabilizes or the load recedes, if the temperature of a certain phase contact box remains higher than that of other phases and decreases slowly, this temperature response is recorded as an abnormal temperature rise event of that phase within that time window.

[0031] Partial discharge response window sequence (i.e., partial discharge sequence): records the response intensity, arrival order, and response duration of each partial discharge sensor within the same evaluation period. For example, if a sensor detects a partial discharge signal amplitude exceeding the background noise threshold within a certain time window, the partial discharge event is recorded as the partial discharge response event at the sensor's location within that time window.

[0032] Simultaneously, a table of adjacent compartment connections and a table of phase conductor paths are generated. The table of adjacent compartment connections describes the connectivity between physical compartments, such as the connection between the circuit breaker compartment and the cable compartment via contact boxes and busbars. The table of phase conductor paths describes the complete paths of each of the three phases A, B, and C in the primary circuit, starting from the circuit breaker contacts, passing through the busbar connection points to the cable terminals, and indicating the phase to which each conductor segment belongs and the compartments it passes through.

[0033] S102. Based on the association between each response and the candidate compartments and phase paths, generate action compartment candidate pairs, temperature rise compartment candidate pairs and partial discharge compartment candidate pairs from the three sequences respectively.

[0034] In this embodiment, based on the association between each response and the candidate compartments and phase paths, candidate pairs of action compartments, candidate pairs of temperature rise compartments, and candidate pairs of partial discharge compartments are generated from the three sequences, specifically including: Based on the location of the acquisition sensors, the structural relationship of the compartments, and the loop connection relationship corresponding to each response, determine the phase to which each response belongs and the candidate compartment; From the action sequence, each action response is associated with its corresponding phase and candidate compartment to generate action compartment candidate pairs, where the candidate compartments of the action response include the circuit breaker compartment and the contact box; From the temperature rise sequence, each temperature rise response is associated with its corresponding phase and candidate compartment to generate candidate temperature rise compartment pairs. The candidate compartments for the temperature rise response include contact boxes, busbar connection points, and cable termination areas. From the partial discharge sequence, each partial discharge response is associated with its corresponding phase and candidate compartment to generate partial discharge compartment candidate pairs. The candidate compartments of the partial discharge response include cable compartments, bus bushings, contact boxes, adjacent insulation areas, and insulation partition areas.

[0035] For example, based on the connection relationship between the acquisition sensor, candidate compartment, and primary loop for each response, three types of response compartment candidate pairs are generated (also referred to as response-compartment candidate pairs; the above-mentioned action compartment candidate pairs, temperature rise compartment candidate pairs, and partial discharge compartment candidate pairs are collectively referred to as response-compartment candidate pairs (or simply candidate pairs)): Action compartment candidate pairs (also known as action response-compartment candidate pairs): Each action response is associated with a candidate compartment from which it may originate. Candidate compartments for action responses include the circuit breaker compartment and the contact box. This is because opening and closing actions occur near the circuit breaker mechanism and contacts, and the direct source of the action positioning deviation can only be the operating mechanism within the circuit breaker compartment or the contact contact state within the contact box.

[0036] Temperature rise compartment candidate pairs (also known as temperature rise response-compartment candidate pairs): Each temperature rise response is associated with a candidate compartment from which it may originate. Candidate compartments for temperature rise responses include contact boxes, busbar compartment connection points, and cable termination areas. This is because current-carrying temperature rises can occur at any conductor connection point along the primary circuit path, including contact points within contact boxes, busbar lap surfaces within busbar compartments, and cable termination connections.

[0037] Partial discharge (PD) compartment candidate pairs (also known as PD response-compartment candidate pairs): Each PD response is associated with a candidate compartment from which it may originate. Candidate compartments for PD responses include cable compartments, bus bushings, adjacent insulation areas of contact boxes, and insulation partition areas. This is because partial discharge can occur in any weakly insulated area along the primary circuit path, including cable termination insulation, bus bushing surfaces, contact box inner walls, and insulation partition surfaces.

[0038] In this way, each response is bound to one or more candidate compartments (specifically, based on a pre-established table of adjacent compartment connections and a table of phase conductor paths, the compartment where the sensor is located and all its directly connected adjacent compartments are considered candidate compartments; if the location of the sensor cannot uniquely determine the compartment's affiliation, all possible compartments within the sensor's signal coverage area are included in the candidate compartment set). Each candidate pair is labeled with a phase number, compartment number, sensor number, time window, and physical path relationship. The candidate pairs clearly define the computational objects for subsequent analysis, avoiding the direct use of scattered, unbound raw data fields in later steps.

[0039] S103. From the candidate pairs, pair the action compartment candidate pairs that are of the same phase and adjacent in time with the temperature rise compartment candidate pairs to form action temperature rise candidate succession pairs, and pair the temperature rise compartment candidate pairs with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs.

[0040] In this embodiment, from the candidate pairs, the action compartment candidate pairs that are in the same phase and adjacent in time are paired with the temperature rise compartment candidate pairs to form action-temperature rise candidate succession pairs, and the temperature rise compartment candidate pairs are paired with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs, specifically including: From the candidate pairs of action compartments and the candidate pairs of temperature rise compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate action temperature rise candidate succession pairs. Among them, the action temperature rise candidate succession pairs are used to characterize whether there is a succession relationship between the deviation of opening and closing and the subsequent current-carrying temperature rise retention. From the candidate pairs of temperature rise compartments and the candidate pairs of partial discharge compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate candidate pairs of temperature rise partial discharge. The candidate pairs of temperature rise partial discharge are used to characterize whether there is a succession relationship between the current-carrying temperature rise retention and the subsequent partial discharge enhancement.

[0041] For example, after forming candidate pairs, adjacent action windows and temperature rise windows on the same phase path are organized into action-temperature rise candidate acceptance pairs (also called action-temperature rise candidate acceptance pairs). Specifically, for the same phase, if an action deviation event is recorded in a certain action window, and an abnormal temperature rise event is recorded at the contact or busbar connection point of that phase in the subsequent adjacent temperature rise window (i.e., the time window numbers are consecutive, or the time window interval does not exceed one window cycle, such as the interval between adjacent windows not exceeding 20 minutes when the window length is 10 minutes)), then these two windows are paired into an action-temperature rise candidate acceptance pair. This acceptance pair is used to subsequently determine whether the opening and closing position deviation is accepted by the subsequent current-carrying temperature rise.

[0042] Similarly, adjacent temperature rise windows and partial discharge windows on the same phase path are grouped into temperature rise-partial discharge candidate acceptance pairs (also known as temperature rise-partial discharge candidate acceptance pairs). Specifically, for the same phase, if an abnormal temperature rise event is recorded in a certain temperature rise window, and a partial discharge enhancement event is recorded in the insulation-adjacent region of that phase in the subsequent adjacent partial discharge window, then these two windows are paired into a temperature rise-partial discharge candidate acceptance pair. This acceptance pair is used to subsequently determine whether the current-carrying temperature rise will further transmit to the insulation-adjacent region as a partial discharge risk.

[0043] By using the above data organization method, subsequent analysis uses candidate pairs and successor pairs with phase, compartment, sensor, time window, and physical path relationships, instead of hashed action, temperature, and partial discharge fields. In this way, each response has a clear candidate physical source and path attribution, providing a unified data object for subsequent determination of whether the same risk source gradually emerges along the action, thermal response, and discharge response.

[0044] S104. For each candidate pair, extract candidate difference terms to characterize the degree of deviation between the response source and the candidate compartment, calculate the source rejection quantity comprehensively, and sort the source rejection quantities of each candidate compartment under the same response. Determine the attribution credibility based on the smallest and second smallest differences.

[0045] In this embodiment, for each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. The source repulsion amount is calculated comprehensively, and the source repulsion amounts of each candidate compartment under the same response are ranked. The attribution reliability is determined based on the smallest and second smallest difference, specifically including: For each candidate response compartment, candidate difference items are extracted relative to the candidate compartment. These candidate difference items include the degree of loop path disconnection, the degree of response timing anomaly, the adjacent deviation of temperature rise points, the deviation of partial discharge propagation attenuation, and the deviation of sensor installation direction. The sum of the candidate differences for the same candidate compartment is divided by the number of candidate differences to obtain the source rejection of the response relative to the candidate compartment. The source rejection factors of each candidate compartment under the same response are sorted from smallest to largest. The candidate compartment corresponding to the smallest source rejection factor is taken as the predicted source of the response, and the candidate compartment corresponding to the second smallest source rejection factor is taken as the competing source of the response. The attribution confidence of the compartment response is obtained by subtracting the source repulsion of the predicted source from the source repulsion of the competing source and then dividing by the source repulsion of the competing source.

[0046] The degree of loop path disconnection is determined based on the length of the connecting path between the candidate compartment and the response source in the primary loop topology table. The value is 0 when there is a direct connection and 1 when there is no connecting path. When there are intermediate nodes, the degree of loop path disconnection is determined based on the number of intermediate nodes, and the degree of loop path disconnection is positively correlated with the number of intermediate nodes. The degree of response timing anomaly is determined based on the deviation between the response timestamp and the expected time window; The adjacent deviation of the temperature rise point is determined based on the length of the heat conduction path between the temperature sensor and the candidate compartment. The deviation of partial discharge propagation attenuation is determined by the difference between the theoretical amplitude calculated by the signal propagation attenuation model and the actual amplitude; The sensor installation orientation deviation is determined based on the angle between the sensor orientation and the candidate compartment direction.

[0047] For example, after the data organization and candidate pair generation are completed, each candidate pair needs to be evaluated to determine whether the response can be reliably assigned to a candidate compartment.

[0048] A partial discharge signal received by a sensor could originate from the A-phase contact box, the B-phase cable compartment, or even be transmitted from a neighboring cabinet. If the source is unclear, all subsequent judgments will be incorrect. Therefore, this step involves identifying the most probable compartment from multiple possible candidates and determining the degree of its "uniqueness."

[0049] For each response-compartment candidate pair, a corresponding candidate difference term is extracted based on the response type. The candidate difference term characterizes the degree of deviation between the response source and the candidate compartment. A larger deviation indicates that the candidate compartment is less likely to explain the source of the response.

[0050] In layman's terms, the candidate difference terms are asking: if this signal really did originate from the A-phase contact box, then it should be "reasonable" in terms of path, time, location, attenuation, direction, and so on. If any aspect is found to be unreasonable, it means that "this signal may not have come from here."

[0051] Specifically, the focus of the action response is on the opening and closing mechanisms and contact areas. The system extracts the corresponding differences between the action phase, mechanism position, candidate contact compartments, and opening / closing position deviations. For example, if a closing action shows that phase B overtravel is too small, the system will associate this action response with multiple candidate compartments, such as the phase B circuit breaker compartment, phase B contact box, and phase A contact box. If the candidate compartment is the phase B contact box, the action phase and mechanism position are consistent with the candidate compartment, with a small difference; if the candidate compartment is the phase A contact box, the phases are inconsistent, with a larger difference.

[0052] The temperature rise response focuses on conductor connections and heat conduction paths under load. It extracts differences in conductor connections and adjacent heat conduction between temperature points and candidate contacts, busbar connections, or cable terminals. For example, if the temperature of the B-phase contact box is significantly higher, the system will associate this temperature rise response with candidate compartments such as the B-phase contact box, the B-phase busbar connection, and the A-phase contact box. If the candidate compartment is the B-phase contact box itself, the temperature point directly corresponds to the candidate location with a small difference; however, if the candidate compartment is the A-phase contact box, heat must be conducted across phases through the conductor to reach that location, which is physically illogical and results in a larger difference.

[0053] The focus of partial discharge (PD) response analysis is on the propagation paths within insulated areas and metal compartments. Differences in propagation paths, signal arrival order deviations, and amplitude attenuation deviations are extracted between the PD sensor and candidate insulated areas. For example, if a UHF sensor on the cabinet top receives a PD signal, the system will associate this PD response with multiple candidate compartments, such as the insulated area adjacent to the B-phase contact box, the A-phase busbar bushing area, and the intermediate joint in the cable compartment. If a candidate compartment is located near the sensor and the propagation path is unobstructed, the difference is small; if a candidate compartment requires multiple reflections through metal partitions before being received by the sensor, the difference is large.

[0054] Action signals, temperature rise, and partial discharge signals within the switchgear exhibit different physical propagation characteristics. Action signals are transmitted along the operating mechanism, temperature rises are conducted along conductors, and partial discharge signals propagate and reflect within the metal compartment. Therefore, candidate differences need to simultaneously reflect the path, timing, adjacency, and installation orientation, avoiding the reliance on a single amplitude value to determine the source.

[0055] The specific methods for determining candidate differences and their values ​​include: The degree of circuit path disconnection is determined based on the length of the connecting path between the candidate compartment and the response source in the primary circuit topology table. If the candidate compartment is directly connected to the phase path containing the response, the value is 0 (indicating complete connectivity with no disconnection issues); if there are no connecting paths, the value is 1 (indicating complete disconnection); if there are intermediate nodes connecting, the value is determined by the number of intermediate nodes, with more intermediate nodes resulting in a larger value, meaning the degree of circuit path disconnection is positively correlated with the number of intermediate nodes. Furthermore, the switchgear has a clearly defined primary circuit topology diagram at the factory, which records the connection relationships between compartments and between phase conductors. In practical implementation, those skilled in the art can determine this value using a monotonically increasing function based on the number of intermediate nodes between two nodes in the topology table. The mapping function can be chosen independently, which is a conventional implementation method.

[0056] Response timing anomaly degree: For action-temperature rise succession, the expected timing relationship is action first, followed by temperature rise. If the action response occurs within time window T, and the temperature rise response occurs within the adjacent time window T+1, the timing is normal; if the temperature rise response occurs before the action response, or the two time windows are too far apart, the timing is abnormal. For temperature rise-partial discharge succession, the expected timing relationship is temperature rise first, followed by partial discharge. If the enhanced partial discharge response occurs within the adjacent time window after the temperature rise retention window, the timing is normal; if the partial discharge occurs before the temperature rise, or the two time windows are too far apart, the timing is abnormal. If the actual time window of the response conforms to the above expected order, the response timing anomaly degree is low; if the timing is reversed or the deviation exceeds the allowable range, the response timing anomaly degree is high, and the greater the deviation, the greater the anomaly degree. Those skilled in the art can map the time deviation to the range of 0 to 1 using a monotonically increasing function based on the actual time difference between the action response and the temperature rise response, or the actual time difference between the temperature rise response and the partial discharge response.

[0057] Temperature rise point adjacency deviation: The shorter the heat conduction path between the temperature sensor and the candidate compartment (e.g., the temperature sensor is directly installed on the surface of the candidate contact box), the smaller the deviation value; the longer the heat conduction path (e.g., the temperature sensor is installed at the cable terminal, and the candidate compartment is the connection point of the busbar compartment, requiring heat to travel a long conductor path for conduction), the larger the deviation value; if there is no direct heat conduction path between the temperature sensor and the candidate compartment (e.g., cross-phase conduction is required), the deviation value is the maximum. Furthermore, the installation positions of the temperature sensor and the candidate compartments in the switchgear are known quantities, and the length of the heat conduction path between them can be directly measured from the switchgear structural drawings or obtained through finite element thermal simulation. Those skilled in the art can map this path length to the range of 0 to 1 using a monotonically increasing function to determine the deviation value.

[0058] Partial Discharge Propagation Attenuation Deviation: A signal propagation attenuation model between each compartment of the switchgear is pre-established (obtainable through simulation or actual measurement). This model adopts a path loss model, where the signal amplitude attenuates exponentially with propagation distance. The attenuation exponent is calibrated according to the propagation medium (metal partition, air gap, insulating material, etc.) between different compartments. Through electromagnetic simulation using the finite-difference time-domain method or actual field testing, the propagation path loss coefficient from each candidate compartment to each sensor is obtained, forming a propagation loss matrix. This model can calculate the theoretical amplitude of the partial discharge signal propagating from the candidate compartment to the sensor based on the propagation path between the candidate compartment and the sensor. After actually receiving the partial discharge signal, the actual amplitude of the signal is obtained, and the difference between the theoretical amplitude and the actual amplitude corresponding to the candidate compartment is calculated. The larger the difference, the larger the deviation, indicating that the candidate compartment is less able to explain the amplitude characteristics of the current partial discharge signal. The partial discharge signal propagation attenuation model between each compartment of the switchgear can be pre-established through electromagnetic simulation (such as the finite-difference time-domain method) or actual field testing. This model can calculate the theoretical amplitude of the discharge signal emitted from any candidate compartment propagating to any sensor. After obtaining the amplitude through actual measurement, the absolute or relative difference between the theoretical amplitude and the actual amplitude is calculated, and the difference is mapped to the range of 0 to 1 through a monotonically increasing function.

[0059] Sensor installation orientation deviation: During sensor installation, its spatial coordinates and orientation (i.e., the sensor's normal direction or optimal receiving direction) are recorded. The spatial azimuth angle of the candidate compartment relative to the sensor (i.e., the direction vector from the sensor's position to the candidate compartment) is calculated. This azimuth angle is compared to the sensor's optimal receiving direction; the larger the angle between the two, the greater the deviation. In simpler terms, if the sensor is directly facing the candidate compartment, the deviation is small; if the candidate compartment is to the side or even behind the sensor, the deviation is large. The sensor's spatial coordinates and orientation (normal direction or optimal receiving direction) are recorded during installation, and the spatial azimuth angle of the candidate compartment relative to the sensor can be calculated from the spatial coordinates of the two points. Technicians calculate the angle between this azimuth angle and the sensor's optimal receiving direction; the deviation value is determined by the proportion of this angle to the maximum angle.

[0060] Regarding the first The first response and the first For each candidate compartment, the five candidate differences are organized into a candidate difference sequence, denoted as [list of candidate compartments]. to ,in, Indicates the first The number of candidate difference items corresponding to each response (usually 5 items, but can be increased or decreased depending on the actual sensor configuration).

[0061] The formula for calculating the source rejection amount is:

[0062] in, For the first The response relative to the first The source rejection of each candidate compartment; For the first One candidate difference item; For the first The number of candidate differences corresponding to each response.

[0063] In other words, the "rejection score" of the candidate compartment is obtained by averaging the five candidate difference items (path disconnection degree, response timing anomaly degree, temperature rise adjacent deviation, partial discharge attenuation deviation, and installation direction deviation). The higher the score, the less likely the candidate compartment is to actually be the signal source.

[0064] The larger the value, the more difficult it is to interpret the current response of the candidate compartment in terms of the corresponding path, timing, adjacency, attenuation, or installation direction. Therefore, The larger the value, the more difficult it is for the candidate compartment to explain the source of the response; conversely, the smaller the value, the more reasonable it is to explain the source of the response.

[0065] By sorting the source repulsion of each candidate compartment under the same response from smallest to largest, we obtain... , to ,in: The candidate compartment with the minimum source exclusion is taken as the most likely source compartment (predicted source) for the current response. The candidate compartment corresponding to the second smallest source rejection factor is considered as the closest competing source compartment (competing source) for that response.

[0066] Because simply looking at the minimum rejection factor is insufficient. If the minimum and second-minimum scores are very close, it indicates that the signal could be interpreted in both compartments simultaneously. This is a manifestation of "source aliasing" or "signal crosstalk," and the attribution is unreliable. Therefore, a separate "attribution reliability" calculation is needed to assess the "degree of uniqueness of attribution."

[0067] The formula for calculating the attribution reliability of compartment responses is as follows:

[0068] in, For the first The reliability of attribution of compartment responses; The minimum source rejection for the same response; The second smallest source rejection amount for the same response; To prevent extremely small positive numbers with a denominator of zero.

[0069] The numerator represents the "gap" between the first and second candidates. The larger the gap, the further ahead the first candidate is from the second, and the more closely they are matched. The denominator normalizes the result to between 0 and 1, making it easier to compare.

[0070] The larger the value, the more significant the exclusion gap between the most likely source and the adjacent competing source. The higher the credibility of the response being uniquely assigned to the most likely source compartment, the more suitable the response is to enter the subsequent same-source link to participate in the link closure judgment.

[0071] The smaller the value, the less significant the difference between the most likely source and the competing source. This response may indicate signal crosstalk or source aliasing between adjacent compartments. This response should be screened out and not included in subsequent link closure judgment.

[0072] S105. Select candidate pairs that meet the requirements of attribution credibility to form candidate links, determine the action temperature rise bearing failure amount corresponding to the action temperature rise candidate receiving pair, and the temperature rise partial discharge bearing failure amount corresponding to the temperature rise partial discharge candidate receiving pair, and determine the link closure based on the attribution credibility of the initial action and the bearing failure amounts of the two segments.

[0073] In this embodiment, candidate pairs that meet the attribution reliability requirements are selected to form candidate links, and the temperature rise failure amount corresponding to the action temperature rise candidate receiving pair and the temperature rise partial discharge failure amount corresponding to the temperature rise partial discharge candidate receiving pair are determined, specifically including: From the candidate pairs whose attribution credibility meets the preset requirements, select the action compartment candidate pairs, temperature rise compartment candidate pairs and partial discharge compartment candidate pairs on the connection path of the same phase or adjacent compartments, and form candidate links along the primary loop path in the order of the action, temperature rise and partial discharge stages. For the candidate temperature rise receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate temperature rise receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of the temperature rise receiving. For the candidate partial discharge receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate partial discharge receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of partial discharge receiving. Among them, the actual degree of fracture of each item in the candidate temperature rise of the action includes the degree of discontinuity of the phase path, the degree of non-stagnation of temperature rise after load correction, and the degree of non-acceptance of the time window. The actual degree of fracture in the partial discharge candidate pair includes the degree of insufficient insulation proximity, the degree of abnormal response timing of partial discharge occurring before temperature rise, the degree of path misalignment of the partial discharge path in other phase compartments, and the degree of interference window not being eliminated.

[0074] The link closure is determined based on the reliability of the initial action and the amount of breakage in the two segments, specifically including: The reliability of the initial action is used as the link support quantity, and the sum of the reliability of the initial action, the temperature rise failure rate, and the temperature rise partial discharge failure rate is used as the total link deviation quantity. The link closure is obtained by dividing the link support quantity by the total link deviation quantity.

[0075] For example, compartment response attribution reliability is used to filter actions, temperature rise, and partial discharge responses that can participate in link closure determination. The system pre-sets an attribution reliability threshold, and only responses that meet this threshold are considered valid. The response is only allowed to enter the link closure judgment process if it exceeds this threshold; if If the response is less than or equal to this threshold, it is considered to have an unreliable source, is marked as "source aliasing" or "signal crosstalk," and is discarded, not participating in subsequent link analysis. The purpose of this screening step is clear: it is better to miss a signal with an ambiguous source than to introduce an incorrect signal into the subsequent link, causing a false alarm.

[0076] It is worth noting that the attribution reliability threshold can be set according to the statistical distribution of historical operating data of the same type of switchgear. For example, it can be 0.6 times the historical attribution reliability average. The specific value can be adjusted according to different switchgear types and operating conditions.

[0077] After completing the attribution reliability calculation and screening, a series of response-compartment candidate pairs were obtained and confirmed as having reliable sources. These candidate pairs recorded the operation deviation event (a phase fails to close), the abnormal temperature rise event (a phase contact box overheats), and the partial discharge enhancement event (discharge in a certain insulation area), respectively. In this step, it is necessary to determine whether these three events—operation deviation, abnormal temperature rise, and enhanced partial discharge—are the same fault evolution process occurring sequentially on the same phase path, or whether they are three unrelated independent events accidentally combined together.

[0078] If a normal temperature rise caused by a load fluctuation and a partial discharge signal from an adjacent phase are captured simultaneously, the system will forcibly associate them as a "fault," resulting in a false alarm. The purpose of this step is to determine whether the action-temperature rise and temperature rise-partial discharge meet the three conditions of "same phase, same sequence, same path" to determine whether the three constitute a complete causal chain.

[0079] From candidate pairs whose compartment response attribution credibility meets the attribution credibility threshold, filter action response-compartment candidate pairs, temperature rise response-compartment candidate pairs, and partial discharge response-compartment candidate pairs on the connection path of the same phase, same compartment, or adjacent compartment.

[0080] In other words, this step is the screening condition: if phase A has an operational deviation, phase B has an abnormal temperature rise, and phase C has enhanced partial discharge, then the system will only pair the phase A operation with the phase A temperature rise (same phase), and will not pair the phase A operation with the phase B temperature rise. If the operation is in phase A, the temperature rise is in phase B, and the partial discharge is in phase C in a candidate link, then these three candidate pairs will not enter the same candidate link because the "same phase path" condition is not met.

[0081] The selected candidate pairs also need to meet the stage connection relationship in the primary circuit of the switchgear: Action-Temperature Rise Connection: When the deviation in the opening and closing of the circuit breaker manifests as a subsequent temperature rise retention at the contacts of the same phase, busbar connection points, or cable terminals, it is more likely due to insufficient contact pressure or a small contact area causing the current-carrying response. In other words, if the circuit breaker fails to close properly (action deviation), and the same phase does indeed subsequently heat up (temperature rise retention), then the causal chain of "action deviation - heat generation" may be valid.

[0082] Temperature rise-partial discharge continuation: When the temperature rise persists and corresponds to enhanced partial discharge in adjacent insulation areas, it is more likely that the thermal environment has affected the local insulation condition. In other words, if the same phase continues to heat up and enhanced partial discharge does indeed occur in the adjacent insulation areas of that phase, it indicates that the causal chain of "heating-insulation degradation-discharge" may be valid.

[0083] If three responses occur simultaneously within the same time period, and their paths, time sequences, or adjacency relationships cannot be closed, they will not be treated as a common-origin state chain.

[0084] Based on the action-temperature rise candidate response, determine whether the deviation in the opening and closing position is absorbed by the thermal response of the same phase current during the subsequent load-bearing stage: If the operational deviation is concentrated in a certain phase contact or circuit breaker compartment, and after the subsequent load stabilizes or falls back, the phase contact, busbar connection point or adjacent cable terminal still remains at a high temperature, falls back slowly or is hot between phases, then the possibility that the operational contact deviation will be further amplified by the current-carrying contact process increases.

[0085] In other words, if the overtravel of phase B of a switchgear is too small when it is closed (operational deviation), and the temperature of the contact box of phase B is significantly higher than that of phases A and C, and the temperature of phase B remains high even after the load is reduced, then it means that the chain of "closed poorly - high contact resistance - severe overheating" is valid.

[0086] Conversely, if the temperature rise only rises and falls synchronously with the load (high load, high temperature; low load, low temperature), or if the temperature rise point is inconsistent with the previous action (action in phase B, heat generation in phase A), then the action deviation and the temperature rise response do not form an effective connection.

[0087] Based on the temperature rise-partial discharge candidate pair, determine whether the current-carrying thermal anomaly is further transmitted to enhance partial discharge in the insulation adjacent region: If the partial discharge enhancement lags behind the temperature rise retention window and occurs in the vicinity of the same phase insulation, the vicinity of the contact box, the cable terminal, or the bus bushing, the possibility of the thermal anomaly further disturbing the local insulation environment increases.

[0088] In other words, if after a period of time (lag) the UHF sensor near the B-phase contact box detects a partial discharge signal after the B-phase contact box has been at a high temperature, it means that the chain of "high temperature - insulation aging - partial discharge" is valid.

[0089] Conversely, if partial discharge occurs before temperature rise (timing reversed), the partial discharge response path is located in another phase compartment (path misalignment), or the partial discharge only occurs briefly within a significant interference window (such as lightning interference), then the temperature rise response and the partial discharge response do not form an effective connection.

[0090] The closure of the action-thermal-discharge link is based on the opening and closing action response as the starting point of the link. In this embodiment, the reliability of the action, temperature rise, and partial discharge is not directly multiplied together. Instead, the reliability of the action response is used as the initial support, and the parts between the action and temperature rise, and between the temperature rise and partial discharge that do not conform to the phase path, time window, load correction, and insulation proximity relationship are organized into the continuity of the breakage quantity.

[0091] In other words, the approach to calculating link closure is as follows: first, examine how tightly the action response source is locked (attribution reliability), then subtract the amount of breakage between the action and temperature rise, and between temperature rise and partial discharge. The greater the initial support and the fewer the breakages, the higher the link closure.

[0092] For adjacent stages and The resulting fault pairs (i.e., action-temperature rise candidate fault pairs or temperature rise-partial discharge candidate fault pairs) are organized into fault fracture terms based on the following characteristics. : The action-temperature rise candidate pair (i.e., the action-temperature rise candidate pair) is checked for three items: discontinuous phase path (the action and temperature rise are not in the same phase), non-connection of time window (the temperature rise does not appear within a reasonable time window after the action), and failure to maintain response after load correction (the temperature rise remains after the load drops, instead of decreasing normally with the load).

[0093] Specifically, phase path discontinuity: This determines whether the action response and temperature rise response belong to the same phase. If the action response is recorded in phase B, while the temperature rise response is recorded in phase A, then the phase path is discontinuous, and the degree of discontinuity is high; if they belong to the same phase, the degree of discontinuity is low. In a specific embodiment, the degree of phase path discontinuity is a binary quantity, with a value of 0 when the action response and temperature rise response belong to the same phase and a value of 1 when they belong to different phases.

[0094] Time window discontinuity: This determines whether the temperature rise response occurs within a reasonable time window after the action response. If the temperature rise response occurs within an adjacent evaluation window after the action deviation, then the time window is considered contiguous; if the time windows of the temperature rise response and the action response are too far apart, or if the temperature rise response occurs before the action response, then the time window is discontinuous, and the degree of discontinuity is set to a higher value. In a specific embodiment, the degree of time window discontinuity is determined based on the time interval between the action window and the temperature rise window. If the temperature rise window and the action window are adjacent or the time interval is within a preset continuation time window (such as within the same evaluation cycle), the value is 0; if the time interval exceeds the continuation window but is still within the same operating cycle, the value increases linearly between 0 and 1 as the interval increases; if the temperature rise window occurs before the action window (time order reversed), the value is 1.

[0095] Load-corrected response not held: This determines whether the temperature rise response is a case of "failed to decrease when it should," rather than normal fluctuations with the load. First, the temperature rise response is corrected based on load current data (specifically, a linear correction method is used, performing a univariate linear regression with load current as the independent variable and temperature rise as the dependent variable to obtain the regression coefficient of temperature rise with load change under normal operating conditions. The actual temperature rise is then subtracted from the product of the regression coefficient and the current load current to obtain the corrected temperature rise value after deducting the load effect; if there is a time delay between load change and temperature rise change, the average load current within the sliding time window (e.g., the first 15 minutes) is used as the correction basis), deducting normal temperature fluctuations caused by load changes. If the temperature remains high or the rate of temperature decrease is significantly slower than normal after the load decreases, the load-corrected response is not held, and the degree of failure is high; if the temperature rise rises and falls synchronously with the load, the degree of failure is low. In a specific embodiment, the degree of non-stagnation of temperature rise after load correction is determined based on the deviation between the rate of temperature decrease after load decrease and the normal rate of temperature decrease. First, the temperature rise response is corrected based on the load current change rate. Then, the difference between the actual temperature rise decrease rate after the load drops and the normal temperature rise decrease rate of the same cabinet type is calculated. The larger the difference, the larger the value of this measure, with a maximum value of 1.

[0096] Four items were checked for the temperature rise-partial discharge candidate pair (i.e., temperature rise-partial discharge candidate pair): insufficient insulation proximity (no insulation proximity between the temperature rise area and the partial discharge area), abnormal timing of partial discharge appearing before temperature rise, misalignment of the partial discharge path in other phase compartments, and interference window not eliminated (such as external factors such as lightning interference not being eliminated).

[0097] Specifically, insufficient insulation proximity is determined by assessing whether an insulation proximity exists between the temperature rise retention area and the partial discharge enhancement area. This means determining whether the temperature rise area can transfer heat to the partial discharge area through insulating materials (contact box walls, bushing surfaces, insulating partitions, etc.), thus affecting the insulation state of that area. If the temperature rise area and the partial discharge area are adjacent and have insulating contact, the insulation proximity is established, and the fracture degree is set to a lower value. If the two areas are far apart or have no insulating contact, the insulation proximity is insufficient, and the fracture degree is set to a higher value. In a specific embodiment, the degree of insufficient insulation proximity is a binary quantity: 0 when there is insulating material contact (such as contact box walls, bushing surfaces, insulating partitions, etc.) between the temperature rise retention area and the partial discharge enhancement area, and 1 when there is no insulating contact.

[0098] Timing anomaly of partial discharge preceding temperature rise: Determine whether partial discharge enhancement occurs after temperature rise retention. If the time window for partial discharge enhancement is later than the temperature rise retention window, the timing is normal; if partial discharge occurs before temperature rise, the timing is abnormal, and the severity of this anomaly is higher. In a specific embodiment, the degree of timing anomaly of partial discharge preceding temperature rise is determined based on the time order of the partial discharge response and the temperature rise response. If partial discharge enhancement occurs after the temperature rise retention window, the value is 0; if partial discharge occurs before temperature rise, the value is 1.

[0099] Partial discharge path misalignment in other phase compartments: This determines whether the propagation path of the partial discharge response points to the phase where the temperature rise area is located. If the partial discharge response, after being located, points to the insulation adjacent area of ​​the phase where the temperature rise is retained, then the path is consistent; if the partial discharge path points to other phase compartments, then the path is misaligned, and the degree of dislocation is higher. In a specific embodiment, the degree of path misalignment of the partial discharge path in other phase compartments is a binary quantity, with a value of 0 when the propagation path of the partial discharge response points to the insulation adjacent area of ​​the phase where the temperature rise is retained, and a value of 1 when it points to other phase compartments.

[0100] Interference Window Not Eliminated: This determines whether the partial discharge response appears within the interference window and has not been eliminated. The interference window refers to the time period during which a known external interference source (such as lightning, corona discharge from other equipment, etc.) may generate a partial discharge signal. If the partial discharge response only appears within the interference window and the window has not been eliminated, the breakage level is higher; if the partial discharge response persists outside the interference window, or if the interference window has been eliminated, the breakage level is lower. In a specific embodiment, the degree of interference window not eliminated is determined based on whether the partial discharge response appears within the interference window. If the partial discharge response appears outside the known interference window (such as lightning, corona discharge from other equipment, etc.), the value is 0; if it only appears within the interference window and has not been eliminated, the value is 1.

[0101] The above-mentioned methods of value determination are all illustrative examples. Those skilled in the art can adopt other equivalent quantification methods based on the actual sensor configuration and switch cabinet structure, as long as the judgment rule of "the greater the degree, the more severe the fracture" is met.

[0102] At the same time, the allowable deviations of the candidate acceptance under the current cabinet type and sensor arrangement are read. In other words, not all deviations are considered "breakage"; the system allows for a certain reasonable margin of error. Measurement errors of the sensor itself, normal load fluctuations, and normal thermal conduction delays are all permissible deviations and are not included in the breakdown.

[0103] The formula for calculating the amount of breakage is:

[0104] in, Indicates the first Adjacent stages in candidate links and The amount of breakage that can be sustained; Indicates the first One fracture item; This indicates the allowable deviation of the corresponding breakage item under the current cabinet type, load and sensor arrangement. Its value is determined in advance through the calibration test of the same type of switch cabinet under normal working conditions, or preset according to the sensor accuracy and load fluctuation range during system initialization. This indicates the number of broken items in the receiving pair.

[0105] It should be noted that the value of the allowable deviation term is determined in advance by the following method: For the same type of switchgear, calibration tests are conducted under normal operating conditions (no operational deviation, no abnormal temperature rise, no partial discharge enhancement). The normal time interval range between operation and temperature rise, the normal temperature rise variation range caused by load fluctuations, and the normal timing relationship between temperature rise and partial discharge are collected. The boundary value of the normal fluctuation range or the boundary value multiplied by a preset safety factor (such as 1.2 times) is used as the allowable deviation.

[0106] For deviations that cannot be directly obtained through calibration tests (such as deviations in sensor installation direction), allowable deviation values ​​can be preset based on the sensor's factory accuracy and installation tolerance range.

[0107] The specific values ​​of the allowable deviation for each item can be set according to different cabinet types, different sensor arrangements and different load levels. Different allowable deviation values ​​can be used for the same cabinet type under different working conditions.

[0108] For each actual fracture extent, subtract the allowable deviation for that individual item. If the result is positive (the actual fracture exceeds the allowable range), retain this "excess portion"; if the result is negative or zero (the actual fracture is within the allowable range), record it as 0 and do not include it in the fracture amount. Finally, sum all the excess portions and divide by the number of items to obtain the average fracture amount of that segment. This formula only retains the portion "exceeding the allowable range," which avoids mistaking ordinary load fluctuations, sensor installation errors, or normal thermal conduction delays for fractures in the same source link.

[0109] After the calculation of the fracture amount was completed, two values ​​were obtained for each of the two joint segments: : Action-temperature rise breakage (caused by discontinuity in the phase path between the action deviation from the contact or circuit breaker compartment and the subsequent temperature rise retention contact, busbar connection point or cable terminal, the absence of temperature rise retention after load correction or the absence of time window. The larger the value, the more severe the breakage in the action-temperature rise segment, that is, the action-temperature rise breakage).

[0110] Temperature rise-partial discharge breakage (caused by insufficient insulation proximity between the temperature rise retention area and the subsequent partial discharge enhancement area, partial discharge appearing before temperature rise, partial discharge path located in other phase compartments or interference window not eliminated; the larger the value, the more severe the breakage in the temperature rise-partial discharge section, i.e., temperature rise-partial discharge breakage).

[0111] At the same time, the credibility of the attribution of the initial action was obtained. (The reliability of locking the source of the initial action; the higher the value, the more uniquely the action response can be attributed to a specific compartment).

[0112] Now we need to combine these three values ​​into a unified metric: link closure. The design concept of link closure is to place the reliable support of the action source in the numerator (the greater the support, the more reliable the link), and also place the amount of breakage at the two connecting points in the denominator (the fewer the breakages, the more reliable the link), and comprehensively evaluate the reliability of the entire link.

[0113] For the The formula for calculating the closure of the action-thermal-discharge link for each candidate link is:

[0114] in, Indicates the first Action-thermal-discharge link closure of candidate links; This indicates the reliability of the compartment response attribution for the initial action response of this link; Indicates the amount of fracture due to temperature rise during the action; Indicates the temperature rise - partial discharge bearing fracture amount; This represents the smallest positive number that prevents the denominator from being zero.

[0115] In other words, in the expression for link closure, the reliability of action attribution appears in both the numerator and the denominator. Therefore... The larger the numerator, the larger the link support, and the larger the denominator, but the increase is relatively small, so the overall link closure increases; while the two-segment breakage only appears in the denominator, the larger their values, the larger the denominator, and the smaller the link closure.

[0116] The higher the value, the more likely the deviation in operation, current-carrying temperature rise, and enhanced partial discharge are to be a continuous state formed along the primary loop path of the same phase. The lower the value, the more likely the link is caused by common load fluctuations, sensor interference, or mismatch of sources from different compartments.

[0117] This embodiment assigns a "credibility score" to each candidate link. The score depends on three factors: how tightly the action response source is locked (…). The larger the better), the less breakage occurs between the action and the temperature rise. The smaller the better), the less fracture occurs between temperature rise and partial discharge. (The smaller the better). If all three aspects are met, the link closure is high, indicating that this link is likely a real "closing deviation-contact heating-insulation deterioration" fault chain; otherwise, it will be screened out and will not participate in subsequent steps.

[0118] S106. Compare the link closure with the cloud baseline, remove links with insufficient closure, calculate the deviation from various baselines, and select the baseline with the smallest deviation as the prediction result.

[0119] In this embodiment, the link closure is compared with the cloud baseline. After eliminating links with insufficient closure, the deviation from various baselines is calculated, and the baseline with the smallest deviation is selected as the prediction result. Specifically, this includes: Compare the link closure with a preset closure threshold and eliminate candidate links whose link closure is lower than the preset threshold; The remaining candidate links are compared with the cloud-based normal heating baseline, contact heating baseline, and insulation discharge baseline, respectively. The deviation between each candidate link and each type of baseline is calculated. The deviation includes the degree of deviation of the phase path, the degree of deviation of the stage sequence, and the degree of deviation of the response intensity evolution. The baseline type with the smallest deviation is selected as the prediction result for the candidate link; If the baseline with the smallest deviation is the contact heating baseline, then output "Contact heating attention"; if it is the insulation discharge baseline and the link closure meets the requirements, then output "Insulation degradation warning"; if no candidate link passes the closure threshold screening, then output "Source mismatch verification".

[0120] For example, after completing the link closure calculation, one or more candidate links and their corresponding link closures are obtained. The core of this embodiment is to determine which type of fault this link belongs to: is it normal load heating, continuous heating caused by poor contact, or has it developed into insulation deterioration and partial discharge?

[0121] The operational data of a single switchgear unit is limited, making it difficult to determine whether the current link status is abnormal based solely on its own data. For example, a contact box temperature of 62℃ for a certain phase may be high for this particular unit, but if other switchgear units of the same model and operating conditions generally reach around 60℃ under the same load, then this indicates normal current-carrying heat generation. Therefore, the role of cloud-based baselines is to provide a reference of "the normal performance range of similar switchgear units under various operating conditions," avoiding misjudgments caused by insufficient data from a single switchgear unit.

[0122] This means that the preset closure threshold can be set according to the statistical distribution of historical normal link closure data of the same type of switchgear. For example, it can be 0.5 times the average value of historical normal link closure. The specific value can be adjusted according to different switchgear types and operating conditions.

[0123] The cloud baseline is formed by link samples from multiple switchgear units of the same type under different sensor arrangements, load levels, and operating conditions, including the following three categories: Normal current-carrying heating link: A link sample in which the temperature rise caused by the load current changes synchronously with the load under normal operating conditions of the switchgear. The characteristics of this type of link are: the temperature rise rises and falls synchronously with the load, the temperature drops normally after the load falls back, there are no preceding events of operational deviation, and no partial discharge is present.

[0124] Contact heating link: This type of link is characterized by insufficient contact pressure or small contact area due to deviations in opening and closing actions (such as insufficient overtravel or excessive closing bounce), resulting in a continuous temperature rise after operation under load. The characteristics of this type of link are: there is an preceding event of the action deviation; the temperature rise occurs after the action; and the temperature remains high or decreases slowly after the load recedes, but partial discharge enhancement has not yet occurred.

[0125] Insulation discharge link: This type of link exhibits enhanced partial discharge due to continuous high temperature caused by contact heating, leading to insulation material degradation. Characteristics of this type of link include: the presence of an antecedent event of operational deviation, an intermediate event of temperature rise stagnation, and a delayed partial discharge enhancement following the temperature rise, with the partial discharge source located near the insulation in the temperature rise region.

[0126] The cloud-based baseline uses established links as comparison objects. For the same cabinet type, normal current-carrying heating under different load levels will form a relatively stable temperature rise and fall pattern. Contact heating links will exhibit in-phase thermal stagnation after operational deviations, while insulation discharge links will show enhanced partial discharge in adjacent insulation areas after thermal stagnation. Comparing the current link with these baselines can prevent the cloud model from ignoring the preceding actions and temperature rise continuity simply because the partial discharge amplitude is high.

[0127] In other words, the cloud baseline is like a "historical medical record," documenting the typical performance of the same type of switchgear under various operating conditions. After the edge side calculates the current link, it compares this link with the three typical cases in the "medical record" to see which type it most closely resembles, and then outputs the corresponding conclusion.

[0128] The cloud baseline does not directly provide the on-site status, but instead puts the closed links already formed on the edge side into the same cabinet-type baseline set for candidate comparison.

[0129] For the From the candidate links, links with insufficient closure are first filtered out based on a link closure threshold. Specifically, the link closure threshold is... With preset threshold Comparison: If This indicates insufficient closure of the link, meaning there are still many breaks between the three segments of action-temperature rise-partial discharge, which is insufficient to form a reliable homologous state chain. Therefore, this link is directly eliminated and will not participate in subsequent comparisons. This indicates that the link closure meets the requirements and allows it to proceed to the subsequent comparison process.

[0130] The remaining links that meet the closure requirements are compared with the baselines of the cloud-based normal current-carrying heating links, contact heating links, and insulation discharge links, respectively, to form the current link relative to the first... Deviation from the baseline of the class state The deviation includes three dimensions: Phase path deviation: The degree of difference between the phase path of the current link and the phase path of the baseline sample. For example, if all the contact heating links in the baseline sample follow the path of "Phase A operation - Phase A heating", and the current link is "Phase B operation - Phase B heating", then the phase path deviation is small (same type, just different phase); if the current link is "Phase A operation - Phase B heating", then the phase path deviation is large (cross-phase, which is physically impossible).

[0131] Stage sequence deviation: The degree of difference between the sequence of the three stages—action, temperature rise, and partial discharge—in the current link and the stage sequence of the baseline sample. For example, the normal stage sequence of an insulation discharge link is "action deviation - temperature rise retention - partial discharge enhancement." If the stage sequence of the current link is "partial discharge enhancement - temperature rise retention - action deviation," then the stage sequence deviation is extremely large (the timing is completely reversed).

[0132] Response intensity evolution deviation: The degree of difference between the intensity change trend of the current link at each stage and the evolution pattern of the baseline sample. For example, the temperature rise response of a contact heating link usually shows an evolution pattern of "rise-stabilize-slow fall" with load changes. If the temperature rise response of the current link shows a short-term pulse characteristic of "sharp rise-sharp fall", then the response intensity evolution deviation is large.

[0133] As a further preferred implementation, deviation amount The specific calculation method is as follows: Phase path deviation: The value is 0 when the phase path of the current link is completely consistent with the phase path of the baseline sample; the value is 0.3 when the phases are different but the path structure is the same (e.g., the baseline is "Phase A operation - Phase A heat generation", and the current is "Phase B operation - Phase B heat generation"); the value is 1 when the path crosses phases (e.g., the baseline is "Phase A operation - Phase A heat generation", and the current is "Phase A operation - Phase B heat generation").

[0134] Stage Sequence Deviation: The sequence of the three stages—action, temperature rise, and partial discharge—is encoded as a three-dimensional vector. The normalized Hamming distance between the current link stage sequence vector and the baseline stage sequence vector is calculated. For example, if the stage sequence of the baseline insulation discharge link is (action, temperature rise, partial discharge), it is encoded as follows: If the current link stage sequence is (action, partial discharge, temperature rise), the encoding is as follows: The degree of deviation in the stage sequence is the proportion of the two vectors at different positions.

[0135] Degree of deviation in response intensity evolution: The intensity change trend of the current link at each stage (such as the rate of temperature rise, the rate of temperature fall, the rate of partial discharge enhancement, etc.) is normalized and the corresponding trend of the baseline sample is calculated, and the average value of the absolute value of the difference at each stage is taken.

[0136] Deviation The result can be a weighted sum of the deviations from the three dimensions, with the weighting coefficients adjustable based on different operational scenarios and priorities. The sum of the weighting coefficients is 1, and the default values ​​are... The specific value of the weighting coefficient can be adjusted based on actual operating experience for different voltage levels and cabinet types.

[0137] The formula for calculating the state prediction result can be:

[0138] in, Indicates the first The state prediction and acceptance results corresponding to each candidate link; This represents the set of cloud-based state baselines that are allowed to participate in the comparison after the link closure requirement is met. Indicates the current link relative to the first Deviation from the baseline of the class state includes phase path deviation, stage sequence deviation, and response intensity evolution deviation; This indicates that the baseline with the smallest deviation is selected.

[0139] In other words, the degree of "unlikeness" (deviation) of the current link is calculated by comparing it with three types of baselines (normal heating, contact heating, and insulation discharge), and then the type with the "smallest degree of unlikeness" is selected as the prediction result. If the deviation of the current link from the insulation discharge baseline is the smallest, it means that it is most likely to be an insulation discharge fault.

[0140] Specifically: like For the corresponding contact heating link, the output "Contact Heating Attention" indicates that the current link exhibits the characteristics of "action deviation → temperature rise retention in the same phase", but has not yet developed into the partial discharge stage. Maintenance personnel should pay attention to the contact status of this phase and check whether there are any problems with the contact pressure or contact area.

[0141] like The corresponding insulation discharge link, and the closure of the preceding link meets the requirements (i.e., the calculated link closure). If the value exceeds the preset threshold, an insulation degradation warning will be output, indicating that the current link fully exhibits the three-stage characteristics of "action deviation - phase temperature rise retention - enhanced partial discharge in adjacent insulation areas". The physical logic of the link is valid, the risk of insulation degradation is high, and maintenance personnel should arrange maintenance in a timely manner.

[0142] If no state baseline enters (That is, the closure of all candidate links is below the threshold and no link passes the screening), then the output source mismatch check or sensor interference check indicates that although there is action, temperature rise or partial discharge response in the current data, they cannot form a closed physical link. This may be caused by signal crosstalk, sensor failure or external interference. The operation and maintenance personnel should check the sensor status and data quality.

[0143] In actual operation, the following special situations may occur: If the current closed link is similar to the contact heating link or insulation discharge link sample in terms of phase path, stage sequence, and response intensity evolution, then the contact heating concern or insulation degradation warning will be output according to the link stage.

[0144] If the partial discharge response is strong but the reliability of the compartmental response attribution is dispersed (i.e., the calculated response is...), then... If the value is low, the source rejection values ​​of multiple candidate compartments are close, and it is impossible to determine a unique source, or if the partial discharge path is inconsistent with the preceding temperature rise path (the partial discharge source is not located near the insulation in the temperature rise area), then the source mismatch verification will be output, prompting the maintenance personnel that the current data may have cross-phase crosstalk or signal aliasing.

[0145] If the temperature rise is abnormally significant but the partial discharge does not generate a homologous response (i.e., temperature rise - partial discharge bearing capacity), If the temperature rise is too high and the partial discharge segment is not effectively connected, then the contact status should be monitored, and subsequent partial discharge windows should be tracked. This indicates that it is not yet possible to confirm whether the temperature rise has caused insulation degradation, and it is necessary to continuously monitor whether the partial discharge will increase in subsequent windows.

[0146] The operating status of the high-voltage switchgear is dynamically changing. When the following triggering events occur, S102-S105 will be re-executed on the edge side: New opening and closing actions occur in the switchgear (such as circuit breaker operation, trolley pushing in or pulling out); the load current changes significantly (such as exceeding the preset change rate threshold); new abnormalities occur at temperature rise points (such as the temperature of a certain phase exceeding the preset alarm threshold); the partial discharge sensor generates a strong response (such as the partial discharge amplitude exceeding the preset threshold); the cloud baseline is updated (if other switchgear of the same model adds valid link samples, the cloud retrains and distributes the updated baseline).

[0147] If the new link result is consistent with the historical link after recalculation (such as the same phase and the same link type appearing again), the risk level of the phase compartment path is increased, indicating that the faulty link is repetitive or persistent, and the risk is higher than that of an occasional event.

[0148] If the new link result is inconsistent with the historical link (e.g., the historical link was judged as contact heating, while the new link is judged as source mismatch), the response is marked as a sample to be separated and uploaded to the cloud to update the multi-cabinet attribution closure model. This indicates that the abnormal pattern has not appeared in the historical baseline and needs to be uploaded to the cloud as a new sample to participate in subsequent baseline updates.

[0149] Through the aforementioned dynamic update mechanism, the system can continuously learn new failure modes, constantly improve the cloud baseline library, and enhance the accuracy and adaptability of state prediction.

[0150] In summary, in this embodiment of the invention, by establishing candidate pairs between the action, temperature rise, and partial discharge response and candidate compartments, and calculating the source rejection quantity and attribution reliability based on multi-dimensional candidate difference terms, the precise locking of the response source is achieved, effectively reducing the risk of phase compartment source mismatch caused by signal crosstalk under a metal-enclosed structure. Furthermore, by pairing time-adjacent action-temperature rise and temperature rise-partial discharge pairs on the same phase path as candidate receiving pairs, and calculating the receiving break quantity and link closure, it is possible to determine whether the three physical causal chains of "operational deviation - current-carrying temperature rise retention - insulation partial discharge enhancement" follow the same phase path. The fully integrated quantitative judgment provides clear and interpretable physical link basis for the status prediction results, avoiding false alarms caused by purely data-driven models that forcibly splice anomalies from different phases and compartments into the same risk event. Furthermore, the deviation comparison between the link closure calculated on the edge side and the cloud-based baseline of the same type of link is made up for by using historical samples from multiple cabinets to compensate for the lack of samples from a single cabinet. At the same time, the edge side retains the real-time judgment right of link closure, while the cloud only serves as a baseline reference. This ensures the real-time nature of on-site decision-making and improves the accuracy of prediction results, significantly enhancing the overall reliability of high-voltage switchgear operating status prediction.

[0151] This invention also proposes an intelligent prediction system for the operating status of high-voltage switchgear; please refer to [link / reference]. Figure 2 The diagram shows a structural diagram of an intelligent prediction system for the working status of a high-voltage switchgear provided in an embodiment of the present invention. The system includes: a data acquisition module 101, an edge processing module 102, and a cloud baseline module 103.

[0152] The data acquisition module 101 is used to treat the circuit paths of each phase in the same physical compartment as independent monitoring units, and collect their action response, temperature rise response and partial discharge response data respectively, and align them according to the time window to form action sequence, temperature rise sequence and partial discharge sequence; The edge processing module 102 is used to generate action compartment candidate pairs, temperature rise compartment candidate pairs and partial discharge compartment candidate pairs from the three sequences respectively, based on the association between each response and the candidate compartment and phase path; From the candidate pairs, the action compartment candidate pairs that are in the same phase and adjacent in time are paired with the temperature rise compartment candidate pairs to form action temperature rise candidate succession pairs, and the temperature rise compartment candidate pairs are paired with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs. For each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. The source rejection quantity is calculated comprehensively, and the source rejection quantities of each candidate compartment under the same response are ranked. The attribution credibility is determined based on the smallest and second smallest difference. Candidate pairs that meet the requirements of attribution credibility are selected to form candidate links. The action temperature rise bearing fracture amount corresponding to the action temperature rise candidate receiving pair and the temperature rise partial discharge bearing fracture amount corresponding to the temperature rise partial discharge candidate receiving pair are determined. The link closure is determined based on the attribution credibility of the initial action and the two bearing fracture amounts. The cloud baseline module 103 is used to compare the link closure with the cloud baseline, remove links with insufficient closure, calculate the deviation from various baselines, and select the baseline with the smallest deviation as the prediction result.

[0153] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer equipment can be divided into different functional modules to complete all or part of the functions described above. In addition, the intelligent prediction system for the working status of high-voltage switchgear and the intelligent prediction method for the working status of high-voltage switchgear provided in the above embodiments belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0154] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0155] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0156] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent prediction of the operating status of a high-voltage switchgear, characterized in that, include: Each phase circuit path in the same physical compartment is treated as an independent monitoring unit, and its action response, temperature rise response and partial discharge response data are collected separately. The action sequence, temperature rise sequence and partial discharge sequence are formed by aligning them according to the time window. Based on the association between each response and the candidate compartments and phase paths, candidate pairs of action compartments, candidate pairs of temperature rise compartments, and candidate pairs of partial discharge compartments are generated from the three sequences, respectively. From the candidate pairs, the action compartment candidate pairs that are in the same phase and adjacent in time are paired with the temperature rise compartment candidate pairs to form action temperature rise candidate succession pairs, and the temperature rise compartment candidate pairs are paired with the partial discharge compartment candidate pairs to form temperature rise partial discharge candidate succession pairs. For each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. The source rejection quantity is calculated comprehensively, and the source rejection quantities of each candidate compartment under the same response are ranked. The attribution credibility is determined based on the smallest and second smallest difference. Candidate pairs that meet the requirements of attribution credibility are selected to form candidate links. The action temperature rise bearing fracture amount corresponding to the action temperature rise candidate receiving pair and the temperature rise partial discharge bearing fracture amount corresponding to the temperature rise partial discharge candidate receiving pair are determined. The link closure is determined based on the attribution credibility of the initial action and the two bearing fracture amounts. By comparing the link closure with the cloud baseline, removing links with insufficient closure, calculating the deviation from various baselines, and selecting the baseline with the smallest deviation as the prediction result.

2. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The method of treating each phase's circuit path within the same physical compartment as an independent monitoring unit, and separately collecting its action response, temperature rise response, and partial discharge response data, aligning them according to time windows to form action sequences, temperature rise sequences, and partial discharge sequences, specifically includes: Treat the circuit paths of each phase within the same physical compartment as independent monitoring units; For each independent monitoring unit, the opening and closing action data, mechanism stroke curve, opening distance, overtravel, closing bounce, opening and closing speed and three-phase synchronicity data of each phase are collected to form an action sequence. The action sequence is used to record the position deviation of each phase contact and mechanism during one opening and closing process. Collect load current, contact temperature, busbar connection point temperature and cable terminal temperature data for each phase to form a temperature rise sequence. The temperature rise sequence is used to record the temperature rise retention of contacts, busbar connection points or cable terminals during the load stage after closing. Data on UHF partial discharge, TEV partial discharge, ultrasonic partial discharge, and pulsed current partial discharge of each phase are collected to form a partial discharge sequence. The partial discharge sequence is used to record the response intensity, arrival order, and response duration of each partial discharge sensor within the same evaluation period. Align the action sequence, temperature rise sequence, and partial discharge sequence according to the switch cabinet number, phase number, compartment number, sensor number, and time window.

3. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The step of generating candidate pairs of action compartments, candidate pairs of temperature rise compartments, and candidate pairs of partial discharge compartments from the three sequences based on the association between each response and the candidate compartments and phase paths specifically includes: Based on the location of the acquisition sensors, the structural relationship of the compartments, and the loop connection relationship corresponding to each response, determine the phase to which each response belongs and the candidate compartment; From the action sequence, each action response is associated with its corresponding phase and candidate compartment to generate action compartment candidate pairs, where the candidate compartments of the action response include the circuit breaker compartment and the contact box; From the temperature rise sequence, each temperature rise response is associated with its corresponding phase and candidate compartment to generate candidate temperature rise compartment pairs. The candidate compartments for the temperature rise response include contact boxes, busbar connection points, and cable termination areas. From the partial discharge sequence, each partial discharge response is associated with its corresponding phase and candidate compartment to generate partial discharge compartment candidate pairs. The candidate compartments of the partial discharge response include cable compartments, bus bushings, contact boxes, adjacent insulation areas, and insulation partition areas.

4. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The process of pairing candidate pairs of action compartments that are in the same phase and temporally adjacent with candidate pairs of temperature rise compartments to form action-temperature rise candidate succession pairs, and pairing candidate pairs of temperature rise compartments with candidate pairs of partial discharge compartments to form temperature rise-partial discharge candidate succession pairs, specifically includes: From the candidate pairs of action compartments and the candidate pairs of temperature rise compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate action temperature rise candidate succession pairs. Among them, the action temperature rise candidate succession pairs are used to characterize whether there is a succession relationship between the deviation of opening and closing and the subsequent current-carrying temperature rise retention. From the candidate pairs of temperature rise compartments and the candidate pairs of partial discharge compartments, candidate pairs belonging to the same phase and adjacent time windows are selected for pairing to generate candidate pairs of temperature rise partial discharge. The candidate pairs of temperature rise partial discharge are used to characterize whether there is a succession relationship between the current-carrying temperature rise retention and the subsequent partial discharge enhancement.

5. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, For each candidate pair, candidate difference terms are extracted to characterize the degree of deviation between the response source and the candidate compartment. Their source rejection quantities are then calculated comprehensively. The source rejection quantities of each candidate compartment under the same response are ranked, and the attribution reliability is determined based on the smallest and second smallest difference. Specifically, this includes: For each candidate response compartment, candidate difference items are extracted relative to the candidate compartment. These candidate difference items include the degree of loop path disconnection, the degree of response timing anomaly, the adjacent deviation of temperature rise points, the deviation of partial discharge propagation attenuation, and the deviation of sensor installation direction. The sum of the candidate differences for the same candidate compartment is divided by the number of candidate differences to obtain the source rejection of the response relative to the candidate compartment. The source rejection factors of each candidate compartment under the same response are sorted from smallest to largest. The candidate compartment corresponding to the smallest source rejection factor is taken as the predicted source of the response, and the candidate compartment corresponding to the second smallest source rejection factor is taken as the competing source of the response. The attribution confidence of the compartment response is obtained by subtracting the source repulsion of the predicted source from the source repulsion of the competing source and then dividing by the source repulsion of the competing source.

6. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 5, characterized in that, The degree of disconnection of the loop path is determined based on the length of the connection path between the candidate compartment and the response source in the primary loop topology table. The value is 0 when there is a direct connection and 1 when there is no connection path. When there are intermediate nodes, the degree of disconnection is determined based on the number of intermediate nodes, and the degree of disconnection of the loop path is positively correlated with the number of intermediate nodes. The degree of response timing anomaly is determined based on the deviation between the response timestamp and the expected time window; The adjacent deviation of the temperature rise point is determined based on the length of the heat conduction path between the temperature sensor and the candidate compartment. The deviation of partial discharge propagation attenuation is determined by the difference between the theoretical amplitude calculated by the signal propagation attenuation model and the actual amplitude; The sensor installation orientation deviation is determined based on the angle between the sensor orientation and the candidate compartment direction.

7. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The candidate pairs whose attribution reliability meets the requirements are formed into candidate links, and the fracture amount of the action temperature rise candidate receiving pair and the fracture amount of the temperature rise partial discharge candidate receiving pair are determined, specifically including: From the candidate pairs whose attribution credibility meets the preset requirements, select the action compartment candidate pairs, temperature rise compartment candidate pairs and partial discharge compartment candidate pairs on the connection path of the same phase or adjacent compartments, and form candidate links along the primary loop path in the order of the action, temperature rise and partial discharge stages. For the candidate temperature rise receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate temperature rise receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of the temperature rise receiving. For the candidate partial discharge receiving pairs in the candidate link, obtain the actual fracture degree and preset allowable deviation value of each item in the candidate partial discharge receiving pairs, and retain the part of the actual fracture degree that exceeds the preset allowable deviation value, and take the average value of all the excess parts as the fracture amount of partial discharge receiving. Among them, the actual degree of fracture of each item in the candidate temperature rise of the action includes the degree of discontinuity of the phase path, the degree of non-stagnation of temperature rise after load correction, and the degree of non-acceptance of the time window. The actual degree of fracture in the partial discharge candidate pair includes the degree of insufficient insulation proximity, the degree of abnormal response timing of partial discharge occurring before temperature rise, the degree of path misalignment of the partial discharge path in other phase compartments, and the degree of interference window not being eliminated.

8. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The determination of link closure based on the reliability of the initial action and the amount of breakage in the two segments specifically includes: The reliability of the initial action is used as the link support quantity, and the sum of the reliability of the initial action, the temperature rise failure rate, and the temperature rise partial discharge failure rate is used as the total link deviation quantity. The link closure is obtained by dividing the link support quantity by the total link deviation quantity.

9. The intelligent prediction method for the operating status of a high-voltage switchgear according to claim 1, characterized in that, The process of comparing link closure with the cloud baseline, eliminating links with insufficient closure, calculating the deviation from various baselines, and selecting the baseline with the smallest deviation as the prediction result specifically includes: Compare the link closure with a preset closure threshold and eliminate candidate links whose link closure is lower than the preset threshold; The remaining candidate links are compared with the cloud-based normal heating baseline, contact heating baseline, and insulation discharge baseline, respectively. The deviation between each candidate link and each type of baseline is calculated. The deviation includes the degree of deviation of the phase path, the degree of deviation of the stage sequence, and the degree of deviation of the response intensity evolution. The baseline type with the smallest deviation is selected as the prediction result for the candidate link; If the baseline with the smallest deviation is the contact heating baseline, then output "Contact heating attention"; if it is the insulation discharge baseline and the link closure meets the requirements, then output "Insulation degradation warning"; if no candidate link passes the closure threshold screening, then output "Source mismatch verification".

10. A high-voltage switchgear operating status intelligent prediction system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent prediction method for the working status of a high-voltage switchgear as described in any one of claims 1-9.

Images