A switch cabinet contact temperature and partial discharge on-line monitoring method based on multi-parameter fusion

Abstract

The application relates to the technical field of switch cabinets and discloses a switch cabinet contact temperature and partial discharge on-line monitoring method based on multi-parameter fusion, which divides a cabinet body into four independent collection partitions of a busbar room, a circuit breaker room, a cable room and a secondary instrument room, configures partition independent collection switches and adopts a time-sharing interlocking polling collection mechanism to completely eliminate the multi-measurement-point wireless collection crosstalk problem; meanwhile, a three-dimensional X / Y / Z space coordinate system is built for each partition, a partition exclusive temperature threshold system and a three-dimensional space temperature correction model are established in combination with the temperature gradient characteristics of different partitions and different height intervals, the cabinet body space heat accumulation temperature rise and the equipment body fault temperature rise are accurately distinguished, and the defects of the traditional temperature measurement, such as upper layer false alarm and lower layer missed alarm, are solved.

Description

Technical Field

[0001] This invention belongs to the field of electrical equipment technology, and in particular relates to an online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion. Background Technology

[0002] High-voltage switchgear has a sealed metal cavity structure, internally divided into multiple independent cavities according to electrical functions. The equipment layout, heat source distribution, ventilation conditions, and electric field environment of different cavities vary significantly, forming a naturally non-uniform three-dimensional temperature field. Contact overheating and partial discharge are the two most critical types of faults in switchgear, and these two types of faults exhibit a strong coupling evolution mechanism: continuous contact overheating reduces the partial discharge initiation voltage of the insulation material, accelerating the development of insulation defects; partial discharge arcs burning the contacts increase contact resistance, further exacerbating the temperature rise, forming a vicious cycle that triggers compound faults.

[0003] Despite years of development, most existing online monitoring technologies for switchgear still suffer from the following shortcomings: First, the temperature measurement mode ignores the three-dimensional temperature gradient within the cabinet, leading to false alarms and missed alarms. Traditional temperature measurement systems use a single global temperature threshold and indiscriminate temperature measurement correction logic, failing to differentiate the spatial characteristics of different functional zones within the switchgear. Due to the principle of hot air rising, the temperature in the upper busbar compartment of the cabinet is generally 5-12°C higher than that in the lower cable compartment. The traditional uniform threshold can cause heat accumulation in the upper space to be misjudged as equipment failure, frequently triggering false alarms; while slight temperature rises in the well-ventilated lower areas are easily masked by the global threshold, resulting in missed early fault detections. Furthermore, existing technology does not establish a correlation between spatial coordinates and temperature measurement data, making it impossible to distinguish between "actual heat generation of the equipment itself" and "heat accumulation in the cabinet's spatial environment," resulting in a flawed temperature measurement principle.

[0004] Secondly, the switch cabinet has dozens of wireless temperature measurement and partial discharge acquisition nodes densely arranged inside. The existing technology adopts a full-node synchronous acquisition mode, which is prone to problems such as wireless crosstalk, time slot conflict, data misalignment, and temperature jump. This leads to distortion of the original temperature measurement data and partial discharge signal, and subsequent data processing and fault diagnosis lose the accurate data foundation and cannot adapt to the complex electromagnetic environment inside the cabinet.

[0005] Third, contact temperature is a slow-changing steady-state parameter on the order of seconds, reflecting the long-term operating status of the equipment; partial discharge is a transient pulse parameter on the order of microseconds, reflecting instantaneous insulation defects in the equipment. Existing technologies generally use a unified timestamp for both types of parameters without performing time-series adaptation and feature matching. Directly fusing them through standard DS evidence theory is prone to causing high levels of evidence conflict. The global conflict probability equalization mechanism of the standard algorithm will lead to oscillations in the fusion results and a significant decrease in credibility, failing to truly reflect the operating status of the equipment.

[0006] Fourth, existing temperature measurement and partial discharge monitoring systems are independent and their data is not interconnected. They can distinguish between simple overheating or simple partial discharge faults, but cannot identify the combined degradation defects resulting from the coupled evolution of these two types of faults. Traditional monitoring technologies cannot achieve fine-grained fault classification, leading to blind troubleshooting and low maintenance efficiency, making it difficult to support the differentiated operation and maintenance needs of smart grids. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide an online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion, thereby solving the problems mentioned in the background technology.

[0008] This invention provides the following technical solution: An online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion includes the following steps: S1: Cabinet Zone Three-Dimensional Spatial Temperature Acquisition: Based on the electrical function and three-dimensional spatial structure of the switchgear, the cabinet is divided into four independent acquisition zones: busbar compartment, circuit breaker compartment, cable compartment, and secondary instrument compartment. Each zone is equipped with an independent controllable acquisition switch. A zone-based time-interlocked polling acquisition mechanism is adopted, with only one zone acquisition switch activated at any given time. The raw temperature data of the temperature measurement nodes in each zone are collected sequentially. Simultaneously, the three-dimensional spatial coordinate information of each zone, the ambient temperature and humidity inside the cabinet, the three-phase load current of the switchgear, and the raw partial discharge signals of the cabinet's UHF, TEV, and AE channels are collected. All acquired data are uniformly timestamped and zonal spatially tagged to complete the preliminary timing matching of heterogeneous parameters. S2: Three-dimensional spatial temperature feature extraction of each zone: Based on the three-dimensional spatial coordinate system and zone attributes established in step S1, match the corresponding zone-specific temperature reference threshold, Z-axis height thermal gradient coefficient k, XY plane ventilation heat transfer coefficient k and shading coefficient S, and complete the layered and step-by-step accurate correction by relying on the preset three-dimensional spatial temperature gradient correction model to eliminate the temperature measurement system error caused by cabinet hot air floating, zone ventilation differences and spatial heat accumulation, and restore the true temperature of the electrical equipment body; S3: Data preprocessing: Adaptive bandpass filtering is applied to the three raw partial discharge signals of UHF, TEV and AE. Based on the real-time noise window, the dynamic 3σ noise threshold is statistically analyzed to remove environmental interference pulses, effectively retain the effective partial discharge signal, and extract the partial discharge pulse amplitude, pulse repetition frequency and power frequency phase distribution characteristics to construct multi-channel partial discharge feature vectors. S4: Layered Data Fusion: The first layer completes the fusion of partial discharge characteristic evidence from three sources: UHF, TEV, and AE, and outputs a single-dimensional partial discharge degradation confidence level; the second layer performs cross-dimensional heterogeneous parameter fusion of partial discharge confidence evidence, three-dimensional temperature characteristic evidence after partition correction, and load current evidence; configures dynamic confidence level assignment rules based on operating condition adaptation, introduces conflict evidence local allocation strategy to resolve evidence conflicts generated by fast and slow heterogeneous parameter fusion, and outputs the comprehensive degradation confidence level of switchgear contacts; S5: Graded Early Warning: Retrieves the dedicated fault threshold library for each zone, and combines the comprehensive degradation confidence, the spatiotemporal evolution trend of zone temperature, and the continuous stability of partial discharge pulse to distinguish between overheating faults caused by increased contact resistance, partial discharge faults in the internal insulation of the switchgear, and temperature-partial discharge symbiotic composite degradation faults. Outputs multi-level early warning signals based on the severity level of the defect, and locally stores monitoring data and spatial temperature field data for fault querying.

[0009] Preferably, the zone-based time-interlocked polling acquisition mechanism described in step S1 is as follows: the system initializes all zone acquisition switches to the off state, and polls and turns on each zone acquisition switch one by one according to the fixed timing sequence of bus room → circuit breaker room → cable room → secondary instrument room. After the acquisition of a single zone is completed, the corresponding switch is immediately turned off, and then the acquisition switch of the next zone is turned on, so as to realize the electrical isolation and wireless signal isolation of the zone acquisition and prevent cross-zone data crosstalk and timing conflicts.

[0010] Preferably, in step S1, a dedicated local three-dimensional spatial coordinate system is built for each independent acquisition zone to achieve accurate modeling and differential correction of the spatial dimension of temperature measurement data. Specifically, this includes: constructing a local rectangular coordinate system with the core current-carrying component of each zone's equipment as the reference origin; defining the X-axis as the left-right longitudinal installation direction of the cabinet to characterize the differences in equipment arrangement and obstruction; defining the Y-axis as the front-to-back air intake and exhaust ventilation direction of the cabinet to characterize the differences in ventilation, convection, and heat transfer; and defining the Z-axis as the vertical height gradient direction of the cabinet to characterize the vertical temperature gradient of rising hot air, which is the core gradient dimension of the cabinet's temperature field. Each temperature measurement node in each zone is bound to a unique (X1, Y1, Z1) three-dimensional spatial coordinate parameter, and the reference coordinates (X, Y, Z) of each zone are preset as the temperature correction reference point for that zone, realizing one-to-one traceability and matching of each set of temperature data, spatial location, zone attributes, and environmental conditions, providing accurate spatial parameter support for the three-dimensional spatial gradient correction in step S2.

[0011] Preferably, the specific temperature ranges for each zone in step S2 are divided as follows: Upper busbar compartment: normal temperature range 25℃~55℃, first-level warning range 55℃~65℃, second-level fault range >65℃; Middle circuit breaker compartment: normal temperature range 25℃~45℃, first-level warning range 45℃~55℃, second-level fault range >55℃; Lower cable compartment: normal temperature range 20℃~40℃, first-level warning range 40℃~48℃, second-level fault range >48℃; Secondary instrument compartment: normal temperature range 20℃~35℃. Temperatures exceeding these ranges are considered abnormal temperature rises.

[0012] Preferably, the formula for the three-dimensional spatial temperature gradient in step S2 is: T real =T raw -k z ∆Zk xy S xy Among them, T real T represents the actual temperature of the device after spatial correction. raw k is the raw temperature measurement data collected by the zone acquisition switch. z The axial height thermal gradient coefficient is used to quantify the inherent temperature deviation caused by the upward movement of hot air from the vertical direction within the cabinet. Differential calibration is applied across zones: upper busbar compartment k∈[0.4,0.6]℃ / cm, middle circuit breaker compartment k∈[0.1,0.2]℃ / cm, lower cable compartment and secondary instrument compartment k∈[-0.2,-0.1]℃ / cm; ∆Z is the height difference between the temperature measurement point and the zone reference height, in cm. A positive value indicates the measurement point is above the reference height, and a negative value indicates it is below the reference height; k xy The X / Y plane ventilation heat transfer coefficient is adaptively selected based on the cabinet opening and closing and vent blockage conditions: 0.2 for smooth ventilation, 0.5 for weak ventilation, and 0.8 for sealed and congested conditions; S xy The coefficient represents the degree of ventilation obstruction at the measuring point plane, with 0 for no obstruction, 0.5 for partial obstruction, and 1 for severe obstruction.

[0013] Preferably, the temperature data preprocessing in step S2 specifically includes: setting the effective temperature measurement range [-20℃, 150℃], removing wireless communication outliers and replacing them with adjacent effective data interpolation; and using a sliding median filtering algorithm with a window length of N=5 to suppress temperature jumps caused by instantaneous electromagnetic pulses and output a stable and reliable filtered temperature sequence.

[0014] Preferably, the dynamic noise threshold in step S3 satisfies V th =V avg +3σ, select a 20s real-time noise sliding window, statistically analyze the mean signal amplitude Vavg and standard deviation σ within the window, adaptively remove environmental interference pulses under different operating conditions, and retain the effective partial discharge pulse sequence.

[0015] Preferably, the dynamic confidence assignment rule in step S4 is as follows: when the temperature rise rate of the core area of ​​the circuit breaker compartment is greater than 1.5℃ / 10min, the weight of the three-dimensional temperature feature evidence is increased; when a valid partial discharge pulse is detected stably for 5 consecutive minutes, the weight of the partial discharge evidence is increased; when the load current reaches more than 90% of the rated current, the weight of the load parameter is increased; the conflict evidence local allocation strategy is as follows: when the temperature evidence and the partial discharge evidence judgment results conflict, the traditional global equal allocation method is abandoned, and the conflict probability is weighted according to the historical fault identification accuracy of the two types of evidence, which effectively suppresses the oscillation of the fusion result.

[0016] Preferably, the coupling fault identification logic in step S5 is as follows: if the actual temperature of the contact continues to rise after correction and there is no effective partial discharge pulse, it is determined to be a contact failure and overheating fault; if an effective partial discharge pulse is continuously detected and there is no continuous abnormal temperature rise, it is determined to be a partial discharge fault in the cabinet insulation; if the temperature continues to exceed the standard and is accompanied by a stable partial discharge pulse, it is determined to be a temperature-partial discharge symbiotic composite degradation fault; at the same time, combined with the overall distribution of the zoned temperature field, it distinguishes between single-point equipment faults and global space heat accumulation.

[0017] Preferably, all data acquisition, temperature correction, feature extraction, multi-parameter fusion, and fault identification processes are deployed on the local edge terminal of the switchgear. Only early warning information, fault segment data, and abnormal spatial temperature field data are uploaded to the remote platform, while the original acquired data is stored locally, enabling on-site real-time monitoring and early warning.

[0018] Compared with the prior art, the present invention has the following beneficial effects: This invention discloses an online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion. The switchgear is configured with independent acquisition switches according to functional zones, and a time-division interlocking polling acquisition mechanism is adopted to achieve electrical and wireless isolation of cross-zone signals, preventing crosstalk, time slot conflicts, and data distortion problems of traditional synchronous acquisition, thereby improving the purity and synchronization of the original temperature measurement data and partial discharge signal acquisition from the source.

[0019] To address the three-dimensional spatial temperature gradient characteristics of the cabinet, a zone-specific temperature threshold system and spatial gradient correction model were established to distinguish between spatial heat accumulation temperature rise and equipment failure temperature rise. This solved the industry problem of false alarms due to upper-level heat accumulation and missed alarms due to lower-level equipment in traditional technologies, and achieved unified normalization of temperature measurement data from different spatial locations, significantly improving temperature measurement accuracy.

[0020] To address the heterogeneous characteristics of gradually changing temperature and transient partial discharge, a two-level hierarchical fusion architecture is adopted to fuse homogeneous and cross-domain parameters step by step. Combined with dynamic confidence assignment and conflict evidence local allocation strategy, the results are stable and reliable, without frequent oscillation issues.

[0021] By combining the spatiotemporal trends of zoned temperature with the characteristics of partial discharge pulses, three types of faults—simple overheating, simple partial discharge, and combined degradation—can be accurately distinguished. The evolution mechanism and risk level of different faults can be clarified, providing maintenance personnel with accurate maintenance basis, significantly improving maintenance efficiency and reducing power outage losses. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a flowchart of the overall method of the present invention.

[0024] Figure 2 This is a schematic diagram of the cabinet partition three-dimensional space acquisition architecture of the present invention.

[0025] Figure 3 This is a schematic diagram of the partition temperature correction and feature extraction process of the present invention.

[0026] Figure 4 This is a flowchart of the fault classification and processing logic of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0028] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0029] Example: like Figure 1-4 As shown, an online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion includes the following specific steps: S1. Cabinet Partitioned Three-Dimensional Spatial Temperature Acquisition: Based on the electrical functions and three-dimensional spatial structure of the switchgear, the cabinet is divided into four independent acquisition zones: busbar compartment, circuit breaker compartment, cable compartment, and secondary instrument compartment. Each zone is equipped with an independent controllable acquisition switch, achieving physical partitioning isolation of the acquisition hardware. A local three-dimensional rectangular coordinate system is constructed for each independent acquisition zone to unify spatial reference standards: the installation center of the core electrical equipment in each zone is taken as the zone reference origin (X, Y, Z). The X-axis is defined as the left and right longitudinal installation direction of the cabinet, representing the difference in equipment arrangement and obstruction; the Y-axis is defined as the front and rear air intake and exhaust ventilation direction of the cabinet, representing the difference in convective heat transfer intensity; the Z-axis is defined as the vertical height gradient direction of the cabinet, which is the core gradient dimension of the cabinet temperature field, representing the vertical temperature deviation caused by the rising of hot air. All temperature measurement nodes in each zone are bound to a unique and fixed (X1, Y1, Z1) three-dimensional spatial coordinate, establishing a unique mapping relationship of "zone-spatial coordinate-measuring point", providing accurate spatial parameter support for subsequent differentiated spatial temperature correction. Based on this, a zone-based, time-interlocked, polling acquisition mechanism is adopted. The system initializes all zone acquisition switches to a fully closed state. Following a fixed timing sequence of busbar compartment → circuit breaker compartment → cable compartment → secondary instrument compartment, the zone acquisition switches are polled and activated one by one, strictly adhering to the principle of "single-zone acquisition, single-zone activation." Only one zone acquisition switch is activated at any given time. After the data acquisition of a single zone is completed, the current zone switch is immediately turned off before the next zone switch is activated, preventing cross-zone wireless crosstalk, time slot conflicts, and data misalignment. The system collects the raw temperature data of the temperature measurement nodes in each zone, and simultaneously collects the three-dimensional spatial coordinate parameters of each zone, the ambient temperature and humidity inside the cabinet, the real-time three-phase load current of the switch cabinet, and the raw partial discharge signals of the cabinet's UHF, TEV, and AE channels. All acquired data are uniformly stamped with high-precision timestamps, zone numbers, and three-dimensional spatial tags, completing the preliminary timing registration and structured encapsulation of multi-source fast and slow heterogeneous parameters.

[0030] Step S1, specifically the calibration of the local 3D spatial coordinate system and the binding of measurement points for each independent acquisition zone, includes: (1) Zoning reference origin calibration: The core current-carrying contact / main busbar installation center of each zone is taken as the reference origin (X,Y,Z) of the zone rectangular coordinate system. During the factory commissioning stage of the switchgear, the three-dimensional coordinate origin calibration is completed by laser rangefinder, and the origin coordinates are solidified and stored in the corresponding zone parameter library of the edge terminal. (2) Standardized definition of three-axis directions: X-axis: The left and right longitudinal installation direction of the cabinet. The X-axis value represents the left and right arrangement and obstruction distance of the temperature measuring points relative to the reference origin. Y-axis: The air intake and exhaust direction of the cabinet. The Y-axis value represents the airflow distance of the measuring point relative to the reference origin. Z-axis: The vertical height gradient direction of the cabinet. The Z-axis value represents the vertical height difference between the measuring point and the reference origin, and serves as the core calculation dimension for the temperature gradient of the rising hot air in the cabinet. (3) Unique coordinate binding process for measuring points: After each wireless temperature measuring node and UHF / TEV / AE partial discharge acquisition sensor in the cabinet is installed, the offset (ΔX, ΔY, ΔZ) of the measuring point relative to the partition reference origin is recorded on site, and the absolute three-dimensional coordinates of the measuring point (X1=X+ΔX, Y1=Y+ΔY, Z1=Z+ΔZ) are calculated in combination with the partition origin (X,Y,Z). (4) Spatiotemporal tag synchronous binding: During the time-sharing polling process, when each round of partition collection starts, the (X1,Y1,Z1) bound to all measurement points in that partition are retrieved synchronously.

[0031] The three-dimensional coordinates encapsulate the raw temperature, partial discharge, and load data with the corresponding three-dimensional coordinates of the measuring points, the partition number, and the high-precision acquisition timestamp into a four-dimensional spatiotemporal dataset, enabling one-to-one traceability and matching of each set of monitoring data with the spatial location of the measuring points.

[0032] S2. Three-dimensional spatial temperature feature extraction of zones: Based on the zone attributes and three-dimensional spatial coordinate system established in S1, combined with the exclusive temperature threshold range and differentiated gradient parameters of each zone, and relying on the customized three-dimensional spatial temperature gradient correction model, the inherent temperature measurement system errors caused by the upward floating of hot air in the cabinet, the difference in ventilation between zones, and the obstruction of equipment are eliminated step by step to accurately restore the true working temperature of the electrical equipment. The specific detailed execution steps are as follows: S201. Adaptive matching of zone parameters: According to the zone to which the temperature measurement point belongs, the standard normal temperature range, warning range, and fault range of the corresponding zone are retrieved. At the same time, the exclusive Z-axis height thermal gradient coefficient k and XY plane ventilation heat transfer coefficient k of the zone are matched as the reference values. Among them, k∈[0.4,0.6]℃ / cm for the upper busbar room, focusing on compensating for the upper layer thermal accumulation deviation; k∈[0.1,0.2]℃ / cm for the middle circuit breaker room, slightly correcting the spatial deviation of the core area; k∈[-0.2,-0.1]℃ / cm for the lower cable room and secondary instrument room, compensating for the temperature measurement error caused by the ventilation and heat dissipation of the lower layer. S202. Spatial Deviation Quantification Calculation: Based on the three-dimensional coordinates (X1, Y1, Z1) of the measuring point and the reference coordinates (X, Y, Z) of the zone, calculate the vertical height deviation ΔZ = Z1 - Z. ΔZ is positive when the measuring point is above the reference height and negative when it is below. S203. Three-Dimensional Model Precise Correction Calculation: The three-dimensional spatial temperature gradient formula satisfies: T real =T raw -k z ∆Zk xy S xy In the formula, T real T represents the actual temperature of the device after spatial correction. raw k is the raw temperature measurement data collected by the zone acquisition switch.z The axial height thermal gradient coefficient is used to quantify the inherent temperature deviation caused by the upward movement of hot air from the vertical direction within the cabinet. Differential calibration is applied across zones: upper busbar compartment k∈[0.4,0.6]℃ / cm, middle circuit breaker compartment k∈[0.1,0.2]℃ / cm, lower cable compartment and secondary instrument compartment k∈[-0.2,-0.1]℃ / cm; ∆Z is the height difference between the temperature measurement point and the zone reference height, in cm. A positive value indicates the measurement point is above the reference height, and a negative value indicates it is below the reference height; k xy The X / Y plane ventilation heat transfer coefficient is adaptively selected based on the cabinet opening and closing and vent blockage conditions: 0.2 for smooth ventilation, 0.5 for weak ventilation, and 0.8 for sealed and congested conditions; S xy The ventilation obstruction coefficient for the measuring point plane is set to 0 for no obstruction, 0.5 for partial obstruction, and 1 for severe obstruction. S204. High-precision preprocessing of temperature measurement data: The effective physical temperature measurement range [-20℃, 150℃] is set, outliers from wireless transmission are removed, and missing or abnormal measuring points are replaced by interpolation of adjacent effective data. A sliding median filtering algorithm with a window length of N=5 is used to suppress temperature jumps caused by instantaneous electromagnetic interference inside the cabinet, outputting a continuous, stable, and accurate real temperature time series. S205. Construction of a three-dimensional temperature feature vector: Based on the corrected steady-state temperature sequence, three core state features are extracted, including real-time temperature amplitude, 10-minute short-term dynamic temperature rise rate, and 24-hour long-term baseline offset, to jointly construct a standardized three-dimensional temperature feature vector. This vector comprehensively characterizes the equipment's steady-state temperature level, dynamic temperature rise trend, and long-term aging baseline offset, providing high-precision, high-dimensional temperature feature support for subsequent heterogeneous parameter fusion.

[0033] S3. Data Preprocessing: Addressing the signal characteristics of strong electromagnetic interference, time-varying noise, and weak transient partial discharge pulses within the switchgear cabinet, layered adaptive noise reduction and standardized feature extraction are performed on the synchronously acquired UHF ultra-high frequency, TEV transient voltage to ground, and AE ultrasonic raw signals. This solves the problems of poor adaptability of traditional fixed filter parameters and the easy submersion of effective pulses by noise. Specific detailed steps are as follows: S301. Differentiated Preprocessing of Multiple Signals: For the frequency domain differences of UHF high-frequency electromagnetic signals, TEV pulse voltage signals, and AE acoustic signals, corresponding adaptive bandpass filters are applied to filter out power frequency harmonics, low-frequency vibrations, and high-frequency spurious electromagnetic interference, while retaining the effective partial discharge frequency band components of each signal; S302. Dynamic Adaptive Noise Threshold Calculation: Using a 20-second real-time sliding noise window, the signal amplitude dataset during periods without effective discharge within the window is continuously statistically analyzed. The mean signal amplitude V and standard deviation σ are calculated. An adaptive dynamic noise threshold is constructed based on the 3σ criterion. The threshold calculation formula is: V th =V avg+3σ; S303, Precise Removal of Interference Pulses: The amplitude of the real-time acquired signal is compared point by point with the dynamic threshold V. Signals below the threshold are judged as random environmental noise and removed by setting them to zero. Pulses above the threshold are judged as valid partial discharge pulses and are retained completely, adapting to the noise fluctuation characteristics of different loads, different time periods, and different electromagnetic environments; S304, Multi-dimensional Partial Discharge Feature Extraction: Three types of core effective features are quantitatively extracted from the denoised pure partial discharge pulse sequence, including the peak amplitude of the partial discharge pulse, the pulse repetition frequency per unit time, and the power frequency phase distribution characteristics; S305, Construction of Partial Discharge Feature Vector: The multi-dimensional features of the three partial discharge signals are normalized to the [0,1] interval to eliminate the difference in dimensions and construct a standardized multi-channel partial discharge feature vector, providing a unified dimensional and high signal-to-noise ratio partial discharge feature data source for subsequent hierarchical evidence fusion.

[0034] S4. Layered Data Fusion: Addressing the heterogeneity between temperature-variable steady-state parameters and partial discharge transient pulse parameters, traditional single-layer fusion methods are prone to issues such as conflicting evidence, oscillating results, and distorted credibility. By constructing a two-level layered progressive fusion architecture, coupled with an adaptive dynamic confidence assignment strategy based on operating conditions and a local allocation strategy for conflicting evidence, accurate fusion of multi-source evidence is achieved. Detailed steps are as follows: S401, First-level homogeneous partial discharge evidence fusion: The feature vectors of UHF, TEV and AE homogeneous partial discharges are fused for the first time to construct a single-dimensional partial discharge identification framework. The basic probability allocation functions m, m, and m of each partial discharge sub-evidence are calculated through the basic trust function. The fused output is a single global partial discharge degradation trust degree m, which realizes the complementarity of homogeneous multi-signal information and eliminates the random error of single partial discharge detection. S402. Construction of multi-dimensional evidence sets: Complete the collection of cross-dimensional heterogeneous evidence and form three types of core evidence sets, namely, the evidence of regional three-dimensional corrected temperature characteristics, the evidence of partial discharge degradation of the first-level fusion output, and the evidence of load current condition correlation. S403. Dynamic Confidence Weight Assignment for Operating Conditions: The weights of each piece of evidence are adaptively adjusted based on real-time operating conditions. The specific assignment rules are as follows: When the short-term temperature rise rate in the core contact area of ​​the circuit breaker compartment is greater than 1.5℃ / 10min, the weight of temperature characteristic evidence is increased; when a valid partial discharge pulse sequence is detected stably for 5 consecutive seconds, the weight of partial discharge evidence is increased; when the real-time load current is ≥90% of the rated current, the weight of load condition evidence is increased, thus achieving a strong binding between the weights and fault-sensitive operating conditions. S404. Quantification and Local Allocation Correction of Heterogeneous Evidence Conflicts: A conflict coefficient K is introduced to quantify the degree of conflict among three types of heterogeneous evidence: temperature, partial discharge, and load. This overcomes the shortcomings of the traditional DS algorithm's global equal distribution of conflict probabilities and adopts a local allocation strategy for conflict evidence, distributing the conflict probability to the corresponding focal element based on the historical fault identification accuracy of each piece of evidence. The formula satisfies: In the formula To compensate for the probability of conflicting localities, the fusion oscillations caused by differences in the temporal characteristics of heterogeneous parameters are effectively resolved. Simultaneously, three-dimensional spatial location parameters are introduced during the two-level hierarchical evidence fusion process to complete dynamic weight correction, and evidence conflicts are resolved based on the spatial distance between measurement points. (1) The evidence weight correction method driven by three-dimensional spatial location is as follows: ①Z-axis height weight correction: When the measuring point is located in the high Z-axis region of the busbar room, the basic weight of temperature characteristic evidence is reduced by the same temperature rise magnitude; when the measuring point is located in the low Z-axis core current-carrying region of the circuit breaker room or cable room, the basic weight of temperature characteristic is increased by the same temperature rise magnitude. ② XY Plane Occlusion Weight Correction: For measuring point Sxy=1, under severe occlusion conditions, heat dissipation is poor, and even a small temperature rise indicates equipment degradation; therefore, the weight of temperature evidence is increased. xy =0 Unobstructed ventilation leads to a higher risk of failure at the same temperature rise, further amplifying the weight of temperature evidence; (2) Strategies for resolving heterogeneous evidence conflicts based on spatial distance between measurement points: ① Calculate the three-dimensional Euclidean spatial distance between the overheating temperature measurement point and the effective partial discharge acquisition sensor. ; , Coordinates of the abnormal temperature measurement point. , , The coordinates of the sensor used to detect partial discharge pulses; ② If L≤10cm, it is determined that the temperature anomaly and the partial discharge pulse originate from the same equipment coupling degradation, thereby increasing the synergistic weight of the two types of evidence, temperature and partial discharge, and reducing the probability of conflict allocation; ③ If L>50cm, the temperature abnormality and partial discharge defect are determined to be two independent faults in the cabinet. The probability of conflict between the two types of evidence is allocated according to their respective historical fault identification accuracy, and the confidence output does not interfere with each other. ④10cm < If L≤50cm, adjust the conflict allocation ratio according to linear interpolation of distance. The closer the distance, the higher the correlation between the two types of evidence.

[0035] S405, Comprehensive Deterioration Confidence Output: After iteratively completing the two-level hierarchical fusion calculation, the comprehensive deterioration confidence of the switchgear contacts and insulation system is output. The confidence range is [0,1]. The larger the value, the higher the risk of equipment deterioration and failure.

[0036] S5. Tiered Early Warning: Based on the dedicated temperature threshold library for each zone, the three-dimensional temperature spatiotemporal evolution law, partial discharge pulse stability, and comprehensive degradation confidence, a multi-dimensional coupled fault discrimination logic is constructed to achieve accurate differentiation and multi-level early warning for three typical faults, preventing misjudgment based on a single parameter and missed judgment of coupled faults. The specific steps are as follows: S501. Zone Temperature Field State Determination: Retrieve the dedicated temperature threshold ranges for the bus room, circuit breaker room, cable room, and secondary instrument room, and combine them with the corrected true temperature amplitude, 10-minute short-time temperature rise rate, and 24-hour baseline offset to differentiate... The system classifies temperature states into four categories: steady-state normal, space heat accumulation, early slight temperature rise, and severe overheating. It also uses multi-point temperature range analysis to determine whether temperature anomalies are due to single-point equipment failure or overall space heat exchange anomalies. S502, Partial Discharge (PD) Operation Status Classification: Based on the duration, amplitude stability, and repetition frequency of effective PD pulses, the insulation status is classified into four levels: defect-free, slight intermittent PD, stable continuous PD, and high-intensity destructive PD. S503, Precise Classification and Identification of Coupling Faults: A temperature-PD dual-parameter coupling discrimination model is established to accurately distinguish between... Three core fault types: First, a continuous temperature rise exceeding the limit without a stable and effective partial discharge pulse, determined to be a simple contact overheating fault caused by increased contact resistance, decreased spring pressure, and oxidation and erosion of the contact surface; Second, a continuous and stable partial discharge pulse, without a continuous abnormal temperature rise or thermal accumulation, determined to be a simple insulation partial discharge fault caused by insulation aging, burrs, or insulation cracks within the cabinet; Third, a continuous temperature exceeding the limit accompanied by a high-intensity stable partial discharge pulse, with the temperature rise trend increasing synchronously with the partial discharge intensity, determined to be a combined temperature-partial discharge fault. S504. Multi-level differentiated early warning output: Combining comprehensive degradation confidence and fault type, five levels are classified: normal, warning, first-level early warning, second-level early warning, and third-level emergency fault, matching different operation and maintenance handling strategies; S505. Fault source tracing data storage: The local edge terminal fully stores the original temperature of the partition, the corrected real temperature, three-dimensional spatial coordinates, partial discharge pulse sequence, load conditions, and fused confidence data of the fault period, constructing a fault dataset with spatial and type labels to prepare for subsequent source tracing; S506. Fault three-dimensional location output: When the system triggers an early warning, it synchronously retrieves the unique three-dimensional coordinates (X1, 1Y, Z1), the partition, and the fault thermal diffusion radius of the abnormal measurement point. The three-dimensional coordinates are uploaded to the remote operation and maintenance platform along with the early warning signal. Operation and maintenance personnel can directly locate the defective component by referring to the cabinet's three-dimensional installation drawings, improving maintenance efficiency.

[0037] Based on specific operating conditions, this implementation uses a general-purpose KYN28A-12 armored withdrawable 10kV switchgear, equipped with a VS1 vacuum circuit breaker with a rated current of 1250A. The test site is a 10kV distribution room of a 110kV urban substation. It fully simulates complex operating conditions such as electromagnetic interference, load fluctuations, seasonal temperature and humidity changes, and cabinet ventilation changes. Multiple control groups of traditional schemes are set up.

[0038] Example 1: The test conditions were normal temperature environment in spring and autumn, the switch cabinet load was stable at 50% of the rated current, the cabinet ventilation was normal, and the equipment had no faults or defects.

[0039] An online monitoring method for contact temperature and partial discharge in switchgear based on multi-parameter fusion is proposed. First, the switchgear is divided into four independent acquisition zones: an upper busbar compartment, a middle circuit breaker compartment, a lower cable compartment, and a secondary instrument compartment. Each zone is independently equipped with a wireless temperature measurement node and a controllable acquisition switch. A three-dimensional X / Y / Z coordinate system is established for each zone, and the spatial coordinate parameters of each temperature measurement node are recorded, completing the equipment initialization. After system startup, all zone acquisition switches are off by default and are turned on sequentially according to a preset timing sequence. The acquisition time for a single zone is 2 seconds. After acquisition is completed, the switch is immediately turned off, and the system switches to the next zone, completely avoiding cross-zone signal crosstalk.

[0040] During the data acquisition process, raw temperature data, ambient temperature and humidity, load current, and multiple partial discharge signals for each zone were collected simultaneously, and the zone number, spatial coordinates, and acquisition timestamp labels were bound to them. The raw temperature data for the upper busbar compartment was 53.5℃, the contact temperature for the middle circuit breaker compartment was 42.8℃, the temperature for the lower cable compartment was 36.2℃, and the temperature for the secondary instrument compartment was 31.5℃.

[0041] The calibration calculation is performed using a three-dimensional spatial temperature gradient calibration model. The model formula is: T real =T raw -k z ∆Zk xy S xy The measuring points in the upper busbar compartment belong to the high-level thermal accumulation zone, and the values ​​are matched with the zone parameter: k. z =0.5℃ / cm, relative height difference between measuring points and reference point ∆Z=6.4cm, cabinet ventilation is weak k xy =0.5, slight equipment obstruction S xy =0.2; Substitute into the calculation: T real=53.5-0.5*6.4-0.5*0.2=50.3℃, the actual temperature of the equipment after correction is 50.3℃, matching the normal temperature range of 25℃~55℃ for the busbar compartment. The middle circuit breaker compartment is the core current-carrying reference area, ∆Z approaches 0, the spatial gradient deviation is extremely small, and the temperature after correction remains basically unchanged at 42.8℃, which is within the normal operating range. The measuring point in the lower cable compartment is too low, matching the low-level zoning parameter: k z =-0.15℃ / cm, ∆Z=-6.0cm, good ventilation k xy =0.2, Unobstructed S xy =0; Substitute into the calculation: T real =36.2-(-0.15)*6.0-0=37.1℃, slightly compensated for heat dissipation loss, no abnormal temperature rise.

[0042] The corrected temperature data underwent outlier removal and median filtering preprocessing. A three-dimensional temperature feature vector was constructed by extracting temperature amplitude, temperature rise rate, and baseline offset. Under these conditions, the temperature rise rate was consistently below 0.3℃ / 10min, and the baseline offset was less than 3℃, showing no signs of degradation. Only background noise was collected from the multiple partial discharge signals; no effective discharge pulses were observed. The overall degradation confidence level after fusion was 0.18, indicating that the system determined the equipment was operating normally.

[0043] Example 2: This technical solution is used to verify the heat accumulation condition of high frequency false alarms in traditional technology. The test condition is a high temperature environment in summer, the cabinet is loaded with 80% of the rated current, the cabinet door is closed and the ventilation is weak, the heat accumulation in the upper part of the cabinet is serious, and the equipment body has no faults or defects.

[0044] Traditional monitoring methods detected an initial temperature of 62°C in the upper busbar compartment, triggering a Level 1 over-temperature warning and identifying it as an equipment overheating fault. This is a typical false alarm and cannot distinguish between heat accumulation in the space and temperature rise due to equipment failure.

[0045] This solution accurately identifies operating conditions through zone identification and three-dimensional correction: First, it identifies the measuring point located in the high-temperature accumulation zone of the upper busbar chamber, retrieves the specific threshold system and correction parameters for that zone, and sets k... z =0.5℃ / cm, ∆Z=9.2cm, summer closed ventilation difference k xy =0.8, partial occlusion S xy =0.3, Substitute into the correction formula: T real=62-0.5*9.2-0.8*0.3=57.2℃. After deducting the inherent heat accumulation temperature difference of 4.8℃ in the upper space, the corrected actual temperature of the equipment body is 57.2℃, which is at the lower limit of the first-level warning range of 55℃~65℃ in the busbar compartment, and has not reached the fault threshold. At the same time, combined with the comparison of the temperature field at multiple points in the zone, it is determined that the uniform temperature rise in the entire cabinet is due to heat accumulation in the space caused by poor ventilation, and not a fault in the equipment body. Only the operating data is recorded, and no fault warning is triggered, thus completely suppressing the false alarm problem of traditional technology.

[0046] At the same time, the system synchronously monitors the core contact area of ​​the central circuit breaker compartment. After correction, the temperature is 46.1℃, accurately capturing the slight temperature rise trend in the core area and marking the potential danger status in advance, achieving the monitoring effect of "no false alarms and no missed alarms".

[0047] Example 3: By fine-tuning the preload of the circuit breaker contact springs, a simple overheating fault with decreased contact pressure and increased contact resistance was simulated. The insulation inside the cabinet was intact, with no partial discharge defects. The test load was 1000A, and the ambient temperature was 35℃. The zone acquisition unit focused on collecting the core contact temperature data of the central circuit breaker compartment. The gradient parameters of the circuit breaker compartment were slightly corrected, and k was set. z =0.15℃ / cm, ∆Z=2.0cm, normal ventilation k xy =0.2, Unobstructed S xy =0, original temperature measurement T raw =58.5℃, Substitute into the formula: T real =58.5 - 0.15 * 2.0 - 0 = 58.2℃. After correction, the actual hot spot temperature of the contact continued to rise and stabilized at 58.2℃, exceeding the 55℃ fault threshold of the circuit breaker compartment; the temperature rise rate stabilized at 1.0℃ / 10min, the baseline offset continued to increase, and the three-dimensional temperature characteristics all showed a significant deterioration trend. Multiple partial discharge signals continuously lacked effective discharge pulses, and the partial discharge fusion confidence level was below 0.1.

[0048] During the two-level hierarchical fusion process, the system automatically increases the weight of temperature feature evidence, eliminates minor evidence conflicts, and ultimately outputs a comprehensive degradation confidence level of 0.69, triggering a level-two warning. Combining the characteristics of "continuously exceeding the temperature limit and no partial discharge signal," the system accurately determines that it is a contact failure due to overheating, and outputs targeted contact maintenance and repair suggestions. It also locally stores complete temperature rise curves and spatial temperature field data for fault tracing.

[0049] Example 4: To maintain good contact between the contacts, a small metal burr defect is set in the insulation partition of the cable compartment to simulate a simple insulation partial discharge fault, and the equipment temperature does not rise abnormally.

[0050] All system zone temperature measurement data are within the normal range for their respective zones. The parameter value for the cable room measurement point is: k. z=-0.15℃ / cm, ∆Z=-5.0cm, the temperature stabilized at 36.5℃ after correction, with no temperature rise degradation; the UHF, TEV and AE three partial discharge sensors continuously captured stable periodic discharge pulses. After adaptive filtering and dynamic threshold denoising, effective partial discharge features were extracted, and the first layer of homogeneous partial discharge fusion output had a high degradation confidence level.

[0051] The second-layer cross-dimensional fusion revealed a slight evidence conflict: "normal temperature, abnormal partial discharge." The system activated a local conflict evidence allocation strategy, weighting the conflict probability based on historical identification accuracy to effectively suppress oscillations in the fusion result. The final output showed a comprehensive degradation confidence level of 0.63, triggering a level-two warning. Based on the characteristics of "no abnormal temperature, consistently stable partial discharge," the system accurately determined it to be a partial discharge fault in the cabinet's insulation, prompting maintenance personnel to check the insulation components. No repair of the contact equipment was required, achieving precise maintenance.

[0052] Example 5 simultaneously reduces the preload of the contact spring and retains the burr defects of the insulating metal to simulate the most dangerous and difficult-to-identify temperature-partial discharge coupling composite fault in the field, verifying the high-order fault identification capability of this solution.

[0053] During the monitoring process, the initial temperature measurement T of the contacts in the central circuit breaker compartment was recorded. raw =58.8℃, matching core partition parameter k z =0.15℃ / cm, ∆Z=4.0cm, slight shading S xy =0.2, Substitute into the formula: T real =58.8-0.15*4.0-0.2*0.2=58.16℃. After correction, the actual temperature continues to exceed the standard, and the temperature rise trend is obvious. At the same time, the three partial discharge sensors continuously detect high-intensity effective discharge pulses, and the temperature and partial discharge characteristics deteriorate synchronously. After the two-level layered fusion, the comprehensive degradation confidence reaches 0.84, reaching the level three severe defect threshold.

[0054] The system combines the temperature field distribution of the zone, the temperature evolution trend, and the duration of the partial discharge pulse to accurately determine that it is a temperature-partial discharge symbiotic composite degradation fault, outputs the highest level emergency warning, and marks the fault coupling evolution mechanism, prompting maintenance personnel to simultaneously inspect the contact structure and the insulation system inside the cabinet.

[0055] It cannot identify complex coupled faults, and can only provide single alerts such as temperature or partial discharge anomalies. It cannot provide feedback on fault coupling relationships and evolution risks, resulting in low efficiency in operation and maintenance troubleshooting.

[0056] Results Comparison and Analysis: The experiment involved synchronous and continuous comparative testing of multiple KYN28A-12 switchgear units of the same model, setting up a control scheme to demonstrate the advantages of this technical solution. The results comparison data are shown in the table below:

[0057] Control group 1 (traditional mainstream solution): unified temperature measurement across the entire area + independent partial discharge monitoring, without spatial correction or fusion optimization; overheat warning accuracy rate of 72.1%, partial discharge identification accuracy rate of 69.5%, average of 31 false alarms per month, completely unable to identify compound faults.

[0058] Control group 2 (conventional improvement scheme): fixed coefficient temperature correction + single-layer standard DS fusion; overheat warning accuracy rate 84.6%, partial discharge identification accuracy rate 82.3%, average of 14 false alarms per month, composite fault identification success rate 47%, poor working condition adaptability and obvious fusion oscillation.

[0059] The experimental group of this solution (new partitioned three-dimensional temperature measurement + layered improved fusion) achieved an overheating hazard warning accuracy of 96.3%, a partial discharge identification accuracy of 94.7%, an average of only 9 false alarms per month, and a composite degradation fault identification success rate of 92.6%. It can adaptively adapt to seasonal temperature and humidity changes, load fluctuations, and cabinet ventilation changes, solving problems such as false and missed temperature measurements, fusion conflicts, and single fault identification.

[0060] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention can be modified and varied in various ways. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for online monitoring of switchgear contact temperature and partial discharge based on multi-parameter fusion, characterized in that, Includes the following steps: S1: Cabinet Zone Three-Dimensional Spatial Temperature Acquisition: Based on the electrical function and three-dimensional spatial structure of the switchgear, the cabinet is divided into four independent acquisition zones: busbar compartment, circuit breaker compartment, cable compartment, and secondary instrument compartment. Each zone is equipped with an independent controllable acquisition switch. A zone-based time-interlocked polling acquisition mechanism is adopted, with only one zone acquisition switch activated at any given time. The raw temperature data of the temperature measurement nodes in each zone are collected sequentially. Simultaneously, the three-dimensional spatial coordinate information of each zone, the ambient temperature and humidity inside the cabinet, the three-phase load current of the switchgear, and the raw partial discharge signals of the cabinet's UHF, TEV, and AE channels are collected. All acquired data are uniformly timestamped and zonal spatially tagged to complete the preliminary timing matching of heterogeneous parameters. S2: Three-dimensional spatial temperature feature extraction of each zone: Based on the three-dimensional spatial coordinate system and zone attributes established in step S1, match the corresponding zone-specific temperature reference threshold, Z-axis height thermal gradient coefficient k, XY plane ventilation heat transfer coefficient k and shading coefficient S, and complete the layered and step-by-step accurate correction by relying on the preset three-dimensional spatial temperature gradient correction model to eliminate the temperature measurement system error caused by cabinet hot air floating, zone ventilation differences and spatial heat accumulation, and restore the true temperature of the electrical equipment body; S3: Data preprocessing: Adaptive bandpass filtering is applied to the three raw partial discharge signals of UHF, TEV and AE. Based on the real-time noise window, the dynamic 3σ noise threshold is statistically analyzed to remove environmental interference pulses, effectively retain the effective partial discharge signal, and extract the partial discharge pulse amplitude, pulse repetition frequency and power frequency phase distribution characteristics to construct multi-channel partial discharge feature vectors. S4: Layered Data Fusion: The first layer completes the fusion of partial discharge characteristic evidence from three sources: UHF, TEV, and AE, and outputs a single-dimensional partial discharge degradation confidence level; the second layer performs cross-dimensional heterogeneous parameter fusion of partial discharge confidence evidence, three-dimensional temperature characteristic evidence after partition correction, and load current evidence; configures dynamic confidence level assignment rules based on operating condition adaptation, introduces conflict evidence local allocation strategy to resolve evidence conflicts generated by fast and slow heterogeneous parameter fusion, and outputs the comprehensive degradation confidence level of switchgear contacts; S5: Graded Early Warning: Retrieves the dedicated fault threshold library for each zone, and combines the comprehensive degradation confidence, the spatiotemporal evolution trend of zone temperature, and the continuous stability of partial discharge pulse to distinguish between overheating faults caused by increased contact resistance, partial discharge faults in the internal insulation of the switchgear, and temperature-partial discharge symbiotic composite degradation faults. Outputs multi-level early warning signals based on the severity level of the defect, and locally stores monitoring data and spatial temperature field data for fault querying.

2. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, The zone-based time-interlocked polling acquisition mechanism described in step S1 is as follows: the system initializes all zone acquisition switches to the off state, and polls and turns on each zone acquisition switch one by one according to the fixed timing sequence of bus room → circuit breaker room → cable room → secondary instrument room. After the acquisition of a single zone is completed, the corresponding switch is immediately turned off, and then the acquisition switch of the next zone is turned on, so as to realize the electrical isolation and wireless signal isolation of the zone acquisition and prevent cross-zone data crosstalk and timing conflicts.

3. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, In step S1, a dedicated local three-dimensional spatial coordinate system is built for each independent acquisition zone to achieve accurate modeling and differential correction of the spatial dimension of temperature measurement data. Specifically, this includes: constructing a local rectangular coordinate system with the core current-carrying components of each zone's equipment as the reference origin; defining the X-axis as the left-right longitudinal installation direction of the cabinet to characterize the differences in equipment arrangement and obstruction; defining the Y-axis as the front-to-back air intake and exhaust direction of the cabinet to characterize the differences in ventilation, convection, and heat transfer; and defining the Z-axis as the vertical height gradient direction of the cabinet to characterize the vertical temperature gradient of rising hot air, which is the core gradient dimension of the cabinet's temperature field. Each temperature measurement node in each zone is bound to a unique (X1, Y1, Z1) three-dimensional spatial coordinate parameter, and the reference coordinates (X, Y, Z) of each zone are preset as the temperature correction reference point for that zone, realizing one-to-one traceability and matching of each set of temperature data, spatial location, zone attributes, and environmental conditions, providing accurate spatial parameter support for the three-dimensional spatial gradient correction in step S2.

4. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, The specific temperature ranges for each zone in step S2 are as follows: Upper busbar compartment: normal temperature range 25℃~55℃, first-level warning range 55℃~65℃, second-level fault range >65℃; Middle circuit breaker compartment: normal temperature range 25℃~45℃, first-level warning range 45℃~55℃, second-level fault range >55℃; Lower cable compartment: normal temperature range 20℃~40℃, first-level warning range 40℃~48℃, second-level fault range >48℃; Secondary instrument compartment: normal temperature range 20℃~35℃. Temperatures exceeding these ranges are considered abnormal temperature rises.

5. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, The temperature data preprocessing in step S2 specifically includes: setting the effective physical temperature measurement range [-20℃, 150℃], removing outliers from wireless communication and replacing them with adjacent effective data interpolation; using a sliding median filtering algorithm with a window length of N=5 to suppress temperature jumps caused by instantaneous electromagnetic pulses and output a stable and reliable filtered temperature sequence.

6. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, In step S3, the dynamic noise threshold satisfies V th =V avg +3σ, select a 20s real-time noise sliding window, statistically analyze the mean signal amplitude Vavg and standard deviation σ within the window, adaptively remove environmental interference pulses under different operating conditions, and retain the effective partial discharge pulse sequence.

7. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, The dynamic confidence assignment rule in step S4 is as follows: when the temperature rise rate of the core area of ​​the circuit breaker compartment is greater than 1.5℃ / 10min, the weight of the three-dimensional temperature feature evidence is increased. When a valid partial discharge pulse is detected stably for 5 consecutive minutes, the weight of the partial discharge evidence is increased; when the load current reaches more than 90% of the rated current, the weight of the load parameter is increased; the conflict evidence local allocation strategy is as follows: when the temperature evidence and the partial discharge evidence judgment results conflict, the traditional global equal allocation method is abandoned, and the conflict probability is weighted according to the historical fault identification accuracy of the two types of evidence, which effectively suppresses the oscillation of the fusion result.

8. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, The coupling fault identification logic in step S5 is as follows: if the actual contact temperature continues to rise after correction and there is no effective partial discharge pulse, it is determined to be a contact failure and overheating fault; if effective partial discharge pulses are continuously detected and there is no continuous abnormal temperature rise, it is determined to be a partial discharge fault in the cabinet insulation; if the temperature continues to exceed the standard and is accompanied by a stable partial discharge pulse, it is determined to be a temperature-partial discharge symbiotic composite degradation fault; at the same time, combined with the overall distribution of the zoned temperature field, it distinguishes between single-point equipment faults and global space heat accumulation.

9. The online monitoring method for switchgear contact temperature and partial discharge based on multi-parameter fusion according to claim 1, characterized in that, All data acquisition, temperature correction, feature extraction, multi-parameter fusion, and fault identification processes are deployed on the local edge terminal of the switchgear. Only early warning information, fault segment data, and abnormal spatial temperature field data are uploaded to the remote platform, while the original acquired data is stored locally, enabling real-time on-site monitoring and early warning.

Images