A method and system for detecting faults in a cable branch box

Abstract

The application discloses a kind of cable branch box fault detection method and system, specifically relates to branch box fault detection technical field, including by multiple sensors being set in cable branch box Real-time acquisition multidimensional operating state data;The data collected are extracted features, obtain multiple dimensions of feature sequence, calculate the correlation coefficient matrix between each dimension feature sequence, generate the comprehensive feature vector reflecting branch box operating state;Based on comprehensive feature vector and dynamic threshold mechanism, real-time detection and early warning of fault are carried out and fault type is judged, the load state of branch box is adjusted in real time in the application Contact temperature dynamic threshold and vibration dynamic threshold, prevent threshold setting in fault detection link from being mostly fixed value, cannot be dynamically adjusted according to branch box real-time operation scene, prone to false alarm or the problem of missing report.

Description

Technical Field

[0001] This invention relates to the field of fault detection technology for cable branch boxes, and more specifically, to a fault detection method and system for cable branch boxes. Background Technology

[0002] Cable branch boxes are key equipment in power distribution networks used to realize the branching, switching and distribution of power cables and power energy. They are usually installed at outdoor or indoor distribution nodes. Through internal busbars, insulating supports and pluggable contacts, they connect the main cable to multiple branch circuits, complete the collection and distribution of load current, and also have the functions of line isolation and fault protection.

[0003] However, existing branch box operation and maintenance systems have systemic shortcomings in status monitoring and fault detection: status monitoring often uses single-dimensional data collection, and the data from each monitoring module are isolated from each other, without forming a correlation analysis mechanism. This makes it impossible to cross-verify the cause and type of fault through multi-dimensional data when a fault occurs. Furthermore, the threshold settings in the fault detection stage are mostly fixed values, which cannot be dynamically adjusted according to the real-time operation of the branch box, easily leading to false alarms or missed alarms and increasing ineffective operation and maintenance costs. Therefore, this invention proposes a fault detection method and system for cable branch boxes to solve the above problems. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present invention provide a fault detection method and system for cable branch boxes to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a fault detection method for cable branch boxes, comprising the following steps: S1. Real-time acquisition of multi-dimensional operating status data through multiple sensors installed in the cable branch box; S2. Extract features from the collected data to obtain feature sequences in multiple dimensions. Calculate the correlation coefficient matrix between the feature sequences in each dimension. When the absolute value of the correlation coefficient between any two dimensions exceeds a preset threshold, it is determined that there is a significant correlation between the two dimensions of data, and a comprehensive feature vector reflecting the operating status of the branch box is generated. S3. Based on the comprehensive feature vector and dynamic threshold mechanism, perform real-time fault detection and early warning and determine the fault type; In step S2, the comprehensive feature vector is obtained in the following way: Extract the instantaneous values ​​of the feature sequences of each dimension at the current moment to form the basic feature vector; Extract the correlation coefficients between each dimension from the correlation coefficient matrix, and determine whether the absolute value of the correlation coefficient reaches the preset threshold. If it does, take the value of the correlation coefficient; otherwise, take zero as the correlation feature value. Arrange and combine multiple correlation feature values ​​into a correlation feature vector. The basic feature vector and the associated feature vector are concatenated to generate a comprehensive feature vector that reflects the operating status of the branch box; The associated feature vector includes at least a first associated feature value representing the correlation between load current and contact temperature, a second associated feature value representing the correlation between load current and vibration, a third associated feature value representing the correlation between load current and humidity, a fourth associated feature value representing the correlation between contact temperature and vibration, a fifth associated feature value representing the correlation between contact temperature and humidity, and a sixth associated feature value representing the correlation between vibration and humidity. Each associated feature value is taken when the absolute value of the corresponding correlation coefficient reaches the preset threshold, otherwise it is taken as zero. In step S3, fault detection is performed through the following steps: Based on the comparison of the contact temperature component, vibration effective value component, load current effective value component, and humidity value component in the comprehensive feature vector with the corresponding dynamic threshold or rated threshold, the temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly are calculated. If the current anomaly is greater than zero and the temperature anomaly is greater than zero, and the first associated characteristic value is not zero, then it is determined to be a contact overheating fault caused by overload. If the temperature anomaly is greater than zero and the vibration anomaly is greater than zero, and the fourth associated characteristic value is not zero, then it is determined to be a local overheating and vibration fault caused by poor contact. If the temperature anomaly is greater than zero and the humidity anomaly is greater than zero, and the fifth associated characteristic value is not zero, then it is determined to be an insulation deterioration and heating fault caused by moisture. If the current anomaly degree is greater than zero and the vibration anomaly degree is greater than zero, and the second associated characteristic value is not zero, then it is determined to be an abnormal vibration fault induced by electromagnetic force. If only the humidity anomaly is greater than zero, and the current anomaly, temperature anomaly, and vibration anomaly are all zero, and the third and fifth associated characteristic values ​​are both zero, then the fault is determined to be due to excessively high ambient humidity. If multiple anomalies are greater than zero and multiple associated feature values ​​are not zero, then it is determined to be a composite fault, and all single fault types that meet the conditions are output.

[0006] Preferably, in step S1, the sensor includes at least a current transformer, a temperature sensor, a vibration sensor, and a temperature and humidity sensor, and the multidimensional operating status data includes at least load current data, contact temperature data, vibration data, and humidity data.

[0007] Preferably, in step S2, the correlation coefficient matrix is ​​calculated in the following manner: Within a set sliding time window, the sample mean and standard deviation of each dimension feature sequence are calculated. For any two dimension feature sequences, the sum of the products of the deviations of each pair of sample values ​​from their respective means in the two dimension feature sequences is calculated, and the ratio of this product to the product of the sample standard deviations of the two dimension feature sequences is obtained. The correlation coefficient between the two dimensions is then formed by iterating through all dimension combinations.

[0008] Preferably, in step S3, the dynamic threshold mechanism includes: Calculate the ambient temperature compensation coefficient, which is obtained by multiplying the difference between the current ambient temperature and the baseline temperature by the temperature sensitivity coefficient. The temperature sensitivity coefficient is determined by the slope of the linear relationship between the change in ambient temperature and the change in contact temperature fitted by a steady-state temperature rise experiment. Calculate the dynamic threshold of the contact temperature, which is the sum of the base temperature reference, the product of the rated allowable temperature rise and the load rate correction factor, and the ambient temperature compensation factor; wherein, the load rate correction factor is determined based on the real-time load rate. Obtain the reference vibration effective value and multiply it by the vibration load correction coefficient determined based on the real-time load rate to obtain the vibration dynamic threshold. The vibration load correction coefficient is the square of the real-time load rate.

[0009] Preferably, in step S3, fault detection is performed through the following steps: The temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly are weighted and summed with their respective weighting coefficients to obtain the comprehensive fault degree; among which, the weighting coefficients are determined by the entropy weighting method based on the information entropy of each anomaly in historical fault data. A fault alarm is triggered when the overall failure rate exceeds the set threshold.

[0010] Preferably, when the contact temperature exceeds the dynamic threshold of the contact temperature, the temperature anomaly is determined based on the ratio of the magnitude of the excess to the difference between the dynamic threshold and the baseline temperature reference. The vibration anomaly degree is determined by the ratio of the excess amplitude to the vibration dynamic threshold when the effective vibration value exceeds the vibration dynamic threshold. The current anomaly is determined by the ratio of the excess magnitude to the rated current threshold when the effective value of the load current exceeds the rated current threshold. The humidity anomaly level is determined by the ratio of the magnitude of the exceedance to the humidity threshold when the humidity value exceeds the preset allowable humidity range.

[0011] Preferably, in step S3, the alarm type is determined by the following method: If the duration is less than the first preset duration and the overall failure rate falls below the preset safety value, it is determined to be an instantaneous fluctuation and the alarm is canceled. If the duration is between the first preset duration and the second preset duration and the overall failure rate shows an upward trend, it is determined that there are signs of failure. If the duration reaches or exceeds the second preset duration and the overall failure rate remains higher than the preset failure confirmation value, it is determined to be a real failure.

[0012] Preferably, the associated feature vector in the current time comprehensive feature vector is extracted, and multiple associated feature values ​​therein are compared with the benchmark associated feature values ​​obtained under normal operating conditions of the branch box; The fault type is determined based on whether each anomaly degree is greater than zero and the degree to which the current associated feature value deviates from the baseline associated feature value.

[0013] A fault detection system for a cable branch box includes the following steps: a data acquisition unit, used to acquire multi-dimensional operating status data in real time through multiple sensors installed inside the cable branch box; Environmental compensation unit, used to calculate the environmental temperature compensation coefficient; The dynamic threshold calculation unit is used to calculate the dynamic threshold of contact temperature and the dynamic threshold of vibration based on the real-time load rate and the ambient temperature compensation coefficient. The fault analysis unit is used to compare the components of the comprehensive feature vector with the corresponding dynamic thresholds or nominal thresholds to calculate the anomaly degree of each item, and then sum them up by weight to obtain the comprehensive fault degree. The verification feedback unit is used to determine the alarm type based on the alarm duration and the overall failure rate trend. The technical effects and advantages of this invention are as follows: This invention adjusts the dynamic thresholds for contact temperature and vibration in real time based on the load status of the branch box. This prevents the fault detection thresholds from being fixed values ​​that cannot be dynamically adjusted according to the real-time operating scenario of the branch box, which could easily lead to false alarms or missed alarms. Furthermore, by weighting and summing temperature anomalies, vibration anomalies, current anomalies, and humidity anomalies with their respective weighting coefficients, a comprehensive fault degree is calculated, thereby detecting faults. This prevents data from monitoring modules from being isolated from each other, which would prevent the cross-validation of faults through multi-dimensional data when a fault occurs. By cross-validating the fault cause based on multiple anomalies and associated feature values, the fault type can be quickly identified, providing clear guidance for rapid repair and root cause analysis, and shortening fault handling time. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the overall process of the fault detection method of the present invention.

[0015] Figure 2 This is a schematic diagram of the dynamic threshold mechanism of the present invention.

[0016] Figure 3This is a schematic diagram of the fault detection and early warning process of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1

[0019] In actual production, the threshold settings for fault detection are mostly fixed values, which cannot be dynamically adjusted according to the real-time operation of the branch box. This can easily lead to false alarms or missed alarms, increasing the cost of ineffective operation and maintenance. This embodiment is invented to solve the above problems.

[0020] Please see Figures 1 to 3 As shown, a fault detection method and system for a cable branch box according to an embodiment of the present invention includes: setting multiple sensors in the cable branch box to collect multi-dimensional operating status data of the cable branch box in real time; the multiple sensors include at least a current transformer, a temperature sensor, a vibration sensor, and a temperature and humidity sensor; and the multi-dimensional operating status data includes at least load current data, contact temperature data, vibration data, and humidity data. Feature extraction is performed on the collected data to obtain feature sequences in multiple dimensions. The correlation coefficient matrix between the feature sequences in each dimension is calculated. When the absolute value of the correlation coefficient between any two data exceeds a preset threshold, it is determined that there is a significant correlation between the two data, and a comprehensive feature vector reflecting the operating status of the branch box is generated. Fault detection steps: Real-time fault detection and early warning are achieved based on comprehensive feature vectors and dynamic threshold mechanisms, and the fault type is determined.

[0021] Feature extraction includes extracting the effective current value I from the load current data and extracting the contact temperature value from the temperature data. and ambient temperature value Extract the effective vibration value V from the vibration data, and extract the humidity value H from the humidity data; Obtaining the correlation coefficient matrix includes: setting a sliding time window and calculating the sample mean and sample standard deviation of the feature sequence for each dimension within the time window; For any two-dimensional feature sequences, calculate the ratio of the sum of the products of the deviations of each pair of sample values ​​from their respective means in the two-dimensional feature sequences to the product of the sample standard deviations of the two-dimensional feature sequences, and obtain the correlation coefficient between the two dimensions. After traversing all dimension combinations, a correlation coefficient matrix is ​​formed. The specific formula is as follows:

[0022] in, Let represent the correlation coefficient between the i-th dimension feature sequence and the j-th dimension feature sequence, and N represent the total number of data samples in the feature sequence, i.e., the number of sampling points within the sliding time window. This represents the data value of the i-th dimension feature sequence at the k-th sampling time. Similarly, This represents the data value of the j-th dimension feature sequence at the k-th sampling time. and Let i and j represent the average values ​​of the i-th dimension feature sequence and the j-th dimension feature sequence in N samples, respectively.

[0023] The instantaneous values ​​of the feature sequences of each dimension at the current moment are extracted to form the basic feature vector, specifically including the effective current value I and the contact temperature value. and ambient temperature value Vibration effective value V and humidity value H; The correlation coefficients between each dimension are extracted from the correlation coefficient matrix, and it is determined whether the absolute value of the correlation coefficient reaches a preset threshold. If the correlation coefficient is greater than the threshold, it is determined that there is a correlation, and the value of the correlation coefficient is retained as a valid correlation feature value. If the correlation coefficient is less than the threshold, it is determined that there is no significant correlation, and the correlation feature value of the dimension pair is assigned to 0 as the correlation feature value. The obtained correlation feature values ​​are arranged in sequence to form a correlation feature vector. The threshold used to determine the correlation coefficient is a preset threshold. The specific value can be determined by those skilled in the art based on the historical statistical analysis results of the actual operation data of the branch box, and is generally set between 0.5 and 0.8. The basic feature vector and the associated feature vector are concatenated to generate a comprehensive feature vector that reflects the operating status of the branch box.

[0024] Environmental compensation calculation: The environmental temperature compensation coefficient is calculated by multiplying the difference between the current environmental temperature and the baseline temperature by a temperature sensitivity coefficient. , or , in, This is the ambient temperature compensation coefficient. The current ambient temperature. Based on the temperature reference, The temperature sensitivity coefficient is obtained through a steady-state temperature rise experiment: the baseline temperature is recorded. With the branch box under no-load current and in a thermally stable state, its ambient temperature is changed and multiple ambient temperature values ​​are recorded simultaneously. and contact temperature value Calculate the ambient temperature deviation and the corresponding contact temperature deviation at each temperature point:

[0025]

[0026] For the obtained multiple groups ( , A univariate linear regression was performed on the data points, with the change in ambient temperature as the independent variable and the change in contact temperature as the dependent variable, to achieve a linear fit.

[0027] The slope of the resulting fitted line As a temperature sensitivity coefficient; the base temperature reference is a pre-set ambient temperature reference value. The specific temperature can be set according to the actual working conditions, and is not limited here. It is usually set to the standard room temperature of 25℃. Calculation of dynamic threshold for contact temperature: Obtain the rated allowable temperature rise, which is the maximum allowable temperature rise limit of the contact under rated load current; determine the load rate correction factor based on the real-time load rate; sum the product of the base temperature reference, the rated allowable temperature rise, and the load rate correction factor, along with the environmental compensation factor, and use the result as the dynamic threshold for contact temperature. The calculation formula is as follows:

[0028] in, This refers to the dynamic threshold temperature of the contact point. Based on the temperature reference, For the rated allowable temperature rise, This is the load factor correction factor. The ambient temperature compensation coefficient and the load rate correction coefficient can be determined by referring to the general model of the relationship between temperature rise and load rate in industry standards. ; in, and To meet the requirements under rated load rate The constraint constant is 1, which can be taken as a reference to industry standards. =1, The value is 0, and the real-time load factor is determined by the ratio of the effective value of the current to the rated current.

[0029] The reference vibration effective value is obtained by extracting the vibration effective value from the vibration signal collected by the vibration sensor when the rated load current is present and the branch box is in normal condition. Obtain the vibration load correction factor, which is determined by the square of the real-time load rate, i.e.:

[0030] in, LR is the load correction factor, and LR is the real-time load rate, representing the physical relationship that the amplitude of the alternating electrodynamic force on the conductor is proportional to the square of the current. The vibration dynamic threshold is obtained by multiplying the effective value of the reference vibration by a vibration load correction factor determined based on the real-time load rate, i.e.:

[0031] in, The vibration dynamic threshold, The reference vibration effective value, This is the load correction factor.

[0032] The overall failure degree is calculated by weighting and summing the temperature anomaly degree, vibration anomaly degree, current anomaly degree, and humidity anomaly degree respectively. The weight coefficients are determined by entropy weighting based on historical failure data. Specifically, the temperature anomaly degree is calculated as follows: when the contact temperature component in the comprehensive feature vector exceeds the dynamic threshold of the contact temperature under the current operating condition, the temperature anomaly degree is determined based on the ratio of the magnitude by which the contact temperature exceeds the dynamic threshold to the difference between the dynamic threshold and the baseline temperature. ; To calculate the vibration anomaly degree, when the effective vibration value component in the comprehensive feature vector exceeds the vibration dynamic threshold under the current operating condition, the vibration anomaly degree is determined based on the ratio of the amplitude of the effective vibration value exceeding the vibration dynamic threshold to the vibration dynamic threshold itself. ; The current anomaly is calculated by determining the load anomaly level based on the ratio of the magnitude of the load effective value exceeding the rated current threshold to the rated current threshold when the load effective value in the comprehensive characteristic vector exceeds the rated current threshold. ; The humidity anomaly is calculated by determining the humidity anomaly level based on the ratio of the magnitude of the humidity value exceeding the humidity threshold to the preset allowable humidity range when the humidity component in the comprehensive feature vector exceeds the preset allowable humidity range. ; The temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly are weighted and summed using their respective weighting coefficients to obtain the comprehensive fault degree:

[0033] in, , These are the weighting coefficients for temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly, respectively. , These are the rated current threshold and humidity threshold, respectively. , , , These are temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly. For overall failure rate; The weighting coefficients are determined based on each anomaly using the entropy weighting method. The steps of the entropy weighting method include: collecting multidimensional anomaly data at the time of the fault occurrence and several times before the fault, as well as anomaly data during normal operation; normalizing each anomaly index; calculating the information entropy of each anomaly index; and determining the entropy weight corresponding to each anomaly index as the weighting coefficient based on the proportion of the information entropy of each anomaly index in the sum of the information entropy of all anomaly indices.

[0034] During use, a fault alarm is triggered when the overall fault level exceeds a set threshold. After the alarm is triggered, the alarm type is determined based on the alarm duration and the trend of the overall fault level: if the duration is less than the first preset duration and the overall fault level falls below the preset safety value, it is determined to be an instantaneous fluctuation and the alarm is canceled; if the duration is between the first and second preset durations and the overall fault level shows an upward trend, it is determined to be a fault symptom; if the duration reaches or exceeds the second preset duration and the overall fault level remains higher than the preset fault confirmation value, it is determined to be a real fault. By adjusting the dynamic thresholds of contact temperature and vibration in real time based on the load status of the branch box, the threshold settings in the fault detection process are prevented from being fixed values ​​that cannot be dynamically adjusted according to the real-time operating scenario of the branch box, which could easily lead to false alarms or missed alarms. At the same time, the overall fault level is calculated by weighting and summing the temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly with their respective weight coefficients, thereby detecting the fault and preventing the data of each monitoring module from being isolated from each other, which would lead to the inability to cross-verify the fault through multi-dimensional data when the fault occurs.

[0035] Example 2

[0036] In practical use, it was found that status monitoring often adopts single-dimensional data collection, and the data of each monitoring module is isolated from each other and no correlation analysis mechanism is formed. This makes it impossible to cross-verify the cause and type of the fault through multi-dimensional data when a fault occurs. Further improvements are made based on the above embodiments.

[0037] Under normal operating conditions, obtain multiple associated feature values ​​from the associated feature vector of the branch box and use them as the reference associated feature values ​​respectively; When in use, when the comprehensive fault degree triggers a fault alarm, the associated feature vector in the comprehensive feature vector at the current moment is extracted; Compare multiple associated feature values ​​in the current associated feature vector with multiple benchmark associated feature values; The fault type is determined based on the degree to which the anomaly degree deviates from the baseline associated feature value from the current associated feature value. The associated feature vector includes at least a first associated feature value representing the correlation between load current and contact temperature, a second associated feature value representing the correlation between load current and vibration, a third associated feature value representing the correlation between load current and humidity, a fourth associated feature value representing the correlation between contact temperature and vibration, a fifth associated feature value representing the correlation between contact temperature and humidity, and a sixth associated feature value representing the correlation between vibration and humidity. Each associated feature value is taken when the absolute value of the corresponding correlation coefficient reaches the preset threshold, otherwise it is taken as zero. If the current anomaly is greater than zero and the temperature anomaly is greater than zero, and the first associated characteristic value is not zero, then it is determined to be a contact overheating fault caused by overload. If the temperature anomaly is greater than zero and the vibration anomaly is greater than zero, and the fourth associated characteristic value is not zero, then it is determined to be a local overheating and vibration fault caused by poor contact. If the temperature anomaly is greater than zero and the humidity anomaly is greater than zero, and the fifth associated characteristic value is not zero, then it is determined to be an insulation deterioration and heating fault caused by moisture. If the current anomaly degree is greater than zero and the vibration anomaly degree is greater than zero, and the second associated characteristic value is not zero, then it is determined to be an abnormal vibration fault induced by electromagnetic force. If only the humidity anomaly is greater than zero, and the current anomaly, temperature anomaly, and vibration anomaly are all zero, and the third and fifth associated characteristic values ​​are both zero, then the fault is determined to be due to excessively high ambient humidity. If multiple anomalies are greater than zero and multiple associated feature values ​​are not zero, it is determined to be a composite fault, and all single fault types that meet the conditions are output. The fault cause is verified by cross-validation of multiple anomalies and associated feature values, thereby quickly confirming the fault type and forming a multi-dimensional state monitoring and fault detection mechanism.

[0038] Example 3

[0039] Based on the above embodiments, this embodiment also provides a fault detection system for cable branch boxes, including: The data acquisition unit is used to collect multi-dimensional operating status data in real time through multiple sensors installed in the cable branch box; An environmental compensation unit is used to calculate the environmental temperature compensation coefficient. The dynamic threshold calculation unit is used to calculate the dynamic threshold of contact temperature and the dynamic threshold of vibration based on the real-time load rate and the ambient temperature compensation coefficient. The comprehensive failure degree analysis unit is used to compare the components of the comprehensive feature vector with the corresponding dynamic threshold or rated threshold to calculate the anomaly degree of each item, and then sum them up by weight to obtain the comprehensive failure degree. The verification feedback unit is used to determine the alarm type based on the alarm duration and the overall failure rate trend.

[0040] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A fault detection method for a cable branch box, characterized in that, Includes the following steps: S1. Real-time acquisition of multi-dimensional operating status data through multiple sensors installed in the cable branch box; S2. Extract features from the collected data to obtain feature sequences in multiple dimensions. Calculate the correlation coefficient matrix between the feature sequences in each dimension. When the absolute value of the correlation coefficient between any two dimensions exceeds a preset threshold, it is determined that there is a significant correlation between the two dimensions of data, and a comprehensive feature vector reflecting the operating status of the branch box is generated. S3. Based on the comprehensive feature vector and dynamic threshold mechanism, perform real-time fault detection and early warning, and determine the fault type; In step S2, the comprehensive feature vector is obtained in the following way: Extract the instantaneous values ​​of the feature sequences of each dimension at the current moment to form the basic feature vector; Extract the correlation coefficients between each dimension from the correlation coefficient matrix, and determine whether the absolute value of the correlation coefficient reaches the preset threshold. If it does, take the value of the correlation coefficient; otherwise, take zero as the correlation feature value. Arrange and combine multiple correlation feature values ​​into a correlation feature vector. The basic feature vector and the associated feature vector are concatenated to generate a comprehensive feature vector that reflects the operating status of the branch box; The associated feature vector includes at least a first associated feature value representing the correlation between load current and contact temperature, a second associated feature value representing the correlation between load current and vibration, a third associated feature value representing the correlation between load current and humidity, a fourth associated feature value representing the correlation between contact temperature and vibration, a fifth associated feature value representing the correlation between contact temperature and humidity, and a sixth associated feature value representing the correlation between vibration and humidity. Each associated feature value is taken when the absolute value of the corresponding correlation coefficient reaches the preset threshold, otherwise it is taken as zero. In step S3, fault detection is performed through the following steps: Based on the comparison of the contact temperature component, vibration effective value component, load current effective value component, and humidity value component in the comprehensive feature vector with the corresponding dynamic threshold or rated threshold, the temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly are calculated. If the current anomaly is greater than zero and the temperature anomaly is greater than zero, and the first associated characteristic value is not zero, then it is determined to be a contact overheating fault caused by overload. If the temperature anomaly is greater than zero and the vibration anomaly is greater than zero, and the fourth associated characteristic value is not zero, then it is determined to be a local overheating and vibration fault caused by poor contact. If the temperature anomaly is greater than zero and the humidity anomaly is greater than zero, and the fifth associated characteristic value is not zero, then it is determined to be an insulation deterioration and heating fault caused by moisture. If the current anomaly degree is greater than zero and the vibration anomaly degree is greater than zero, and the second associated characteristic value is not zero, then it is determined to be an abnormal vibration fault induced by electromagnetic force. If only the humidity anomaly is greater than zero, and the current anomaly, temperature anomaly, and vibration anomaly are all zero, and the third and fifth associated characteristic values ​​are both zero, then the fault is determined to be due to excessively high ambient humidity. If multiple anomalies are greater than zero and multiple associated feature values ​​are not zero, then it is determined to be a composite fault, and all single fault types that meet the conditions are output.

2. The fault detection method for cable branch boxes according to claim 1, characterized in that: In step S1, the sensor includes at least a current transformer, a temperature sensor, a vibration sensor, and a temperature and humidity sensor, and the multidimensional operating status data includes at least load current data, contact temperature data, vibration data, and humidity data.

3. The fault detection method for cable branch boxes according to claim 1, characterized in that: In step S2, the correlation coefficient matrix is ​​calculated in the following manner: Within a set sliding time window, the sample mean and standard deviation of each dimension feature sequence are calculated. For any two dimension feature sequences, the sum of the products of the deviations of each pair of sample values ​​from their respective means in the two dimension feature sequences is calculated, and the ratio of this product to the product of the sample standard deviations of the two dimension feature sequences is obtained. The correlation coefficient between the two dimensions is then formed by iterating through all dimension combinations.

4. The fault detection method for cable branch boxes according to claim 1, characterized in that: In step S3, the dynamic threshold mechanism includes: Calculate the ambient temperature compensation coefficient, which is obtained by multiplying the difference between the current ambient temperature and the baseline temperature by a temperature sensitivity coefficient. The temperature sensitivity coefficient is determined by the slope of the linear relationship between the change in ambient temperature and the change in contact temperature fitted by a steady-state temperature rise experiment. Calculate the dynamic threshold of the contact temperature, which is the sum of the product of the base temperature reference, the rated allowable temperature rise and the load rate correction factor, and the ambient temperature compensation factor; wherein the load rate correction factor is determined based on the real-time load rate. Obtain the reference vibration effective value and multiply it by the vibration load correction coefficient determined based on the real-time load rate to obtain the vibration dynamic threshold, wherein the vibration load correction coefficient is the square of the real-time load rate.

5. The fault detection method for cable branch boxes according to claim 4, characterized in that: In step S3, fault detection is performed through the following steps: The temperature anomaly, vibration anomaly, current anomaly, and humidity anomaly are weighted and summed with their respective weighting coefficients to obtain the comprehensive fault degree; wherein, the weighting coefficients are determined by the entropy weighting method based on the information entropy of each anomaly degree in historical fault data. A fault alarm is triggered when the overall failure rate exceeds a set threshold.

6. The fault detection method for cable branch boxes according to claim 5, characterized in that: The temperature anomaly degree is determined based on the ratio of the magnitude of the exceedance to the difference between the dynamic threshold and the baseline temperature reference when the contact temperature exceeds the dynamic threshold. The vibration anomaly degree is determined based on the ratio of the excess amplitude to the vibration dynamic threshold when the effective vibration value exceeds the vibration dynamic threshold. The current anomaly degree is determined by the ratio of the excess magnitude to the rated current threshold when the effective value of the load current exceeds the rated current threshold. The humidity anomaly is determined by the ratio of the magnitude of the exceedance to the humidity threshold when the humidity value exceeds the preset allowable humidity range.

7. The fault detection method for cable branch boxes according to claim 6, characterized in that: In step S3, the alarm type is determined as follows: If the duration is less than the first preset duration and the overall failure rate falls below the preset safety value, it is determined to be an instantaneous fluctuation and the alarm is canceled. If the duration is between the first preset duration and the second preset duration and the overall failure rate shows an upward trend, it is determined that there are signs of failure. If the duration reaches or exceeds the second preset duration and the overall failure rate continues to be higher than the preset failure confirmation value, it is determined to be a real failure.

8. The fault detection method for cable branch boxes according to claim 7, characterized in that: In S3, the fault type is determined in the following way: Extract the associated feature vector from the current time-integrated feature vector, and compare multiple associated feature values ​​with the baseline associated feature values ​​obtained under normal branch box operation conditions; The fault type is determined based on whether each anomaly degree is greater than zero and the degree to which the current associated feature value deviates from the baseline associated feature value.

9. A fault detection system for a cable branch box, employing the fault detection method for a cable branch box as described in claim 8, characterized in that, include: The data acquisition unit is used to collect multi-dimensional operating status data in real time through multiple sensors installed in the cable branch box; An environmental compensation unit is used to calculate the environmental temperature compensation coefficient. The dynamic threshold calculation unit is used to calculate the contact temperature dynamic threshold and vibration dynamic threshold based on the real-time load rate and the ambient temperature compensation coefficient. The comprehensive failure degree analysis unit is used to compare the components of the comprehensive feature vector with the corresponding dynamic threshold or rated threshold to calculate the anomaly degree of each item, and to obtain the comprehensive failure degree by weighted summation. The verification feedback unit is used to determine the alarm type based on the alarm duration and the overall failure rate trend.

Images