Power distribution cabinet wire fault detection and identification method

Abstract

The invention relates to the technical field of power distribution. The invention discloses a power distribution cabinet wire fault detection and identification method, which is characterized in that a plurality of wires with different specifications are arranged in a power distribution cabinet, and the method comprises the following steps: obtaining real-time operation parameter data of the plurality of wires, and determining a fault wire based on a reference parameter set of normal work of a to-be-detected wire and a deviation characteristic value relation of real-time parameters; fault parameter data of the faulty electric wire are input into a multi-layer fault identification model, the model comprises a plurality of sub-identification units connected in sequence, each unit detects different types of faults, and finally a fault mode is obtained. According to the method, at least three parameters are acquired through distributed sensing, and data reliability is guaranteed through noise reduction and normalization processing. On the basis of a double-threshold criterion and dynamic weight calculation, faulty electric wires are accurately distinguished, and misjudgment and missed judgment are completely eradicated. The multi-layer model carries out progressive detection, an emergency fault is preferentially recognized, and the sub-recognition units are matched with an exclusive algorithm to improve the accuracy.

Description

Technical Field

[0001] This invention relates to the field of power distribution technology, and in particular to a method for detecting and identifying electrical wire faults in a power distribution cabinet. Background Technology

[0002] As a crucial link in power transmission, the safe operation of the wiring in distribution cabinets directly impacts the overall stability of the power supply. Traditional wiring fault detection relies heavily on manual inspections, using methods such as infrared thermography and multimeter measurements to identify potential problems. This approach is not only time-consuming and labor-intensive but also limited by the experience of maintenance personnel, making it difficult to detect early-stage, hidden faults. Manual inspection cannot achieve real-time monitoring, often resulting in reactive responses after a fault occurs, missing the optimal time for intervention. Furthermore, the compact internal structure and dense wiring of distribution cabinets create blind spots during manual inspections, making it easy to miss hidden faults such as poor contact or aging insulation, thus creating potential safety hazards for the power distribution system.

[0003] Current automated detection solutions suffer from limitations such as relying on single parameter acquisition methods and weak anti-interference capabilities. They depend solely on a single operating parameter for fault diagnosis, failing to comprehensively reflect the working status of electrical wires. Furthermore, they are susceptible to electromagnetic interference and environmental noise within the distribution cabinet, leading to data distortion and misjudgments. In fault identification, existing models often employ a single algorithm to handle various faults, lacking specificity and failing to accurately distinguish between different types of faults, particularly struggling to quickly pinpoint emergency faults such as short-circuit precursors and overloads. Moreover, most solutions lack precise spatial positioning, requiring a step-by-step inspection of each wire after a fault occurs, extending repair time and failing to predict fault development trends, thus hindering proactive risk prevention.

[0004] Existing testing solutions are mostly designed for wires of specific voltage levels and cross-sectional specifications, resulting in a narrow range of applicability and difficulty in meeting the testing needs of different types of distribution cabinets. For wires of different materials such as copper core, aluminum core, and alloy conductors, a differentiated benchmark parameter system has not been established, leading to testing standards that do not align with actual operating conditions. Furthermore, existing solutions do not consider performance degradation due to wire aging; benchmark parameters remain fixed over a long period, and testing accuracy gradually decreases with increasing service life. In addition, some solutions are complex to operate and have high maintenance thresholds, lacking targeted emergency handling suggestions for alarm information, hindering rapid on-site response and failing to meet the practical needs of various power distribution scenarios. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention discloses a method for detecting and identifying electrical cable faults in power distribution cabinets, which achieves more accurate fault detection, more efficient fault identification, more proactive fault prevention and control, and wider adaptability to various scenarios.

[0006] This invention discloses a method for detecting and identifying electrical wiring faults in a power distribution cabinet, comprising:

[0007] The distribution cabinet contains multiple wires of different specifications, and the methods include:

[0008] Obtain real-time operating parameter data for multiple power lines;

[0009] Based on the relationship between the set of reference parameters of the wire under normal operating conditions and the deviation characteristic values ​​corresponding to the real-time operating parameter data of the wire under test, the faulty wire is identified.

[0010] The fault parameter data of the faulty wire is input into the multi-layer fault identification model to obtain the fault mode of the distribution cabinet wire.

[0011] The multi-layer fault identification model includes multiple sequentially connected sub-identification units, each of which is used to detect different types of fault modes in the wire.

[0012] Furthermore, real-time operating parameter data of multiple power lines are acquired. The real-time operating parameter data includes, but is not limited to, at least three of the following: conductor temperature, real-time current carrying capacity, conductor voltage drop, partial discharge pulse amplitude, and leakage current. This data is collected by distributed sensing units deployed in the distribution cabinet, with the distributed sensing units spaced apart at key nodes and heat-prone sections of the power lines.

[0013] Furthermore, the method for obtaining the set of reference parameters for the wire under normal operating conditions is as follows:

[0014] Based on the rated cross-sectional specifications, conductor material, insulation type, and rated operating environment parameters of the distribution cabinet, combined with the historical normal operation data of this type of wire, statistical regression analysis was conducted to obtain the parameters, and independent subsets of benchmark parameters were established for copper core wires, aluminum core wires, and alloy conductor wires.

[0015] Furthermore, based on the relationship between the set of reference parameters under normal operating conditions of the wire under test and the deviation characteristic values ​​corresponding to the real-time operating parameter data of the wire under test, the faulty wire is determined, including: calculating the single-parameter deviation coefficient of each real-time operating parameter relative to the set of reference parameters, and the comprehensive deviation characteristic value of all parameters.

[0016] Based on the dual threshold criterion combining static threshold and dynamic floating threshold, the allowable deviation range for each parameter is obtained from the benchmark parameter set;

[0017] The faulty wire is determined by the relationship between the comprehensive deviation characteristic value and the allowable deviation range.

[0018] Furthermore, based on the relationship between the comprehensive deviation characteristic value and the allowable deviation range, the faulty wire is identified, including: when the comprehensive deviation characteristic value is within the allowable deviation range and the deviation coefficient of each single parameter does not exceed the safety threshold of its respective parameter, the wire under test is identified as a normal wire;

[0019] When the comprehensive deviation characteristic value exceeds the allowable deviation range, or when the deviation coefficient of any single parameter exceeds the corresponding safety threshold and the duration reaches the preset duration, the wire under test is determined to be a faulty wire.

[0020] Furthermore, the first sub-identification unit in the multi-layer fault identification model is used to preferentially receive fault parameter data of the faulty wire;

[0021] When the fault parameter data meets the identification conditions of the first sub-identification unit, the fault mode of the power distribution cabinet wire is determined to be the fault type corresponding to the first sub-identification unit. The fault type corresponding to the first sub-identification unit is an emergency fault such as a short circuit precursor or an overload fault.

[0022] When the fault parameter data does not meet the identification conditions of the first sub-identification unit, the fault parameter data is sequentially input into the subsequent sub-identification units. The subsequent sub-identification units correspond to non-emergency fault types such as insulation layer aging, poor contact, excessive partial discharge, and abnormal voltage drop.

[0023] Furthermore, the sub-identification units employ random forest algorithm, logistic regression model, or gradient boosting tree algorithm. Each sub-identification unit is trained using sample data corresponding to the fault type, and the identification threshold of each sub-identification unit can be dynamically adjusted according to the actual operating load of the distribution cabinet.

[0024] Furthermore, after acquiring real-time operating parameter data for multiple power lines, the process also includes:

[0025] Wavelet transform noise reduction is performed on real-time operating parameter data to remove noise data caused by electromagnetic interference in the power distribution cabinet, environmental clutter and measurement equipment errors, while retaining effective parameter characteristics.

[0026] At the same time, the parameter data after noise reduction is normalized to unify the data units, which facilitates the subsequent calculation of deviation feature values ​​and the processing of fault identification models.

[0027] Furthermore, after inputting the fault parameter data of the faulty wire into the multi-layer fault identification model to obtain the fault mode of the distribution cabinet wire, the model also includes:

[0028] A fault development trend model is constructed by performing polynomial fitting on historical and real-time fault parameter data of faulty wires within a preset time period.

[0029] Based on the slope and curvature of the fitted curve of the model, the fault deterioration rate of the faulty wire is predicted, the early warning time window for fault escalation is output, and key risk points are marked.

[0030] Furthermore, by combining the internal spatial layout data of the distribution cabinet and the wiring path information, the identified faulty wires are spatially located.

[0031] Simultaneously, based on the fault mode and predicted deterioration trend, a graded alarm mechanism is triggered. The alarm information includes the faulty wire number, specific fault type, fault location, expected deterioration time, and emergency handling suggestions. It can also be adapted to different rated voltage levels and different cross-sectional specifications of distribution cabinet wire detection scenarios.

[0032] The beneficial effects of this invention are:

[0033] This invention utilizes distributed sensing units to collect at least three parameters of electrical wires, including conductor temperature and real-time current carrying capacity. Combined with wavelet transform denoising and normalization, it effectively eliminates electromagnetic interference and equipment errors, ensuring data accuracy and reliability. Based on static and dynamic dual-threshold criteria, and combined with single-parameter deviation coefficients and dynamic weights, it distinguishes between normal and faulty wires, avoiding false positives and false negatives. A multi-layer fault identification model progressively detects faults according to risk level, prioritizing short-circuit precursors and overloads to buy time for emergency response. Each sub-identification unit is adapted to different fault characteristics, using algorithms such as random forests and gradient boosting trees to match fault types and improve identification accuracy.

[0034] The solution incorporates a built-in parameter library, flexibly matching distribution cabinet wires of different rated voltage levels and cross-sectional specifications, and is compatible with various materials such as copper core, aluminum core, and alloy conductors. A tiered alarm mechanism integrates fault number, type, estimated deterioration time, and emergency handling suggestions, intuitively conveying key information and lowering the maintenance threshold. The aging compensation and dynamic optimization mechanism for baseline parameters ensures that the testing standards closely match the actual condition of the wires during long-term use, eliminating the need for frequent adjustments. The overall process is highly automated, easy to operate, shortens the testing and maintenance cycle, reduces labor and time costs, and is suitable for fault prevention needs in various power distribution scenarios. Attached Figure Description

[0035] Figure 1 This is a flowchart of a method for detecting and identifying electrical wire faults in a power distribution cabinet according to an embodiment of this application. Detailed Implementation

[0036] To enable those skilled in the art to better understand the present invention, the technical solutions in the specific embodiments of the present invention will be clearly and completely described below.

[0037] This invention discloses a method for detecting and identifying electrical wire faults in a power distribution cabinet. The power distribution cabinet contains multiple wires of different specifications. The method includes:

[0038] Obtain real-time operating parameter data for multiple power lines;

[0039] Based on the relationship between the set of reference parameters of the wire under normal operating conditions and the deviation characteristic values ​​corresponding to the real-time operating parameter data of the wire under test, the faulty wire is identified.

[0040] The fault parameter data of the faulty wire is input into the multi-layer fault identification model to obtain the fault mode of the distribution cabinet wire.

[0041] The multi-layer fault identification model includes multiple sequentially connected sub-identification units, each of which is used to detect different types of fault modes in the wire.

[0042] First, sensors deployed within the distribution cabinet collect real-time operating parameter data for multiple power lines of different specifications, ensuring data coverage of the operating status of each power line. Next, a set of baseline parameters for the power line under test under normal operating conditions is retrieved, and the deviation characteristic value between the real-time operating parameter data and this baseline set is calculated. By analyzing the matching relationship between the two, faulty power lines exhibiting abnormalities are identified. Finally, the fault parameter data of the faulty power lines is input into a multi-layer fault identification model. This model consists of multiple sub-identification units connected sequentially as serial data streams. The fault parameter data is first input into the first sub-identification unit; if a match is found, the result is output and the process terminates; if not, the data is automatically passed to the second sub-identification unit, and so on. Each sub-identification unit corresponds to a different type of fault detection function. After the data is processed sequentially by each sub-identification unit, the specific fault mode of the power distribution cabinet's power lines is output.

[0043] Real-time operational parameter data collection covers all wires of different specifications within the distribution cabinet, ensuring comprehensive coverage. A baseline parameter set serves as a standard reference for the normal operating state of the wire under test, while deviation characteristic values ​​directly reflect the degree of deviation between real-time parameters and the standard state. Correlation analysis between the two allows for the identification of faulty wires. The sub-identification units of the multi-layer fault identification model are connected in a preset order, with each sub-identification unit focusing on detecting a specific type of fault. Data is transferred systematically between units, gradually clarifying the specific type of fault.

[0044] It can systematically complete the entire process from data acquisition to fault location and pattern recognition, with clear steps and tight connections. Based on the comparison of benchmark parameters and deviation feature values, it can quickly distinguish between normal and faulty wires, avoiding misjudgment. The multi-layer fault recognition model, through the division of labor and cooperation of sub-recognition units, can specifically detect different types of fault modes, adapting to the detection needs of multiple wires of different specifications in the distribution cabinet, improving the targeting and effectiveness of fault recognition.

[0045] Due to differences in specifications such as cross-sectional area and conductor material, the key monitoring points of the multiple wires within the distribution cabinet differ. Distributed sensing units are arranged at intervals along the wire laying path: temperature and current sensors are deployed at key nodes such as terminals and cable branches to capture real-time changes in conductor temperature and current carrying capacity fluctuations, as these locations are prone to localized overheating due to contact resistance; insulation resistance detectors and partial discharge sensors are added in areas prone to heat generation, such as densely laid sections and bends, to continuously collect partial discharge pulse amplitudes and prevent accelerated insulation aging due to poor heat dissipation; for long-distance wire laying, voltage drop monitoring modules are installed every 1.5-2 meters along the line to synchronously record conductor voltage drop data. All sensing units are connected to a data acquisition terminal via wired or wireless means, uploading parameters at a frequency of 50-100ms / time to ensure real-time reflection of the dynamic operating status of each wire.

[0046] The deviation characteristic value is calculated with reference to the baseline parameter set of the wire under test: First, the single-parameter deviation coefficient is calculated, which is the ratio of (real-time operating parameter value - baseline parameter value) to the baseline parameter value. The larger the absolute value, the more significant the deviation of the parameter from the normal state. Then, a comprehensive deviation characteristic value is obtained through weighted summation. The dynamic adjustment mechanism of the weights is set based on the current load type of the distribution cabinet, not the fault type. The load type of the distribution cabinet, such as lighting load, power load, and special equipment load, is determined during installation. The system will preset corresponding weight parameters according to the load type of the distribution cabinet. For example, the insulation parameter weight of the lighting load distribution cabinet is set to 0.45, the temperature / current carrying capacity weight of the power load distribution cabinet is set to 0.45, and the voltage drop weight of the special equipment load distribution cabinet is set to 0.4. During operation, the system automatically calls the preset weights according to the real-time monitored load type, and optimizes and adjusts the weight parameters every six months through historical fault data analysis. This setting avoids the logical loop between fault type and weight adjustment, ensuring the accuracy of weight adjustment and the stability of system operation. Simultaneously, a dual-threshold criterion is used to determine the allowable deviation range: the static threshold is set based on the parameter fluctuation range during the historical normal operation of similar power lines, while the dynamic threshold is dynamically adjusted according to the real-time load of the distribution cabinet. The static threshold is initially used as the basic allowable range, and then the threshold boundary is dynamically adjusted based on the real-time load rate. The threshold is narrowed when the load rate exceeds 80% and widened when it is below 50%, avoiding overlap and thus improving the adaptability and accuracy of deviation judgment. The dynamic floating threshold is calculated based on the ratio of the real-time load to the rated load of the distribution cabinet. When the real-time load ≥ the rated load, the dynamic threshold = static threshold × (1 - 0.05 × (real-time load / rated load - 1)); when the real-time load < the rated load, the dynamic threshold remains the same as the static threshold. This calculation is updated in real-time every 500ms in the embedded controller of the distribution cabinet to ensure that the threshold dynamically adapts to load changes.

[0047] As one implementation method, real-time operating parameter data of multiple power lines is acquired. This real-time operating parameter data includes, but is not limited to, at least three of the following: conductor temperature, real-time current carrying capacity, conductor voltage drop, partial discharge pulse amplitude, and leakage current. Data is collected by distributed sensing units deployed within the distribution cabinet, with these units spaced at critical nodes and heat-prone sections of the power lines. Sensor accuracy: temperature ±0.5℃, current carrying capacity ±0.2%FS, partial discharge ≤1pC, voltage drop ±0.1%FS; all conforming to GB / T13850, calibrated before shipment, and the report is retained.

[0048] First, define the scope of real-time operating parameter data collection, selecting at least three parameters from: conductor temperature, real-time current carrying capacity, conductor voltage drop, partial discharge pulse amplitude, and leakage current. Partial discharge pulse amplitude is only applicable to 10kV and above high-voltage distribution cabinets; this monitoring is not performed in low-voltage distribution cabinets. Next, deploy distributed sensing units within the distribution cabinet, installing them at intervals at key nodes such as terminals and cable branches, as well as in areas prone to heat generation, such as densely packed wires and bends. Finally, through these distributed sensing units, simultaneously collect selected real-time operating parameter data from multiple wires of different specifications within the distribution cabinet, ensuring data collection covers high-risk monitoring areas of each wire.

[0049] The selection of acquisition parameters focuses on key dimensions of safe operation of power lines. Each parameter reflects a specific operating state of the power line, and a combination of at least three parameters avoids the limitations of monitoring a single parameter. The deployment of distributed sensing units is highly targeted, with key nodes being areas prone to abnormal contact resistance and sections prone to overheating being high-risk fault areas. This layout captures the most valuable operational data. The acquisition process simultaneously covers all power lines of different specifications within the distribution cabinet, ensuring comprehensive monitoring and providing data support for subsequent fault diagnosis.

[0050] With a rich variety of parameters that directly address key issues, it can present the operating status of power lines from multiple dimensions, avoiding missed fault detections due to a single monitoring dimension. Distributed sensing units are strategically placed in high-risk areas, ensuring highly targeted and accurate data collection and reducing interference from invalid data. It supports flexible selection of at least three parameter combinations to adapt to the monitoring needs of different power line specifications. Furthermore, its simple deployment method allows for rapid adaptation to the internal structures of various distribution cabinets, enhancing the versatility and practicality of the detection method.

[0051] Sensing units at critical nodes are installed close to the node, such as at terminals, cable branch points, and busbar connections, within a range of 0.1-0.3 meters. This ensures the detection of parameter anomalies caused by changes in node contact resistance. In areas prone to heat generation, including densely laid wiring areas, bends, and areas near heat sources, sensing units are evenly spaced at 0.5-1 meter intervals to monitor parameter fluctuations caused by poor heat dissipation. For ordinary wiring sections without special risks, the spacing between sensing units can be increased to 1.5-2 meters, ensuring monitoring coverage while avoiding resource waste. The overall installation spacing can be flexibly adjusted according to the internal space layout of the distribution cabinet and the wire specifications.

[0052] Parameter acquisition is prioritized according to the urgency of its impact on the safe operation of power lines. The first priority is conductor temperature, real-time current carrying capacity, and partial discharge pulse amplitude. These three parameters are directly related to emergency faults such as overload, short-circuit precursors, and insulation breakdown, and are monitoring indicators to ensure the safe operation of power lines. The real-time accuracy of these parameters must be prioritized. The second priority is leakage current, which reflects the health of the insulation layer and is related to the long-term stable operation of the power line. This parameter is acquired synchronously after the initial parameter acquisition. The third priority is conductor voltage drop. This parameter can help identify problems such as poor contact and abnormal line losses. As a supplementary monitoring indicator, it works in conjunction with the first two types of parameters to improve the data system and ensure that no fault is missed in the detection process.

[0053] To accommodate the varying characteristics of different parameters, a tiered acquisition frequency is employed: Emergency parameters are acquired at 1ms / time, using a high-frequency current-type partial discharge sensor with a measurement range of 0.1pC-1000pC and an error ≤1pC, ensuring pulse signal capture; Other parameters are acquired at 10ms / time, using a PT1000 platinum resistance temperature sensor with a measurement range of -20℃-200℃ and an error of ±0.5℃, and a Hall effect current sensor with a measurement range of 0-1000A and an error ≤0.2%FS; Auxiliary parameters are acquired at 50ms / time, using a voltage drop sensor with a measurement range of 0-100V and an error of ±0.1%FS. The acquisition frequency can be dynamically adjusted according to the actual operating load of the distribution cabinet. When the load exceeds 80% of the rated value, the acquisition frequency for all parameters is increased by 50%, ensuring that parameter changes under high load are captured.

[0054] As one implementation method, the set of reference parameters for the wire under normal operating conditions is obtained by: based on the wire's rated cross-sectional specifications, conductor material, insulation layer type, and the rated operating environment parameters of the distribution cabinet, combined with the historical normal operating data of this type of wire, through statistical regression analysis, and establishing independent subsets of reference parameters for copper core wires, aluminum core wires, and alloy conductor wires respectively.

[0055] First, the properties and parameters of the wire under test are analyzed to clarify its rated cross-sectional area, conductor material, and insulation type. Simultaneously, the rated operating environment parameters of the distribution cabinet are collected. Next, historical normal operating data of this type of wire with the same properties as the wire under test are retrieved to ensure that the data covers stable operating conditions under different workloads and environmental conditions. Statistical regression analysis is used to model and calculate the above properties and parameters against the historical normal operating data, generating a set of benchmark parameters for the normal operating conditions of the wire under test. Finally, independent subsets of benchmark parameters are established for copper core wires, aluminum core wires, and alloy conductor wires to ensure that each type of wire has its own dedicated normal operating condition reference standard.

[0056] The rated cross-sectional area and conductor material of electrical wires directly determine their conductivity, heat dissipation, and insulation performance. The rated operating environment parameters of the distribution cabinet affect the external conditions for wire operation. These are all influencing factors in the set of benchmark parameters. Historical normal operation data provides accurate and effective data support for the benchmark parameters, while statistical regression analysis can uncover the inherent correlations between data points, making the benchmark parameters more closely reflect the actual operating patterns of electrical wires. The physical properties of wires with different conductor materials vary significantly; establishing separate subsets of benchmark parameters can avoid benchmark deviations caused by material differences and ensure the reference accuracy of each subset.

[0057] The generation of the benchmark parameter set comprehensively considers the inherent properties of the wires and external environmental factors, combining historical data with scientific analysis methods to make the benchmark standards more objective and targeted. Independent subsets are established according to conductor material classification to accommodate the differences in characteristics of different wire materials, solving the benchmark reference problem for mixed-material distribution cabinets. The benchmark parameter set provides a reliable basis for subsequent deviation characteristic value calculations, reducing the misjudgment rate of fault diagnosis and improving the scientific rigor of the overall detection method.

[0058] The subset of benchmark parameters is not fixed; it is periodically updated with new historical normal operating data for the wire material, and dynamic optimization is achieved by updating the statistical regression analysis model. Benchmark parameters are supplemented with aging compensation: annual aging coefficient 0.99 for copper core, 0.98 for aluminum core, and 0.975 for alloy. The calibrated benchmark = original benchmark × (aging coefficient^ years used). Calibration is performed annually, and the years are reset after fault repair. The optimization cycle can be adjusted according to the fluctuations in the distribution cabinet's operating load. The optimization cycle is shortened when load fluctuations are large to ensure that the benchmark parameters remain synchronized with the actual operating status of the wire, further improving the timeliness and accuracy of the benchmark reference.

[0059] As one implementation method, the faulty wire is determined based on the relationship between the set of reference parameters under normal operating conditions of the wire under test and the deviation characteristic values ​​corresponding to the real-time operating parameter data of the wire under test. This includes: calculating the single-parameter deviation coefficient of each real-time operating parameter relative to the set of reference parameters, and the comprehensive deviation characteristic value of all parameters.

[0060] Based on the dual threshold criterion combining static threshold and dynamic floating threshold, the allowable deviation range for each parameter is obtained from the benchmark parameter set;

[0061] The faulty wire is determined by the relationship between the comprehensive deviation characteristic value and the allowable deviation range.

[0062] First, for each power line under test, the single-parameter deviation coefficient of each real-time operating parameter relative to its reference parameter set is calculated. The calculation method is to divide the difference between the real-time parameter value and the reference parameter value by the reference parameter value, and then obtain the comprehensive deviation characteristic value of all parameters through weighted summation. The weights are set according to the degree of influence of the parameters on power line safety. Next, a dual-threshold criterion combining static and dynamic floating thresholds is used. The static threshold is taken from the parameter fluctuation range of similar power lines during historical normal operation, while the dynamic floating threshold is adjusted according to the real-time operating load of the distribution cabinet. The combination of these two criteria determines the allowable deviation range for each parameter. Finally, the comprehensive deviation characteristic value is compared with the allowable deviation range. If the comprehensive deviation characteristic value exceeds the allowable range, the power line is determined to be faulty.

[0063] The dynamic floating threshold adjustment mechanism can be specifically based on the ratio of the total current of the distribution cabinet to the rated current. The system has a preset baseline threshold. When the real-time load rate increases to 80%-100%, the dynamic threshold is narrowed by 10%-15% based on the baseline value; when the load rate is below 50%, the threshold is widened by 5%-10% accordingly. This adjustment can be achieved through a preset lookup table, without the need for complex real-time calculations, thus ensuring the threshold's adaptability to environmental changes while also taking into account the processing capabilities of the embedded system.

[0064] Single-parameter deviation coefficients intuitively reflect the degree of deviation of a single operating indicator, while comprehensive deviation characteristic values ​​integrate multi-dimensional parameter information, avoiding the one-sidedness of judgment based on a single parameter. The dual-threshold criterion balances stability and flexibility; the static threshold ensures a consistent basic judgment standard, while the dynamic floating threshold adapts to the operational differences of the distribution cabinet under different loads, making the allowable deviation range more closely match actual working conditions. By directly comparing the comprehensive deviation characteristic value with the allowable deviation range, a clear basis for fault determination is established, ensuring that the faulty wire screening process is systematic and the results are accurate.

[0065] The calculation combining single-parameter and comprehensive deviation characteristic values ​​reflects both the anomalies of individual parameters and the overall operating status, reducing the risk of missed detections. Compared to a single threshold, the dual-threshold criterion better handles complex situations such as distribution cabinet load fluctuations and environmental changes, improving the adaptability of fault diagnosis. The judgment logic is clear and quantifiable, avoiding subjective judgment errors, while the operation process is simple, enabling rapid fault screening of multiple wires and improving detection efficiency.

[0066] The calculation of the single-parameter deviation coefficient strictly references the unique set of benchmark parameters for each wire, ensuring that the calculation benchmarks for wires of different specifications and materials do not interfere with each other. For parameters such as conductor temperature and real-time current carrying capacity, the calculation results are retained to three decimal places to ensure numerical accuracy and provide a reliable basis for subsequent calculation of comprehensive deviation characteristic values ​​and fault diagnosis.

[0067] The static threshold is determined by statistically analyzing the normal operating parameters of similar power lines over three consecutive months, taking the maximum value of the fluctuation range as the upper limit and the minimum value as the lower limit. The dynamic floating threshold is adjusted based on the ratio of the current total operating load of the distribution cabinet to the rated load. For every 10% increase in the load ratio, the dynamic threshold is narrowed by 5% from the previous level, ensuring stricter fault judgment standards under high load conditions and avoiding misjudgments under low load conditions.

[0068] As one implementation method, the faulty wire is determined based on the relationship between the comprehensive deviation characteristic value and the allowable deviation range, including: when the comprehensive deviation characteristic value is within the allowable deviation range and the deviation coefficient of each single parameter does not exceed the safety threshold of its respective parameter, the wire under test is determined to be a normal wire;

[0069] When the comprehensive deviation characteristic value exceeds the allowable deviation range, or when the deviation coefficient of any single parameter exceeds the corresponding safety threshold and the duration reaches the preset duration, the wire under test is determined to be a faulty wire.

[0070] First, two criteria are established for judgment: one is the matching between the comprehensive deviation characteristic value and the allowable deviation range; the other is the comparison result and duration of each individual parameter deviation coefficient with its corresponding safety threshold. For each wire under test, first check whether its comprehensive deviation characteristic value is within the allowable deviation range, and simultaneously confirm whether each individual parameter deviation coefficient does not exceed its respective safety threshold. If both conditions are met, the wire is judged to be a normal wire. If the comprehensive deviation characteristic value exceeds the allowable deviation range, it is directly judged as a faulty wire; if any individual parameter deviation coefficient exceeds the corresponding safety threshold, the duration of this abnormal state must be recorded. When the duration reaches the preset duration, it is also judged as a faulty wire.

[0071] The determination of normal power lines adopts a dual compliance principle: it requires that the overall operating status (comprehensive deviation characteristic value) meets the standard, and it also ensures that each individual operating indicator (single parameter deviation coefficient) is normal, avoiding the omission of hidden risks due to a single-dimensional judgment. The determination of faulty power lines uses any trigger principle: a comprehensive deviation exceeding the range indicates a serious abnormality in overall operation, while a single parameter continuously exceeding the threshold represents a persistent risk to a key indicator; both situations meet the fault determination criteria. The preset duration is used to filter out misjudgments caused by instantaneous parameter fluctuations, ensuring that faults are only calibrated for truly existing persistent anomalies, closely reflecting the dynamic characteristics of actual power line operation.

[0072] The judgment logic considers both the overall state and individual indicators, avoiding the overlooking of serious anomalies in a single key parameter due to normal overall indicators, and preventing misjudgments based on instantaneous fluctuations in a single parameter, thus improving judgment accuracy. The introduction of preset durations adapts to the parameter fluctuation characteristics during power line operation, effectively eliminating interference factors and reducing the false judgment rate. Two fault judgment conditions cover different types of abnormal scenarios, comprehensively capturing both overall instability and localized severe anomalies, ensuring no faulty power lines are missed, and providing a basis for subsequent processing.

[0073] The preset durations are set differently based on parameter priority. For first-priority parameters such as conductor temperature, real-time current carrying capacity, and partial discharge pulse amplitude, the preset duration is 3-5 seconds; these parameters require rapid response to abnormalities. For second-priority parameters such as leakage current, the preset duration is 10-15 seconds, balancing response speed and anti-interference requirements. For third-priority parameters such as conductor voltage drop, the preset duration is 20-30 seconds to avoid misjudgments caused by instantaneous fluctuations in non-critical parameters. All preset durations can be fine-tuned based on the actual operational stability of the distribution cabinet.

[0074] Safety thresholds are set based on a set of baseline parameters for each wire. The safety threshold for conductor temperature is 1.3 times the baseline parameter value, the safety threshold for real-time current carrying capacity is 1.2 times the baseline parameter value, and the safety threshold for partial discharge pulse amplitude is 1.5 times the baseline parameter value. The safety threshold for leakage current is 0.8 times the baseline parameter value, and the safety threshold for conductor voltage drop is 1.4 times the baseline parameter value. The threshold settings fully consider the degree of influence of the parameters on the safe operation of the wires, making the parameter thresholds more stringent.

[0075] In one implementation, the first sub-identification unit in the multi-layer fault identification model is used to preferentially receive fault parameter data of the faulty wire;

[0076] When the fault parameter data meets the identification conditions of the first sub-identification unit, the fault mode of the power distribution cabinet wire is determined to be the fault type corresponding to the first sub-identification unit. The fault type corresponding to the first sub-identification unit is an emergency fault such as a short circuit precursor or an overload fault.

[0077] When the fault parameter data does not meet the identification conditions of the first sub-identification unit, the fault parameter data is sequentially input into the subsequent sub-identification units. The subsequent sub-identification units correspond to non-emergency fault types such as insulation layer aging, poor contact, excessive partial discharge, and abnormal voltage drop.

[0078] In the multi-layer fault identification model, the first sub-identification unit is designated as the priority node for data reception, and the fault parameter data of the faulty wire is first transmitted to this unit. The first sub-identification unit has built-in identification conditions for emergency faults such as short-circuit precursors and overload faults. It judges the fault by comparing the matching degree between the fault parameter data and these conditions: if the data meets the identification conditions, the corresponding emergency fault type is directly output as the final fault mode; if the data does not meet the conditions, the model automatically transmits the fault parameter data to subsequent sub-identification units in a preset order. Subsequent sub-identification units are each preset with identification logic for non-emergency faults such as insulation aging, poor contact, excessive partial discharge, and abnormal voltage drop. The data is sequentially detected by each unit until a corresponding fault type is matched and the result is output.

[0079] To prevent misjudgments under multiple fault conditions, each sub-identification unit employs a priority-based decision principle. Fault parameter data flows through each unit from highest to lowest urgency, and the process terminates once identified by a unit. In rare cases where parameter features simultaneously match multiple fault types, the model outputs the fault type with the highest priority and marks it in the unit's alarm information as potentially accompanied by other faults, prompting maintenance personnel to conduct a comprehensive investigation. Furthermore, the system records secondary fault features that also reach thresholds in other sub-identification units during this identification process, forming an auxiliary diagnostic log for reference during in-depth analysis. This design ensures rapid response to urgent faults while avoiding confusion in diagnostic results.

[0080] The model employs a priority-based identification mechanism, placing the first sub-identification unit corresponding to emergency faults at the forefront to ensure that faults potentially causing serious consequences are detected first, thus shortening emergency response time. Sub-identification units for emergency and non-emergency faults are sequentially linked according to risk level, with data only flowing into subsequent units when the identification conditions of the preceding unit are not met, avoiding interference between different fault types. Each sub-identification unit focuses on feature matching for a specific fault type, achieving fault mode localization through clear division of responsibilities, and closely aligning with the characteristic differences of various wire faults.

[0081] The design prioritizes handling emergency faults, enabling rapid identification of high-risk types in the early stages of a fault. This buys time for timely power outages, repairs, and other emergency measures, reducing the risk of the accident escalating. Sub-identification units are specialized by type, avoiding the accuracy degradation caused by a single model identifying multiple faults and improving the accuracy of identifying various fault modes. A progressive data transmission logic adapts to the risk level of fault types, ensuring both rapid response to emergency faults and comprehensive coverage of non-emergency fault identification needs, enhancing the model's adaptability to complex fault scenarios.

[0082] The identification criteria for short-circuit precursors include a sudden increase in real-time current carrying capacity to more than twice the baseline value and a conductor temperature rise exceeding 20°C within 5 seconds; the identification criteria for overload faults are a real-time current carrying capacity exceeding 1.5 times the baseline value for more than 10 seconds and a conductor temperature simultaneously exceeding the safety threshold. These criteria are set based on typical parameter characteristics of emergency faults to ensure rapid triggering of judgments.

[0083] As one implementation method, the sub-identification units employ random forest, logistic regression, or gradient boosting tree algorithms. Each sub-identification unit is trained using sample data corresponding to its respective fault type, and the identification threshold for each sub-identification unit can be dynamically adjusted based on the actual operating load of the distribution cabinet. The training data originates from a historical fault case database, laboratory simulated fault data, and field test data. Data augmentation techniques are used to expand the sample for rare fault types.

[0084] The sub-identification units are constructed using either the Random Forest algorithm, Logistic Regression model, or Gradient Boosting Tree algorithm. For the first sub-identification unit, which handles emergency faults such as short-circuit precursors and overload faults, the Random Forest algorithm is selected to handle multi-parameter abrupt changes. For subsequent sub-identification units handling non-emergency faults such as poor contact and excessive partial discharge, either the Logistic Regression model or the Gradient Boosting Tree algorithm is chosen based on the complexity of the fault characteristics. The training process for each sub-identification unit is based on sample data corresponding to the fault type. The sample data includes parameters such as conductor temperature, current carrying capacity, and partial discharge amplitude at the time of the fault, as well as fault labels. Model parameters are optimized through iterative training until the identification accuracy meets the target. Simultaneously, the system monitors the actual operating load of the distribution cabinet in real time. When the load rises above 80% of the rated load, the identification threshold of each sub-identification unit automatically narrows by 10%-15%; when the load is below 50%, the threshold is widened by 5%-10%, achieving dynamic adjustment.

[0085] All selected algorithms possess the ability to handle multi-parameter inputs and capture feature correlations, adapting to the complexity of power line fault parameters. Random forests exhibit strong anti-overfitting capabilities, making them suitable for identifying abrupt changes in emergency fault characteristics. Logistic regression models are simple and efficient, adapting to fault types with linear feature correlations. Gradient boosting trees offer high accuracy and can uncover deep relationships between nonlinear features. Training with sample data corresponding to specific fault types ensures that sub-identification units specialize in specific faults, avoiding cross-type interference. The identification threshold is dynamically adjusted according to the actual operating load because load changes alter the fluctuation range of normal power line parameters. Threshold adaptation ensures consistency in fault identification under different loads, reducing misjudgments caused by environmental interference.

[0086] Diverse algorithms adapt to different fault characteristics, improving the recognition accuracy and specificity of each sub-identification unit and avoiding the limitations of a single algorithm. Training based on corresponding fault sample data makes the model more sensitive to the characteristics of target faults, resulting in higher recognition efficiency. The recognition threshold is dynamically adjusted according to the operating load, allowing the model to adapt to the load fluctuations of the distribution cabinet, strictly controlling performance under high loads and reducing false alarms under low loads, thus enhancing the environmental adaptability and reliability of fault identification.

[0087] Emergency fault parameters exhibit strong abrupt changes and multi-dimensional correlations. Random forests, through ensemble learning of multiple decision trees, can effectively filter noise and enhance abrupt features, making them the preferred choice. For contact failures, the parameters show an approximately linear relationship with the fault severity; logistic regression models can quickly establish a linear mapping, making them suitable for this type of fault. Fault parameters such as insulation aging and excessive partial discharge exhibit complex features and significant non-linear correlations. Gradient boosting trees, through iterative optimization of weak classifiers, can fit non-linear relationships, making them the preferred choice.

[0088] The threshold adjustment is based on the ratio of the real-time total current to the rated current of the distribution cabinet: when the ratio is between 50% and 80%, the threshold remains at the baseline value; when the ratio is greater than 80%, the threshold is adjusted by narrowing it by 3% for every 5% increase in load, with a maximum narrowing of 15%; when the ratio is less than 50%, the threshold is adjusted by widening it by 2% for every 10% decrease in load, with a maximum widening of 10%. The adjusted threshold is synchronized to the corresponding sub-identification unit in real time to ensure real-time matching with the current load status.

[0089] As one implementation method, after acquiring real-time operating parameter data for multiple power lines, noise reduction and normalization preprocessing are required. Due to strong electromagnetic interference within the distribution cabinet, noise in the original parameter data significantly affects the accuracy of deviation characteristic value calculation. Simultaneously, the significant differences in dimensions between parameters mean that without normalization, parameters with larger numerical values ​​will dominate the comprehensive calculation, potentially masking the abnormal characteristics of other parameters. This solution sets the preprocessing time to within 100ms to ensure no impact on system real-time performance. It also includes: performing wavelet transform noise reduction on the real-time operating parameter data to remove noise data caused by electromagnetic interference within the distribution cabinet, environmental clutter, and measurement equipment errors, while retaining effective parameter characteristics.

[0090] At the same time, the parameter data after noise reduction is normalized to unify the data units, which facilitates the subsequent calculation of deviation feature values ​​and the processing of fault identification models.

[0091] After acquiring real-time operating parameter data for multiple power lines, wavelet transform noise reduction processing is first performed on the data. A wavelet basis function adapted to the signal characteristics of the distribution cabinet parameters is selected to decompose the original data into signal components of different frequencies. High-frequency noise components generated by electromagnetic interference, environmental clutter, and measurement equipment errors are separated and removed, retaining the effective parameter characteristics reflecting the true operating state of the power lines. Simultaneously, the noise-reduced parameter data undergoes normalization processing. A unified numerical scaling method is used to map parameters with different dimensions, such as conductor temperature and current carrying capacity, to the same standard numerical range, completing data preprocessing to adapt to subsequent operations.

[0092] Wavelet transform has the ability to distinguish between valid signals and noise, removing interference without destroying the characteristics of valid parameters, thus ensuring data authenticity. Real-time operating parameters come from different monitoring dimensions with significant differences in units; directly using them for calculations or model inputs would lead to weight imbalances. Normalization, by unifying the numerical range, ensures that each parameter has equal influence in subsequent deviation feature value calculations and fault identification models, guaranteeing fairness in the processing logic. These two processing steps are sequentially linked: first, data quality is purified, then the data format is standardized, laying a reliable foundation for subsequent processes.

[0093] Noise reduction effectively filters out various interference factors, reduces the influence of invalid data on subsequent judgments, and improves the purity and reliability of parameter data. Normalization eliminates the dimensional differences between different parameters, avoids calculation deviations caused by large differences in numerical ranges, ensures the accuracy of deviation feature value calculation results, and reduces the training difficulty and computational load of the fault identification model. The preprocessing process is standardized to adapt to parameter data of all different specifications of wires, enhancing the stability and consistency of the overall detection method.

[0094] The db4 wavelet basis function was selected as the transform to decompose the original parameter data into 3-5 levels of signal components. The noise threshold was set to the mean of the high-frequency components of each decomposition level plus 3 times the standard deviation. High-frequency components below this threshold were identified as noise and set to zero. The signal was then reconstructed through inverse wavelet transform to obtain the denoised effective data.

[0095] A min-max normalization method is used to uniformly map all denoised parameter data to the 0-1 range. The calculation method is that the target data equals the original data minus the historical minimum value of the parameter, and then divided by the difference between the historical maximum and historical minimum values ​​of the parameter. The historical extreme values ​​of all parameters are taken from the reference parameter set of the corresponding wire type and historical normal operation data to ensure that the scaling standard fits the actual working conditions.

[0096] As one implementation method, after inputting the fault parameter data of the faulty wire into the multi-layer fault identification model to obtain the fault mode of the distribution cabinet wire, the method further includes: performing polynomial fitting on the historical fault parameter data and real-time fault parameter data of the faulty wire within a preset time period to construct a fault development trend model.

[0097] Based on the slope and curvature of the fitted curve of the model, the fault deterioration rate of the faulty wire is predicted, the early warning time window for fault escalation is output, and key risk points are marked.

[0098] After obtaining the fault modes of the distribution cabinet's wiring, a preset time range is defined, and historical and real-time fault parameter data of the faulty wiring within this range are collected to ensure data coverage of key change stages after the fault occurs. The collected fault parameter data is then subjected to polynomial fitting, and the fitting effect is optimized by adjusting the polynomial order to construct a fault development trend model that reflects the changing patterns of fault parameters over time. Based on the fitted curve output by this model, the slope and curvature of the curve are calculated: the slope reflects the rate of change of the fault parameters, and the curvature reflects the increasing or decreasing trend of the rate of change. Combining these two factors quantifies the rate of fault deterioration, thereby calculating the early warning time window for fault escalation. Simultaneously, key risk points are marked based on abnormal parameter sections and key monitoring points.

[0099] The pre-set fault data set includes historical and real-time information, comprehensively presenting the entire process of a fault from its emergence to its development, providing ample data support for trend analysis. Polynomial fitting captures the nonlinear variation of fault parameters, better reflecting the actual characteristics of fault development compared to linear fitting. The slope of the fitted curve directly corresponds to the rate of fault deterioration, while the curvature can predict whether the deterioration rate will accelerate; combining both allows for precise quantification of the fault's development trend. The output of early warning time windows and key risk points is based on data pattern derivation rather than subjective judgment, ensuring the objectivity and reliability of the prediction results.

[0100] To prevent overfitting of noisy data by polynomial fitting when constructing the fault development trend model, the system employs an adaptive order selection mechanism. The fitting process starts with a second-order model, and by calculating the goodness of fit and observing the residuals, it gradually increases to a third or fourth-order model only when the data change pattern becomes significantly nonlinear, typically not exceeding a fifth-order model. Simultaneously, the system applies a moving average filter to the input fault parameter data to smooth out instantaneous fluctuations. Regarding computational resource allocation, for faults identified as urgent, the system will invoke a simplified fitting model, such as a linear or second-order fit, sacrificing some accuracy for prediction speed. This ensures that a timely warning window can still be output even when the fault rapidly deteriorates, achieving a balance between safety and efficiency.

[0101] Breaking away from the limitations of traditional fault detection methods that only identify the current state and cannot predict the future, this system uses trend models to proactively predict fault escalation, allowing ample preparation time for maintenance. Early warning time windows clearly identify critical points of fault escalation, helping staff rationally prioritize maintenance and avoid reactive fault responses. Marking key risk points pinpoints the affected areas, reducing troubleshooting time and workload, improving fault handling efficiency, and simultaneously lowering the probability of fault escalation leading to serious accidents.

[0102] The preset duration is set differently based on the fault mode: the preset duration for emergency faults is 30-60 minutes, focusing on the rapid deterioration process; the preset duration for non-emergency faults is 24-72 hours, covering the slow development cycle. The polynomial order is selected according to the degree of data fluctuation; a third-order polynomial is used when the parameter fluctuation is mild, and a fifth-order polynomial is used when the fluctuation is severe, to ensure the fit of the fitted curve to the actual data.

[0103] The warning time window is graded according to the rate of fault deterioration: when the absolute value of the slope is greater than a preset threshold and the curvature is positive, it is determined to be an emergency warning window, lasting 1-3 hours; when the absolute value of the slope is in the medium range, it is determined to be a regular warning window, lasting 4-8 hours; when the absolute value of the slope is small and the curvature tends to be stable, it is determined to be a delayed warning window, lasting 12-24 hours. Key risk points are prioritized for marking monitoring points with the largest exceedances of fault parameters, locations prone to exacerbating faults such as wire joints and branches, and sections near distribution cabinet components. An absolute value of slope > 0.5 (parameter unit / hour) indicates rapid deterioration, 0.2-0.5 indicates medium speed, and ≤ 0.2 indicates slow speed.

[0104] As one implementation method, the method further includes: spatially locating the identified faulty wire by combining the internal spatial layout data of the distribution cabinet and the wiring path information.

[0105] Simultaneously, based on the fault mode and predicted deterioration trend, a graded alarm mechanism is triggered. The alarm information includes the faulty wire number, specific fault type, fault location, expected deterioration time, and emergency handling suggestions. It can also be adapted to different rated voltage levels and different cross-sectional specifications of distribution cabinet wire detection scenarios.

[0106] First, data on the internal spatial layout of the distribution cabinet is collected to clarify the installation location of each component, the distribution of compartments, and the internal passageway. Simultaneously, the laying path information of each wire is analyzed to determine its specific route, fixed points, and relative positional relationship with other components. The location coordinates of the sensor unit corresponding to the identified faulty wire are then correlated and matched with the aforementioned spatial layout data and laying path information. A spatial coordinate conversion algorithm is used to calculate the specific location of the fault, ensuring high positioning accuracy. Next, based on the urgency of the fault mode and the predicted fault escalation trend, different alarm levels are classified, triggering a tiered alarm mechanism. Alarm information is integrated into a fixed structure, including the faulty wire number, specific fault type, fault location section, estimated escalation time, and emergency handling suggestions. The system has a built-in adaptation parameter library for wires of different rated voltage levels and cross-sectional specifications, automatically matching the current detection scenario to ensure that the alarm information is compatible with the detection object. Emergency handling suggestions are derived from the system's embedded expert knowledge base. This knowledge base uses fault type, fault severity, faulty wire specifications, and their location within the distribution cabinet as a joint index. Once a fault is identified and located, the system automatically matches the most suitable pre-processing suggestions from the knowledge base. For example, for an overload fault, the system may suggest checking for abnormalities in the end load and considering switching to a backup circuit; for insulation aging, it may suggest paying close attention to ambient humidity and planning a power outage for inspection. The knowledge base can be updated and maintained by operations and maintenance personnel according to the organization's actual safety procedures.

[0107] Spatial positioning establishes the correspondence between the location of the sensing unit and the physical space of the distribution cabinet. Through coordinate conversion, abstract data is transformed into specific spatial locations, enabling visualized fault location. The tiered alarm mechanism is based on the fault risk level: urgent and rapidly deteriorating faults correspond to high-level alarms, while non-urgent and slowly developing faults correspond to low-level alarms, ensuring that resources are prioritized for high-risk issues. Alarm information integration revolves around the fault handling needs of staff, including key decision-making information. The design of the adaptive parameter library addresses the detection differences of wires of different specifications and voltage levels, achieving universal application through automatic matching.

[0108] The accuracy of spatial location of faulty wires depends primarily on the deployment density of sensor units and the selection of positioning technology. With sensor units typically deployed at 1-2 meter intervals, segment-level positioning can be achieved through signal strength indication or topology analysis, pinpointing the specific wire and approximate segment where the fault occurred. This level of accuracy significantly reduces the investigation area and meets on-site maintenance needs. For more precise positioning, high-precision positioning technologies such as UWB must be integrated, and the sensor unit density increased. This approach is suitable for scenarios with specific requirements.

[0109] The sensor unit's three-dimensional spatial coordinates are recorded synchronously during installation. A multi-source data fusion algorithm is used to calibrate the deviation between the spatial layout data and the laying path information. Combining the sensor unit number corresponding to the faulty wire and the node position during data acquisition, the relative distance between the fault point and the sensor unit is calculated in reverse, ultimately determining the coordinates of the fault point. Simultaneously, a schematic diagram of the fault location is displayed on the system interface.

[0110] The alarm levels are categorized as follows: Level 1: Emergency fault / deterioration ≤ 3 hours, audible and visual alarm, 1-hour response; Level 2: Partial discharge / 3-8 hours, audible alarm + yellow light, 4-hour response; Level 3: Aging / ≥ 8 hours, green light, 24-hour response. Alarm levels are divided into three levels: Level 1 alarms correspond to emergency faults such as short-circuit precursors and overloads, or faults with an expected deterioration time of 1-3 hours, triggering an audible and visual alarm and pushing alarm information to the mobile phone of the maintenance manager; Level 2 alarms correspond to faults such as excessive partial discharge and poor contact, or faults with an expected deterioration time of 4-8 hours, triggering an audible alarm and highlighting it on the system interface; Level 3 alarms correspond to faults such as insulation aging and abnormal voltage drop, or faults with an expected deterioration time of 12-24 hours, displaying alarm information only on the system interface. The response time limits for different alarm levels are set to 1 hour, 4 hours, and 24 hours respectively to ensure orderly fault handling.

[0111] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for detecting and identifying electrical wire faults in an electrical distribution cabinet, the method comprising: The method comprises the following steps: The power distribution cabinet is provided with a plurality of wires of different specifications, and the method comprises the following steps: Obtain real-time operating parameter data of the plurality of wires; Determine the faulty wire based on the relationship between the deviation characteristic value corresponding to the real-time operating parameter data of the wire to be tested and the reference parameter set of the wire to be tested under normal working conditions; Input the fault parameter data of the faulty wire into a multi-layer fault identification model to obtain the fault mode of the wire of the power distribution cabinet; The multi-layer fault identification model comprises a plurality of sub-identification units connected in sequence, and each sub-identification unit is used to detect different types of fault modes of the wire.

2. The power distribution cabinet wire fault detection and identification method according to claim 1, wherein: The real-time operating parameter data of the plurality of wires includes at least three of wire conductor temperature, real-time carrying capacity, wire voltage drop, partial discharge pulse amplitude, and leakage current, and is collected by a distributed sensing unit arranged in the power distribution cabinet, which is arranged at key nodes and heat prone sections of the wire.

3. The power distribution cabinet wire fault detection and identification method according to claim 1, wherein: The reference parameter set of the wire to be tested under normal working conditions is obtained in the following manner: Based on the rated cross-sectional specification, conductor material, insulation layer type of the wire, and the rated working environment parameters of the power distribution cabinet, combined with the historical normal operation data of the wire, the reference parameter set is obtained by statistical regression analysis, and independent reference parameter subsets are established for copper core wires, aluminum core wires, and alloy conductor wires.

4. The power distribution cabinet wire fault detection and identification method according to claim 1, wherein: The relationship between the deviation characteristic value corresponding to the real-time operating parameter data of the wire to be tested and the reference parameter set of the wire to be tested under normal working conditions is used to determine the faulty wire, which comprises: calculating the single parameter deviation coefficient of each real-time operating parameter relative to the reference parameter set, and the comprehensive deviation characteristic value of all parameters; Based on the double-threshold criterion combining the static threshold value and the dynamic floating threshold value, the deviation allowed range corresponding to each parameter is obtained from the reference parameter set; The relationship between the comprehensive deviation characteristic value and the deviation allowed range is used to determine the faulty wire.

5. The power distribution cabinet wire fault detection and identification method according to claim 4, wherein: Based on the relationship between the comprehensive deviation characteristic value and the deviation allowed range, the faulty wire is determined, which comprises: when the comprehensive deviation characteristic value is within the deviation allowed range and the single parameter deviation coefficient does not exceed the safety threshold value of the corresponding parameter, the wire to be tested is determined to be a normal wire; When the comprehensive deviation characteristic value exceeds the deviation allowed range, or any single parameter deviation coefficient exceeds the corresponding safety threshold value and the duration reaches the preset time length, the wire to be tested is determined to be a faulty wire.

6. The power distribution cabinet wire fault detection and identification method according to claim 1, wherein: The first sub-identification unit in the multi-layer fault identification model is used to preferentially receive the fault parameter data of the faulty wire. When the fault parameter data meets the identification condition of the first sub-identification unit, the fault mode of the power distribution cabinet wire is determined as the fault type corresponding to the first sub-identification unit, and the fault type corresponding to the first sub-identification unit is a short-circuit precursor or an emergency fault such as overload fault; When the fault parameter data does not meet the identification condition of the first sub-identification unit, the fault parameter data is sequentially input into subsequent sub-identification units, and the subsequent sub-identification units correspond to non-emergency fault types such as insulation layer aging, poor contact, excessive partial discharge, and abnormal voltage drop.

7. The power distribution cabinet wire fault detection and identification method according to claim 6, characterized in that: The sub-identification units use random forest algorithm, logistic regression model or gradient boosting tree algorithm, each sub-identification unit is trained by sample data corresponding to the fault type, and the identification threshold of each sub-identification unit can be dynamically adjusted according to the actual operating load of the power distribution cabinet.

8. The power distribution cabinet wire fault detection and identification method according to claim 1, characterized in that: After obtaining the real-time operating parameter data of the plurality of wires, the method further comprises: Performing wavelet transform noise reduction processing on the real-time operating parameter data to remove noise data caused by electromagnetic interference, environmental clutter and measurement equipment errors in the power distribution cabinet, and retaining effective parameter features; Meanwhile, the parameter data after noise reduction is normalized to unify the data dimension, which is convenient for subsequent deviation eigenvalue calculation and fault identification model processing.

9. The power distribution cabinet wire fault detection and identification method according to claim 1, characterized in that: After inputting the fault parameter data of the fault wire into the multi-layer fault identification model to obtain the fault mode of the power distribution cabinet wire, the method further comprises: Performing polynomial fitting on the historical fault parameter data and real-time fault parameter data of the fault wire within a preset time period to construct a fault development trend model; Based on the slope and curvature of the fitting curve of the model, the fault deterioration speed of the fault wire is predicted, the early warning time window of fault escalation is output, and the key risk points are labeled.

10. The power distribution cabinet wire fault detection and identification method according to claim 1, characterized in that: Combined with the internal space layout data of the power distribution cabinet and the laying path information of the wire, the determined fault wire is spatially positioned; Meanwhile, according to the fault mode and the predicted deterioration trend, a hierarchical alarm mechanism is triggered, the alarm information includes the fault wire number, the specific fault type, the fault section, the predicted deterioration time and the emergency treatment suggestion, and can adapt to different rated voltage levels, different cross-section specifications of power distribution cabinet wire detection scenarios.

Images