A fault detection method for a distribution box

Abstract

The application relates to the technical field of power equipment fault detection, in particular to a fault detection method for a distribution box, and aims to solve the problem that, although a traditional method can realize basic fault type identification by adopting a multi-sensor fusion technology combined with dynamic weight distribution, the method cannot position a fault to a specific circuit or element level. The application effectively screens key parameters through multi-dimensional data acquisition and dynamic weight calculation, guarantees data integrity through a hierarchical transmission protocol, establishes accurate mapping of fault positioning through circuit-element correlation analysis, and continuously optimizes a detection process through a closed-loop feedback mechanism. The method refines the fault positioning level from the circuit level to the element level, significantly reduces the troubleshooting range during equipment maintenance, and improves the operation and maintenance efficiency of an industrial power distribution system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment fault detection, in particular to a fault detection method of a distribution box. BACKGROUND

[0002] In the actual industrial application of distribution box fault detection, such as industrial plant, outdoor power distribution scene, etc., the existing technology has the core contradiction between the insufficient reliability of multi-parameter data transmission in complex network environment and the demand for precise fault positioning. Specifically, on the one hand, although the existing technology realizes the transmission of multi-parameter data and the classification of fault types through multi-sensor collection, dynamic weight calculation, NB-IoT transmission and CRC check, etc., such as being able to identify over-temperature, humidity abnormality and other fault types, this judgment only stays at the fault type level and cannot be precisely positioned to the specific fault circuit or component, which leads to the need for maintenance personnel to spend a lot of time to check one by one, seriously affecting the fault handling efficiency.

[0003] On the other hand, although some technical solutions try to realize the precise positioning of fault circuit through circuit-component correlation matching, influence correlation analysis and power variable testing, etc., these methods are difficult to guarantee the reliability of data transmission in complex network environment (low bandwidth, high electromagnetic interference), and the collected circuit state and component state data are easy to be distorted due to packet loss, interference and other problems in the transmission process, which further leads to serious deviation of the subsequent correlation analysis and positioning results, and the false positive rate is high in actual application.

[0004] This technical defect makes the distribution box fault detection system difficult to play its due role in actual application, which not only increases the maintenance cost, but also affects the stable operation of the power system. In view of the above problems, the existing technology needs to be improved. SUMMARY

[0005] (I) Technical problems to be solved

[0006] To solve the above problems, the present application proposes a fault detection method of a distribution box, which aims to solve the problem that the traditional method adopts multi-sensor fusion technology combined with dynamic weight distribution, which can realize basic fault type identification, but cannot locate the fault to the specific circuit or component level.

[0007] (II) Technical solutions

[0008] The application discloses a fault detection method of a distribution box, and has the technical scheme as follows: a fault detection method of a distribution box, comprising: collecting environmental operation parameters and element state parameters through a plurality of sensors arranged in the distribution box, wherein the environmental operation parameters include temperature, humidity, sulfur hexafluoride gas concentration, current and voltage, and the element state parameters include contact resistance of a contactor, capacitance of a capacitor group and insulation resistance of a line; performing dynamic weight calculation on the collected original sensor data, wherein the dynamic weight includes stability weight based on standard deviation of parameter historical data and correlation weight based on correlation coefficient of parameters and circuits or elements, and the calculated dynamic weight is used to generate a weighted data set; encapsulating the weighted data set through an NB-IoT network protocol, wherein the data packet contains device identification, a time stamp, the weighted data set and a CRC check code, and is transmitted to a remote server; performing CRC check on the data received by the server, and performing circuit-element correlation analysis based on the weighted data set after the check is passed, identifying abnormal circuits and sorting them according to influence weight; performing power variable test on the abnormal circuits with high sorting order, and realizing element-level fault positioning in combination with the element state parameters; and feeding back the fault positioning result to the dynamic weight calculation process, and using the result to optimize weight distribution in subsequent data collection.

[0009] Further, the application further provides that the calculation manner of the dynamic weight is the product of the stability weight and the correlation weight, wherein the stability weight is calculated based on the historical standard deviation of the sensor data in a preset time period, and the smaller the standard deviation is, the higher the weight is; and the correlation weight is calculated based on the Pearson correlation coefficient, and the maximum value in the correlation coefficient of the parameter and each circuit or element is taken.

[0010] Further, the application further provides that the hierarchical protocol structure used in the data transmission step comprises: the physical layer adopts a modulation mode conforming to the NB-IoT standard; the network layer uses the NB-IoT protocol; the transmission layer uses the UDP protocol; and the application layer uses the CoAP protocol; the retransmission mechanism is triggered when the server check fails, and the maximum retransmission number is three.

[0011] Further, the application further provides that the circuit-element correlation analysis comprises: calculating the Pearson correlation coefficient of the element state parameters and the running circuit based on the weighted data set, and establishing a correlation relationship if the correlation coefficient exceeds a set threshold value, wherein the set threshold value is determined based on historical fault data statistics of the same type of distribution box and is not less than 0.7; analyzing the mutual influence relationship between the circuits based on historical current and voltage data to form an influence correlation table; comparing the weighted data with a preset normal parameter range to identify abnormal circuits and sort them according to their influence weight.

[0012] Further, the application also proposes that the normal parameter range is based on the weighted data statistics during the fault-free operation of the equipment, the mean plus 3 times the standard deviation is used as the initial range, and is updated regularly to adapt to the parameter drift caused by the aging of the element.

[0013] Further, the application also proposes that the power variable test includes: preferentially adjusting the environmental operation parameter with the highest weight in the weighted data; monitoring the parameter change of the related circuit after adjustment in real time, and if the parameter is stable in the normal range for 10 seconds continuously, the element is removed from the abnormal list; in combination with the circuit-element correlation, the weighted parameter of the corresponding element is detected, and if it is out of the normal range, the element is determined to be faulty.

[0014] Further, the application also proposes that the weight distribution optimization includes: if the fault positioning is accurate, the correlation weight of the related sensor parameter is increased; if a certain circuit is excluded as a non-faulty circuit for a preset number of times, the correlation weight of the related parameter is reduced; the upper and lower limits of the correlation weight are set, wherein the upper limit is not more than 0.9, and the lower limit is not less than 0.1, and the weight adjustment effect is evaluated regularly.

[0015] Further, the application also proposes an electronic device including a processor and a memory, the memory storing a computer program, and the computer program is executed by the processor to realize the fault detection method of the distribution box.

[0016] Further, the application also proposes a computer readable storage medium, which stores a computer program, and the computer program is executed by the processor to realize the fault detection method of the distribution box.

[0017] (Three) beneficial effects

[0018] Compared with the prior art, the beneficial effects of the application are:

[0019] In the application, through multi-sensor data acquisition, dynamic weight calculation, NB-IoT protocol transmission and feedback optimization mechanism, combined with circuit-element correlation analysis and power variable test, element-level fault positioning is realized while ensuring data transmission reliability, which has the advantages of realizing accurate positioning of faulty elements and reliable data transmission in complex network environment.

[0020] In the application, key parameters are effectively screened through multi-dimensional data acquisition and dynamic weight calculation, layered transmission protocol guarantees data integrity, circuit-element correlation analysis establishes accurate mapping for fault positioning, and closed-loop feedback mechanism continuously optimizes the detection process. The method refines the fault positioning level from the circuit level to the element level, significantly reduces the troubleshooting range during equipment maintenance, and improves the operation and maintenance efficiency of the industrial power distribution system. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to make the technical solutions of the embodiments of the present application or the prior art clearer, the accompanying drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the accompanying drawings in the following description only represent some of the embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0022] Figure 1 The schematic diagram in Example 1. DETAILED DESCRIPTION

[0023] The technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments only represent some of the embodiments of the present application, rather than all the embodiments. The components of the present application described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. Meanwhile, in the description of the present application, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0024] The power distribution box fault detection field has long faced technical challenges in complex industrial environments. There is low-bandwidth network coverage and high-intensity electromagnetic interference in industrial plant and outdoor power distribution scenarios, which leads to packet loss and distortion phenomena in the sensor data transmission process. Traditional methods use multi-sensor fusion technology combined with dynamic weight distribution, which can achieve basic fault type identification, but cannot locate the fault to the specific circuit or component level. The existing solutions have a split between data collection and transmission links, and the parameter weight distribution does not consider the fault positioning requirements, which leads to maintenance personnel still needing to manually investigate suspected fault areas, seriously affecting detection efficiency.

[0025] To solve the above problems, through field testing, it is found that data transmission distortion and improper weight distribution are the core factors restricting positioning accuracy. In view of the data integrity requirement in complex network environment, a hierarchical verification mechanism is tried to be introduced into the transmission protocol. At the same time, it is observed that different parameters have different contribution degrees to fault location, for example, although the contact resistance parameter exists normal fluctuation, it has strong correlation with circuit on-off fault. Therefore, the technical idea is formed: through dynamic weight calculation to screen key parameters, combined with reliable transmission to guarantee data authenticity, a bidirectional association model of circuit and component is constructed, and finally a closed-loop optimization mechanism is formed to improve positioning accuracy.

[0026] Therefore as Figure 1 shown, the application proposes a fault detection method of distribution box, including the following method steps:

[0027] S100, collecting environmental operation parameters and component state parameters through multiple types of sensors deployed in the distribution box, the environmental operation parameters including temperature, humidity, sulfur hexafluoride gas concentration, current and voltage, and the component state parameters including contact resistance of contactor, capacitance of capacitor group and insulation resistance of line;

[0028] S200, performing dynamic weight calculation on the collected original sensor data, the dynamic weight including stability weight based on parameter historical data standard deviation and correlation weight based on correlation coefficient between parameter and circuit or component, and the calculated dynamic weight is used to generate a weighted data set;

[0029] S300, packaging the weighted data set through NB-IoT network protocol, the data packet containing device identification, time stamp, weighted data set and CRC check code, and transmitting to a remote server;

[0030] S400, the server receives the data and performs CRC check, and after the check is passed, performs circuit-component association analysis based on the weighted data set, identifies abnormal circuits and sorts them according to influence weight;

[0031] S500, performing power variable test on the abnormal circuits ranked at the top, and realizing component-level fault location combined with component state parameters;

[0032] S600, feeding back the fault location result to the dynamic weight calculation process for optimizing weight distribution in subsequent data collection.

[0033] Among them, dynamic weight calculation refers to determining data importance through dual evaluation of parameter stability and correlation, which can be realized by combining historical standard deviation and Pearson correlation coefficient, the standard deviation reflecting parameter fluctuation degree, and the correlation coefficient representing the association strength between parameter and fault.

[0034] NB-IoT network protocol encapsulation refers to the use of narrowband IoT communication standards for data transmission. Specifically, a layered protocol architecture can be used to achieve physical layer anti-interference and transport layer real-time performance assurance. CRC checksum refers to Cyclic Redundancy Check encoding, which can be generated using the CRC-32 algorithm to verify the integrity of transmitted data packets.

[0035] Circuit-component correlation analysis refers to establishing a mapping relationship between operating parameters and equipment components. Specifically, it can be achieved by calculating the correlation coefficient between weighted parameters and circuit status to identify abnormal circuits.

[0036] Specifically, multiple sensors synchronously collect operating status data of the distribution box. Environmental parameters reflect the external operating conditions of the equipment, while component parameters characterize the performance of internal components. During the dynamic weight calculation process, the standard deviation of historical data for each parameter is calculated; the smaller the standard deviation, the higher the stability weight. At the same time, the Pearson correlation coefficient between the parameter and each circuit component is calculated, and the maximum value is taken as the correlation weight.

[0037] The two weights are multiplied to obtain a comprehensive weight, which is used to generate a weighted dataset. The data encapsulation adopts the layered structure of the NB-IoT protocol, with OFDM modulation used at the physical layer to enhance anti-interference capabilities and UDP protocol used at the transport layer to ensure real-time transmission. After the server performs CRC verification, it analyzes the correlation strength between each parameter and circuit components based on the weighted dataset. When the correlation coefficient exceeds a set threshold, an abnormal circuit is identified. Power variable tests are performed on abnormal circuits with high impact weights. By adjusting operating parameters and observing changes in component status, the location of the faulty component is ultimately determined. The location results are fed back to the weight calculation module, dynamically adjusting the weight allocation strategy for relevant parameters.

[0038] Traditional methods, which use only a single weighting index, may lead to the omission of key parameters. This solution ensures that important parameters are given priority by using a dual evaluation mechanism of stability and correlation.

[0039] Existing transmission schemes fail to consider data distortion caused by industrial environmental characteristics. This scheme's layered protocol design effectively balances transmission reliability and real-time requirements. Conventional fault location methods lack feedback optimization mechanisms; this scheme's closed-loop weight adjustment significantly improves long-term detection accuracy.

[0040] This application achieves precise fault location in distribution boxes under complex industrial environments. Multi-dimensional data acquisition and dynamic weight calculation effectively filter key parameters, a layered transmission protocol ensures data integrity, circuit-component correlation analysis establishes a precise mapping for fault location, and a closed-loop feedback mechanism continuously optimizes the detection process. This method refines the fault location hierarchy from the circuit level to the component level, significantly reducing the scope of troubleshooting during equipment maintenance and improving the operation and maintenance efficiency of industrial power distribution systems.

[0041] The application further proposes a calculation method of the dynamic weight as a product of a stability weight and a correlation weight, wherein the stability weight is calculated based on a historical standard deviation of sensor data in a preset time period, and the smaller the standard deviation, the higher the weight; and the correlation weight is calculated based on a Pearson correlation coefficient, and the maximum value in the correlation coefficient of each circuit or element is taken.

[0042] The stability weight refers to an index for evaluating the reliability of the historical data fluctuation degree, and can be realized by setting a differentiated time period corresponding to the load difference of the distribution box, for example, 10 minutes is set as the calculation period for the distribution box with small load fluctuation, and 5 minutes is shortened for the distribution box with frequent load change, and the unstable data in the start-stop stage of the equipment is removed before calculation. The design ensures that the weight calculation can dynamically adapt to parameter changes in different scenarios, avoiding the problem of data lag caused by fixed time period.

[0043] The correlation weight refers to an index for measuring the degree of association between the parameter and the circuit or element fault, and can be realized by importing the default coefficient of the new equipment and the later calibration mechanism, for example, the preset correlation coefficient of the same type of equipment is imported for the new distribution box, and the correlation coefficient is recalibrated after the equipment runs for 72 hours. The mechanism solves the problem of weight failure caused by the lack of historical data in the initial stage of the new equipment, and ensures the accuracy of the association analysis in the initial stage.

[0044] Specifically, the dynamic weight calculation process first selects a preset time period according to the load type of the distribution box, calculates the standard deviation of each parameter after removing the start-stop stage data, and the smaller the standard deviation, the higher the stability weight. Then, the correlation between the parameter and each circuit or element is calculated by Pearson coefficient, and the maximum value is selected as the correlation weight. Finally, the stability weight and the correlation weight are multiplied to generate a comprehensive weight value. For newly installed equipment, the preset correlation coefficient is used in the initial stage for calculation, and the calibration process is triggered by simulating the fault after accumulating enough running data, and the correlation coefficient is updated. This double weight mechanism not only eliminates the interference of high fluctuation parameters, but also strengthens the decision weight of strong correlation parameters, forming a dynamic model that takes into account data quality and correlation strength.

[0045] The traditional scheme only uses a single weight dimension, for example, relying only on the stability weight leads to underestimation of high correlation but fluctuation parameters, or using only the correlation weight ignores data reliability. The present scheme simultaneously constrains data stability and fault correlation by the multiplication mechanism, combined with differentiated time period setting and data removal rules, effectively avoiding misjudgment caused by environmental interference or sensor drift. In addition, the calibration mechanism of the new equipment fills the gap in the scene without historical data in the prior art, ensuring the applicability of weight calculation in the whole life cycle of the equipment.

[0046] The application solves the problem of fault correlation analysis deviation caused by single weight dimension, filters out key parameters with high stability and high correlation through double weight product, and significantly reduces positioning errors caused by data fluctuations or correlation misjudgment. The new device calibration mechanism ensures the effectiveness of the weight in the initial operation stage, avoiding the failure of correlation analysis caused by no historical data. The differentiated time period setting enables the weight calculation to dynamically adapt to different load scenarios, improving the fault positioning accuracy in complex industrial environments.

[0047] The application further proposes a hierarchical protocol structure adopted in the data transmission step, including that the physical layer adopts a modulation mode conforming to the NB-IoT standard, the network layer uses the NB-IoT protocol, the transmission layer uses the UDP protocol, the application layer uses the CoAP protocol, the server triggers a retransmission mechanism when the verification fails, and the maximum number of retransmissions is three.

[0048] Among them, the physical layer adopts a modulation mode conforming to the NB-IoT standard, which means that the modulation technology for maintaining stable connection in a low signal-to-noise ratio environment is adopted. Specifically, an industrial-grade NB-IoT module can be used to achieve this. The module integrates an aluminum shielding layer and passes electromagnetic compatibility testing, and is used to maintain the stability of physical signals in a high electromagnetic interference environment of a distribution box.

[0049] The network layer uses the NB-IoT protocol, which means that a communication protocol based on wide-area coverage and deep penetration characteristics is used. Specifically, the NB-IoT Cat-NB2 protocol can be used to achieve this. It supports power saving mode and extended discontinuous reception function, and is used to establish a remote communication link in an outdoor power distribution scenario.

[0050] The transmission layer uses the UDP protocol, which means a lightweight protocol based on connectionless transmission mechanism. Specifically, a low-latency data transmission mode can be used to achieve this, which is used to avoid data accumulation caused by network congestion. The application layer uses the CoAP protocol, which means an application layer protocol based on lightweight design. Specifically, a block transmission mode can be used to achieve this, which is used to adapt to the low bandwidth characteristics of NB-IoT by transmitting large volume data in blocks.

[0051] The server triggers the retransmission mechanism, which means a data recovery strategy based on verification failure. Specifically, a three-retransmission upper limit mechanism can be used to achieve this, which is used to avoid excessive consumption of network resources while ensuring data integrity.

[0052] Specifically, the physical layer resists electromagnetic interference through the shielding design of the industrial-grade NB-IoT module, the network layer establishes a stable connection using the wide coverage characteristics of the Cat-NB2 protocol, the transmission layer reduces transmission delay through the UDP protocol, the application layer transmits the weighted data set in blocks using the block transmission mode of the CoAP protocol, and the server triggers three retransmissions when the CRC verification fails.

[0053] When each protocol layer works together, the physical layer and the network layer guarantee the basic communication quality, the transmission layer and the application layer optimize the data transmission efficiency, and a complete transmission link suitable for low-bandwidth and high-interference environment is formed. When the data packet fails to pass the check due to network fluctuations, the retransmission mechanism recovers the missing data within a limited number of times, avoiding resource waste caused by unlimited retransmission.

[0054] The existing scheme usually only adopts single protocol layer optimization or basic retransmission mechanism, for example, only relies on NB-IoT transmission without designing block transmission and local emergency processing. The present scheme builds a complete reliable transmission system through multi-dimensional cooperation of the physical layer anti-interference module, the application layer block transmission, and the local alarm upon check failure, solving the problem of inaccurate subsequent analysis caused by transmission distortion in the prior art.

[0055] The present application realizes high-reliability transmission of distribution box operation data in a complex network environment, resists electromagnetic interference and network fluctuations through a layered protocol architecture, ensures data integrity through limited retransmission and local emergency mechanism, and provides accurate data basis for subsequent circuit-element correlation analysis, thereby reducing the fault positioning misjudgment rate caused by transmission distortion.

[0056] The present application further proposes that the circuit-element correlation analysis includes calculating the Pearson correlation coefficient of the element state parameter and the operating circuit based on the weighted data set, and if the correlation coefficient exceeds a set threshold, a correlation relationship is established. The set threshold is determined based on historical fault data statistics of the same type of distribution box and is not less than 0.7. The mutual influence relationship between circuits is analyzed based on historical current and voltage data to form an influence correlation table. The weighted data are compared with the preset normal parameter range to identify abnormal circuits, and the abnormal circuits are sorted according to their influence weights.

[0057] The Pearson correlation coefficient refers to a statistical quantity for measuring the degree of linear correlation between two variables. It can be calculated by using the weighted element state parameter and the circuit operating parameter to calculate the covariance and standard deviation. The correlation is calculated by using the parameter optimized by dynamic weight, which can effectively filter out interference data with low stability or low correlation.

[0058] The influence correlation table refers to structured data recording the mutual influence relationship between circuits. It can be implemented by using the correlation probability of different circuit parameter abnormalities in the weighted current and voltage data within nearly 24 hours. The real influence relationship is reflected by the weighted data, avoiding misjudgment caused by distortion of original data.

[0059] The normal parameter range refers to the reasonable fluctuation interval of the parameter when the device is operating without failure. It can be implemented by calculating the mean and standard deviation range using the baseline data during the device running-in period, and periodically updating the parameter benchmark of the aging device. Dynamic calibration design can adapt to the parameter drift caused by changes in component performance.

[0060] Specifically, in the circuit-element correlation analysis process, first, the correlation calculation is performed using the parameters optimized by dynamic weight, such as the weighted parameter combination of contact resistance and main loop current, and weakly correlated data is filtered by setting a scenario-adjusted threshold, such as a threshold of 0.75 for outdoor power distribution scenarios to reduce false negatives caused by electromagnetic interference.

[0061] Secondly, based on the weighted historical current and voltage data, an influence correlation table is constructed, for example, when the weighted current of the main loop is abnormal, the probability of voltage abnormality of the branch is counted to see if it exceeds the preset threshold, thereby establishing the implicit correlation relationship between circuits.

[0062] Finally, the real-time weighted data is compared with the dynamically updated normal range, for example, if the current value of the contact resistance exceeds the baseline range during the running-in period, it is determined to be abnormal, and a sorting weight is generated according to the number of associated circuits recorded in the influence correlation table and the priority of the key loop, for example, the abnormal circuit containing the production line power loop automatically promotes the ranking level.

[0063] The traditional method directly uses the original sensor data for correlation analysis without considering the influence of data distortion during transmission on correlation calculation, resulting in a high false negative rate of abnormal circuits.

[0064] And the scheme uses weighted data sets for correlation analysis, effectively suppressing noise interference; at the same time, the dynamically updated normal parameter range and the scenario-based threshold adjustment mechanism solve the problem that the fixed parameter range cannot adapt to equipment aging and environmental differences; in addition, through the key loop prioritization mechanism, the resource allocation efficiency of fault troubleshooting is optimized.

[0065] The present application realizes high-precision circuit-element correlation analysis in a complex network environment, solving the problem of correlation misalignment caused by data transmission distortion. For example, in the outdoor distribution box scenario, the weighted data filters the abnormal fluctuations caused by electromagnetic interference, significantly reducing the correlation coefficient calculation error; the dynamic parameter range adapts to the contact resistance drift of aging equipment, avoiding false negatives; the key loop prioritization mechanism shortens the troubleshooting time of production line power faults, improving the operation and maintenance efficiency of industrial scenarios.

[0066] The present application further proposes that the normal parameter range is based on the weighted data statistics during the fault-free operation of the equipment, and the mean plus or minus 3 times the standard deviation is used as the initial range, and is updated regularly to adapt to the parameter drift caused by component aging.

[0067] Among them, the normal parameter range refers to the reference interval for determining whether the running state of the circuit or component is abnormal, which can be specifically realized by statistical calculation of the weighted data set collected during the fault-free operation of the equipment, ensuring that the parameter range is derived from real data of the healthy state of the equipment.

[0068] The mean plus or minus 3 times the standard deviation refers to a mathematical method for determining the parameter fluctuation range based on the principle of statistical normal distribution, and can be implemented by calculating the arithmetic mean and standard deviation of the weighted data set, and taking the mean plus or minus 3 times the standard deviation as the upper and lower limits to cover 99.7% of the data distribution interval under the normal operation state of the equipment.

[0069] The periodic update refers to an operation of dynamically adjusting the normal range according to the equipment running time or parameter change trend, and can be implemented in a quarterly period update or an abnormal trigger update manner, by re-collecting the fault-free running data and re-calculating the statistical value, so that the parameter range is synchronized with the element aging process.

[0070] Specifically, in the initial running stage of the equipment, the initial parameter range is generated by continuously collecting stable state data after the running-in period, and the influence of unstable data in the starting stage on the statistical result is excluded. During operation, the element state parameters are continuously monitored based on the weighted data set, and when it is detected that the parameters are continuously close to the current range boundary, the data re-collection and range update process is triggered. The same statistical method as in the initial stage is used for updating, but only the data during the recent fault-free operation is used, so as to ensure that the parameter range dynamically reflects the performance change caused by element aging.

[0071] The traditional method uses a fixed threshold or a statistical range based on raw data, which cannot adapt to the parameter drift caused by element aging. The present scheme cooperates with the weighted data through a dynamic updating mechanism, so that the normal range always matches the actual state of the equipment, and avoids the misjudgment problem caused by material performance degradation.

[0072] The present application can accurately distinguish between normal parameter drift caused by element aging and abnormal fluctuation caused by real failure, and effectively reduce the misjudgment rate caused by inaccurate range setting. The dynamic range generated based on the statistical method can adapt to the state change of the equipment throughout the life cycle, ensure that the fault judgment standard is consistent with the actual performance degradation trend of the element, and improve the reliability of the fault positioning result.

[0073] The present application further proposes that the power variable test includes preferentially adjusting the environmental operating parameters with the highest weight in the weighted data, monitoring the parameter changes of the related circuit after adjustment in real time, removing the circuit from the abnormal list if the parameter is stable in the normal range for 10 seconds, detecting the weighted parameters of the corresponding element based on the circuit-element correlation, and determining that the element is faulty if the parameter exceeds the normal range.

[0074] The priority adjustment of the environment operating parameter with the highest weight refers to determining the parameter adjustment order according to the dynamic weight calculation result, and can be specifically implemented by using a dynamic weight calculation module, and the parameter with the maximum weight value is tested first to quickly focus on the key variable that has the greatest impact on the system. The real-time monitoring of the changed parameter refers to continuously obtaining the adjusted operating data by using a data acquisition module, and can be specifically implemented by using a timing sampling mechanism to ensure the real-time and continuity of the parameter stability determination.

[0075] The stability in the normal range for 10 seconds refers to setting a quantitative time threshold as a state release condition, which can be specifically implemented by using a timer and a data comparison module to avoid instantaneous fluctuation interference in the determination result. The circuit-element correlation relationship detection refers to matching the abnormal circuit and the associated element based on the preset mapping relationship, which can be specifically implemented by using an associated database query to locate the fault from the circuit level to the element level.

[0076] Specifically, in the power variable test process, first, the environment operating parameter with the highest weight is selected according to the dynamic weight calculation result for adjustment, for example, for a low-voltage distribution box, the current is adjusted in steps, and the adjustment step is controlled within 5% of the rated current each time. After adjustment, the data is collected after maintaining a stable state for 30 seconds.

[0077] After the parameter adjustment is completed, the operating data of the related circuit is monitored in real time, and if the parameter is in the normal range for 10 seconds, the abnormal state is released, wherein the normal range uses the mean value ± 1.5 times the standard deviation as the determination boundary.

[0078] For the abnormal circuit that has not been restored, the weighted parameters of the corresponding element are detected in combination with the preset circuit-element correlation relationship, for example, when detecting the contact resistance of the contactor, an infrared thermal imager is used to verify the contact temperature, and only when both the sensor data and the physical characteristic detection result are abnormal is the element determined to be faulty.

[0079] The traditional scheme does not set a safety strategy for parameter adjustment, and there is a risk of equipment damage caused by blind adjustment. The present scheme sets differentiated adjustment rules by distinguishing between high-voltage and low-voltage distribution scenarios, for example, forced power-off testing in a high-voltage scenario to ensure operation safety. The prior art lacks a two-way standard for stability determination, and the present scheme forms a double determination mechanism by using a continuous stability time threshold and an abnormal number threshold, for example, 2 times of over-limit within 5 seconds is directly marked as not restored, which improves the troubleshooting efficiency. The prior art relies on single sensor data to determine the fault, and the present scheme introduces a physical characteristic verification link, for example, the capacitance detection needs to be retested by a special tester, to avoid misjudgment caused by sensor errors.

[0080] This application addresses the inefficiency caused by improper parameter adjustment order during power variable testing. It reduces the number of invalid tests through a weighted priority mechanism; avoids misjudgments by using a quantitative stability judgment standard to ensure the reliability of test results; and improves positioning accuracy through a dual verification mechanism for component-level faults to avoid erroneous diagnosis caused by distortion of a single data source.

[0081] This application further proposes weight allocation optimization including the following steps: if the fault location is accurate, increase the correlation weight of the relevant sensor parameters; if a circuit is excluded as a non-faulty circuit a certain number of times, decrease the correlation weight of its relevant parameters; set upper and lower limits for the correlation weight and periodically evaluate the effect of weight adjustment.

[0082] The correlation weight adjustment refers to dynamically correcting the contribution of sensor parameters in data analysis based on the fault location results. Specifically, this can be achieved using a proportional coefficient based on location accuracy; for example, increasing the weight by 20% when the location is perfectly accurate. This adjustment mechanism enhances the sensitivity of key parameters, thereby improving subsequent detection accuracy.

[0083] The preset number of attempts refers to the cumulative number of non-faulty circuits that need to be eliminated before a weight reduction is triggered. This can be achieved using a three-threshold method commonly used in industrial maintenance to avoid weight fluctuations caused by a single misjudgment. Upper and lower limit settings refer to setting numerical boundaries for the relevance weights. This can be achieved using a range of 0.1 to 0.9 to prevent a single parameter from excessively dominating the analysis process. Periodic evaluation refers to periodically verifying the effectiveness of the weight adjustment. This can be achieved using three quantitative indicators: positioning accuracy, false alarm rate, and fault troubleshooting time, ensuring that the weight strategy continuously adapts to the equipment status.

[0084] Specifically, a weight adjustment operation is performed immediately after each fault location is completed, dynamically determining the adjustment ratio based on the accuracy of the location results. When on-site verification confirms accurate location, the relevance weight of associated parameters is increased proportionally, while the weight values ​​of non-associated parameters are decreased.

[0085] For circuits that are repeatedly ruled out as non-faulty, the weights of their relevant parameters are reduced after a preset threshold number of failures is reached, thus minimizing interference from invalid data. By setting upper and lower limits for the weights, analytical biases caused by imbalances in parameter sensitivity are avoided. During periodic evaluations, the weight adjustment strategy is recalibrated when core indicators deviate from preset thresholds, forming a continuously optimized closed-loop control system.

[0086] Existing solutions only use unidirectional weight calculation and lack feedback on location results, leading to a disconnect between weight allocation and fault location. This solution addresses the problem of multiple rounds of misjudgment caused by adjustment delays in traditional methods by establishing an instant adjustment mechanism to optimize weight allocation immediately after each location.

[0087] Meanwhile, quantitative evaluation indicators and threshold triggering mechanisms are introduced to make the weight optimization process verifiable and controllable, overcoming the defects of blind adjustment in existing technologies.

[0088] The application effectively reduces the deviation of fault location caused by data transmission distortion and reduces non-critical parameter interference through dynamic optimization of weight distribution. The closed-loop feedback mechanism ensures real-time linkage between weight adjustment and positioning results, improving the relevance and accuracy of anomaly detection. The upper and lower limits of the weight constraint maintain the synergistic effect of multiple parameters, preventing the analysis process from being dominated by a single parameter. The periodic evaluation mechanism ensures that the weight strategy is effective in the long term, adapting to parameter drift caused by equipment aging and environmental changes.

[0089] The application further provides an electronic device, including a processor and a memory, the memory storing a computer program, the computer program being executed by the processor to implement the fault detection method of the distribution box.

[0090] The processor refers to a hardware unit that performs arithmetic logic operations and control instructions, which can be implemented by a multi-core CPU or an embedded microcontroller, and is used to run dynamic weight calculation, circuit-element association analysis and fault location logic in the fault detection algorithm.

[0091] The memory refers to a physical medium that stores program code and running data, which can be implemented by a flash memory chip or a solid state disk, and is used to save raw data collected by sensors, dynamic weight calculation results and fault location rule library.

[0092] The computer program refers to a specific instruction set that implements the fault detection method, which can be implemented by a C++ or Python language compiled executable file, including data encapsulation transmission, CRC check failure retransmission control and weight distribution optimization module.

[0093] Specifically, the processor performs dynamic weight calculation on the environmental operating parameters and element state parameters collected by the sensors in the distribution box by executing the computer program in the memory, generates a weighted data set, and encapsulates and transmits it through the NB-IoT network protocol. After receiving the data, the server performs CRC check, and if the check is passed, it performs circuit-element association analysis based on the weighted data set, identifies abnormal circuits and sorts them by impact weight, and finally locates the element-level fault through power variable testing.

[0094] In a complex network environment, the processor ensures data transmission integrity through program-controlled hierarchical protocol structure and CRC check mechanism, and optimizes the identification priority of abnormal circuits based on dynamic weight adjustment, suppressing error accumulation caused by electromagnetic interference and data packet loss.

[0095] The traditional scheme relies on a single software algorithm to realize fault positioning, without considering the influence of the hardware execution environment on the reliability of data transmission, resulting in inaccurate abnormal circuit identification results due to network interference. The scheme realizes anti-interference transmission at the data acquisition end through the hardware cooperation of the processor and the memory, combined with layered protocol packaging and CRC check retransmission mechanism, and improves the anti-noise ability of abnormal circuit identification through a dynamic weight optimization algorithm.

[0096] The application solves the problem of inaccurate fault circuit or component positioning and high false alarm rate caused by data transmission distortion in complex network environment during power distribution box fault detection. Through the hardware and software cooperative data processing mechanism, the false judgment probability caused by network packet loss or electromagnetic interference is reduced, and the accuracy and maintenance efficiency of component-level fault positioning are improved.

[0097] The application further provides a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to realize the fault detection method of the power distribution box, including collecting environmental operation parameters and component state parameters through multiple types of sensors deployed in the power distribution box, the environmental operation parameters including temperature, humidity, sulfur hexafluoride gas concentration, current and voltage, and the component state parameters including contact resistance of the contactor, capacitance of the capacitor bank and insulation resistance of the line; performing dynamic weight calculation on the collected raw sensor data, the dynamic weight including stability weight based on standard deviation of parameter historical data and correlation weight based on correlation coefficient between the parameter and the circuit or component, and the calculated dynamic weight is used to generate a weighted data set; packaging the weighted data set through NB-IoT network protocol, the data packet containing device identification, timestamp, weighted data set and CRC check code, and transmitting to a remote server; performing CRC check on the received data by the server, and performing circuit-component correlation analysis based on the weighted data set after the check is passed, identifying abnormal circuits and ranking them according to the influence weight; performing power variable test on the abnormal circuits ranked at the top, and realizing component-level fault positioning combined with the component state parameters; feeding back the fault positioning result to the dynamic weight calculation process for optimizing the weight distribution in subsequent data collection.

[0098] The computer readable storage medium refers to a physical carrier capable of storing computer program code, which can be implemented by solid state disk, flash chip or optical disc, and its function is to ensure the integrity and consistency of the fault detection method in complex network environment through solidification of program logic.

[0099] Dynamic weight calculation refers to assigning weight values to different sensor parameters through mathematical operations, which can be realized by the product of stability weight and correlation weight, the stability weight is calculated based on historical standard deviation, and the correlation weight is calculated based on the maximum value of Pearson correlation coefficient, and its function is to enhance the representation ability of key parameters through data screening.

[0100] NB-IoT network protocol encapsulation refers to packing data in a communication standard format, which can be implemented by adopting a layered protocol structure, the physical layer adopts NB-IoT modulation mode, and the application layer adopts CoAP protocol, and its function is to guarantee the reliability of data transmission in high interference environment through low-power wide-area network technology.

[0101] CRC check code refers to cyclic redundancy check code, which can be implemented by generating check value by polynomial division, and its function is to avoid distorted data from entering the analysis link through data integrity verification.

[0102] Specifically, when the computer program is loaded from the storage medium to the processor for execution, first, the temperature and humidity sensor in the distribution box collects environmental parameters, and at the same time, the element state data is obtained through the contact resistance measurement module.

[0103] The collected raw data is calculated by dynamic weight, for example, the historical standard deviation of the capacitance parameter is calculated, if the standard deviation is lower than the set threshold, a high stability weight is given, and at the same time, the correlation coefficient of the parameter and each circuit is calculated, and the maximum value is taken as the correlation weight. The weighted data set is encapsulated into a data packet containing the unique identification of the device and the timestamp through the NB-IoT protocol, and is transmitted to the server after adding the CRC check code. The server performs CRC check on the received data, and if the check fails, the data packet is discarded and the retransmission mechanism is triggered.

[0104] After the check is passed, the server analyzes the relevance of the circuit and the element based on the weighted data set, for example, when the weighted value of the line insulation resistance exceeds the normal range and the correlation coefficient with a certain circuit exceeds 0.7, the circuit is marked as abnormal.

[0105] When performing power variable test on the abnormal circuit, the parameter with the highest weight is adjusted first and the element state change is monitored, for example, if the contact resistance returns to normal after adjusting the voltage, it is determined that the contactor is faulty. The final positioning result is fed back to the dynamic weight calculation module, for example, after a certain element is confirmed as the fault source, the correlation weight of its related sensor parameters is improved in the subsequent collection.

[0106] In the prior art, the storage medium is only used to store the basic detection program, and the closed-loop optimization mechanism is not designed for complex network environment, which leads to that the data collection and transmission link is easy to be disturbed when the program is executed. The present scheme solidifies the program containing dynamic weight optimization logic through the storage medium, so that the weight distribution can be adjusted based on the historical fault positioning result every time data collection is performed.

[0107] The prior art data transmission lacks a CRC check combined with a layered protocol encapsulation mechanism, resulting in data packets being directly entered into the analysis link after being lost or damaged in a high interference environment. The present scheme provides double protection through the layered encapsulation and CRC check of the NB-IoT protocol, automatically triggering retransmission when the data packet is damaged, and ensuring the integrity of the data input into the analysis module.

[0108] The prior art fault positioning does not establish a weight feedback mechanism, resulting in the inability to continuously improve data effectiveness. The present scheme forms a closed-loop optimization of parameter screening by feeding the positioning results back to the weight calculation, for example, the weight of the related parameters of a certain circuit gradually decreases to reduce the proportion of invalid data after being excluded from fault multiple times.

[0109] The present application solves the problems of high misjudgment rate and rough positioning caused by data transmission distortion in complex network environment. By solidifying the closed-loop detection logic in the storage medium, it ensures the coordinated operation of dynamic weight calculation, data transmission verification, and fault positioning link during program execution, improves the accuracy of abnormal circuit identification, and improves the accuracy of component-level fault positioning.

[0110] At the same time, CRC check and layered protocol encapsulation reduce the data transmission distortion rate and avoid misjudgment caused by damaged data packets. The dynamic weight feedback mechanism continuously optimizes data collection effectiveness and reduces the time for manual troubleshooting during maintenance.

[0111] The present application further proposes a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement a fault detection method for a distribution box, including collecting environmental operation parameters and component state parameters through multiple types of sensors deployed in the distribution box; performing dynamic weight calculation on the collected raw sensor data; encapsulating and transmitting the weighted data set through the NB-IoT network protocol to a remote server; performing CRC check on the received data by the server, and performing circuit-component correlation analysis based on the weighted data set; performing power variable testing on abnormal circuits to achieve component-level fault positioning; feeding the fault positioning results back to the dynamic weight calculation process to optimize weight distribution.

[0112] The dynamic weight calculation refers to combining the historical stability of sensor data with the correlation of circuit components to generate a weighted value, which can specifically use standard deviation to calculate stability weight, Pearson coefficient to calculate correlation weight, and the product of the two as the final weight for screening key parameters in complex data. The NB-IoT network protocol encapsulation refers to using a layered communication architecture for data transmission, which can specifically use a layered structure of physical layer NB-IoT modulation, network layer NB-IoT protocol, transport layer UDP protocol, and application layer CoAP protocol, in combination with CRC check and three times retransmission mechanism, to ensure transmission integrity in low bandwidth and high interference environment.

[0113] The circuit-element correlation analysis refers to establishing a mathematical correlation model of operating parameters and device elements. Specifically, the Pearson coefficient can be used to calculate the correlation between parameters and circuits, and the normal range threshold formed by historical data is used to identify abnormal circuits to improve the accuracy of fault location. The weight distribution optimization refers to adjusting the data acquisition strategy in reverse according to the fault location results. Specifically, the weight coefficient of the accurate fault-related parameter can be increased, and the weight coefficient of the false alarm circuit can be reduced to form a closed-loop feedback mechanism for continuously improving detection accuracy.

[0114] Specifically, when the computer program is executed, first, multi-dimensional data is collected through temperature, humidity, gas concentration, and other environmental sensors and contact resistance, capacitance, and other element sensors. The dynamic weight calculation module generates a weighted data set according to the historical fluctuation degree of each parameter and its correlation strength with circuit faults, for example, assigning higher weights to parameters with high stability and assigning higher weights to parameters strongly related to specific element faults.

[0115] The weighted data is transmitted through the hierarchical encapsulation of the NB-IoT protocol, for example, BPSK modulation for anti-interference at the physical layer, CoAP protocol for compressing data packet size at the application layer, and CRC check code for verifying data integrity. If the check fails, it triggers three retransmissions. After the server completes the check, it establishes a correlation model by calculating the correlation coefficient of each parameter and circuit element, for example, when the Pearson coefficient of a certain line current parameter and the contactor resistance exceeds 0.7, a strong correlation is established, and abnormal circuits are identified in combination with the preset normal range threshold and sorted by impact degree.

[0116] The power variable test module preferentially adjusts high-weight parameters and monitors the response, for example, after adjusting the highest-weight voltage parameter, if the related circuit parameters return to normal within 10 seconds, the circuit fault is excluded, otherwise, the fault element is located in combination with the element state parameter detection result. Finally, the positioning result is fed back to the weight calculation module, for example, when a certain capacitor fault is accurately identified, the correlation weight coefficient of the capacitance parameter is increased, forming a closed-loop optimization mechanism.

[0117] The existing method transmits raw sensor data in a complex network environment, which lacks hierarchical anti-interference mechanism and data integrity verification, resulting in subsequent analysis relying on distorted data, causing more than 20% false alarm rate. The present scheme uses the hierarchical encapsulation structure of the NB-IoT protocol, for example, narrowband anti-interference modulation at the physical layer and lightweight CoAP protocol at the application layer, combined with CRC check and three retransmissions, to reduce the data transmission packet loss rate to below 5%.

[0118] The existing method cannot adapt to the parameter drift caused by equipment aging when analyzing the fault circuit with fixed weight, and the scheme dynamically calculates the weight and adopts the closed-loop feedback mechanism, for example, dynamically adjusts the weight according to the historical data standard deviation, optimizes the correlation coefficient based on the fault location result, so that the element-level positioning accuracy is improved by more than 40%.

[0119] The application solves the problem of inaccurate fault location caused by data transmission distortion in a complex network environment, ensures data integrity through hierarchical protocol encapsulation and multiple verification mechanisms, and improves the reliability of basic data for fault analysis. At the same time, through dynamic weight calculation and closed-loop feedback optimization, the fault location accuracy is improved from the circuit level to the element level, and maintenance personnel can directly replace the faulty components without checking the entire circuit, and the detection efficiency is improved by more than 50%. In addition, the adaptive adjustment mechanism of weight distribution can automatically adapt to the parameter changes caused by equipment aging, so that the detection method remains stable and reliable throughout the life cycle of the equipment.

[0120] The above only describes the embodiments of the application and does not limit the protection scope of the application. For those skilled in the art, the application can have various changes and variations. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the application shall be included in the protection scope of the application.

Claims

1. A fault detection method for a distribution box, characterized in that, include: Environmental operating parameters and component status parameters are collected by various sensors deployed in the distribution box. The environmental operating parameters include temperature, humidity, sulfur hexafluoride gas concentration, current and voltage. The component status parameters include contact resistance of contactors, capacitance of capacitor banks and insulation resistance of lines. Dynamic weights are calculated on the collected raw sensor data. The dynamic weights include stability weights based on the standard deviation of historical parameter data and correlation weights based on the correlation coefficient between the parameter and the circuit or component. The calculated dynamic weights are used to generate a weighted dataset. The weighted dataset is encapsulated using a network protocol. The data packet contains the device identifier, timestamp, weighted dataset, and CRC checksum, and is then transmitted to the remote server. After receiving the data, the server performs CRC verification. If the verification passes, it performs circuit-component correlation analysis based on the weighted dataset to identify abnormal circuits and sort them according to their impact weight. Power variable tests are performed on the faulty circuits that are ranked first, and component-level fault location is achieved by combining component status parameters. The fault location results are fed back to the dynamic weight calculation process to optimize the weight allocation in subsequent data collection. The circuit-component correlation analysis includes: The Pearson correlation coefficient between component status parameters and operating circuits is calculated based on a weighted dataset. If the correlation coefficient exceeds a set threshold, an association is established. The set threshold is determined based on historical fault data of similar distribution boxes and is not less than 0.

7. Based on historical current and voltage data, the mutual influence relationships between circuits are analyzed to form an influence correlation table; The weighted data is compared with the preset normal parameter range to identify abnormal circuits and sort them according to their impact weight. The power variable test includes: Prioritize adjusting the environmental operating parameters with the highest weight in the weighted data; Real-time monitoring of parameter changes in related circuits after adjustment; if the parameters remain stable within the normal range for 10 consecutive seconds, they are removed from the anomaly list. By combining the circuit-component relationship, the weighted parameters of the corresponding components are detected. If they exceed the normal range, the component is determined to be faulty.

2. The fault detection method for a distribution box according to claim 1, characterized in that, The dynamic weight is calculated as the product of the stability weight and the correlation weight. The stability weight is calculated based on the historical standard deviation of sensor data within a preset time period; the smaller the standard deviation, the higher the weight. The correlation weight is calculated based on the Pearson correlation coefficient, taking the maximum value among the correlation coefficients of this parameter with each circuit or component.

3. The fault detection method for a distribution box according to claim 1, characterized in that, The weighted dataset is encapsulated using a network protocol. The data packet contains a device identifier, timestamp, weighted dataset, and CRC checksum, and is transmitted to a remote server. A layered protocol structure is employed, including: The physical layer adopts a modulation method that conforms to the NB-IoT standard; The network layer uses the NB-IoT protocol; The transport layer uses the UDP protocol; The application layer uses the CoAP protocol; A retransmission mechanism is triggered when the server-side verification fails, with a maximum of three retransmissions.

4. The fault detection method for a distribution box according to claim 1, characterized in that, The normal parameter range is derived from weighted data statistics during the equipment's trouble-free operation. The initial range is set at the mean ± 3 times the standard deviation and is updated periodically to accommodate parameter drift caused by component aging.

5. The fault detection method for a distribution box according to claim 1, characterized in that, The optimization of the weight allocation includes: If the fault location is accurate, increase the correlation weight of the relevant sensor parameters; If a circuit is excluded from being a non-faulty circuit a certain number of times, the correlation weight of its relevant parameters will be reduced. Set upper and lower limits for relevance weights, with the upper limit not exceeding 0.9 and the lower limit not lower than 0.1, and periodically evaluate the effect of weight adjustments.

6. An electronic device comprising a processor and a memory, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it implements the fault detection method for the distribution box as described in any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the fault detection method for the distribution box as described in any one of claims 1 to 5.

Images