Distribution box fault detection method and device based on artificial intelligence

Abstract

The invention discloses a distribution box fault detection method and device based on artificial intelligence, and relates to the technical field of distribution box fault detection. The method comprises the steps of synchronously collecting grounding and zero-sequence current signals in periodic sampling of a power transmission line of a distribution box, calculating a root-mean-square value according to the grounding current signals, judging a fault by combining the root-mean-square value and a grounding current value at a certain moment, if the fault exists, storing the grounding current and the zero-sequence current into a fault data packet, and if the fault exists, storing the zero-sequence current into the fault data packet; the fault duration and type are determined by using information in the data packet, and meanwhile, the optimal delay amount is calculated through grounding and zero-sequence current data, so that accurate fault positioning is realized; by monitoring the grounding current and the zero-sequence current of the power transmission line of the distribution box, the fault is accurately judged, data are integrated, the duration time and the type of the fault are determined according to a root-mean-square value, the optimal delay amount is calculated by combining the grounding current and the zero-sequence current so as to accurately position the fault, the maintenance efficiency is improved, and economic losses caused by the fault are reduced.

Description

Technical Field

[0001] This invention belongs to the field of power distribution box fault detection technology, specifically relating to a power distribution box fault detection method and device based on artificial intelligence. Background Technology

[0002] A distribution box is a device for distributing and controlling electricity. It mainly includes circuit breakers, residual current devices (RCDs), and transmission lines, and has overload and short-circuit protection functions. In modern life, distribution boxes are widely used in homes, office buildings, factories, and other places to provide power distribution and protection for various electrical devices. It ensures a stable power supply. If a fault occurs and is not repaired in time, it can lead to power outages, equipment damage, increased safety hazards, and even serious accidents such as fires. Distribution boxes play a vital role in the power system and are key nodes in power transmission and distribution, ensuring the safe, reliable, and efficient use of electricity.

[0003] In large industrial plants, complex industrial environments (such as strong electromagnetic interference and mechanical vibration) can easily cause distribution boxes to malfunction. Strong electromagnetic interference can lead to high noise and poor quality of data collected by sensors, causing the current value collected by current sensors to jump instantaneously. Most existing distribution box fault detection technologies rely on the computing power of the detection model to analyze power data. If the input data quality is low, the output fault detection results will be inaccurate, making it difficult to accurately locate the fault area. Moreover, the detection model is difficult to detect the initial fault, which can easily develop into a permanent fault, reducing maintenance efficiency and seriously affecting the operation of the distribution box. Summary of the Invention

[0004] The purpose of this invention is to solve the problem of inaccurate fault detection and fault location of distribution boxes in complex industrial environments, and to propose a method and device for fault detection of distribution boxes based on artificial intelligence.

[0005] In a first aspect of this invention, a method for detecting faults in distribution boxes based on artificial intelligence is first proposed, the method comprising:

[0006] The grounding current signal and zero-sequence current signal of the transmission line in the distribution box are collected within the periodic time. The root mean square value of the grounding current is calculated based on the grounding current signal.

[0007] Fault detection is performed based on the root mean square value of the grounding current and the grounding current value at a certain moment. If a fault exists, the grounding current value and zero-sequence current value at that moment are stored in the fault data packet.

[0008] The fault duration is calculated based on the root mean square value of the grounding current, and the fault type is determined; the fault type includes initial faults and permanent faults.

[0009] The optimal delay between the ground current and the zero-sequence current is calculated using the ground current value and the zero-sequence current value in the fault data packet, and the fault is located based on the optimal delay.

[0010] Optionally, the root mean square value of the grounding current is calculated based on the grounding current signal, including:

[0011] Determine the current target sampling point m, and truncate forward from m. One continuous sampling point;

[0012] according to The root mean square value of the grounding current corresponding to each m is calculated from the given grounding current values. The process is as follows:

[0013]

[0014] in, This represents the root mean square value of the m-th target sampling point in the j-th line segment. This represents the grounding current value at the nth sampling point of the j-th section of the line.

[0015] Optionally, fault determination is performed based on the root mean square value of the grounding current and the grounding current, including:

[0016] The amplitude threshold, sudden change threshold, and instantaneous sudden change threshold are calculated using preset coefficients, grounding current, and the rated root mean square value of grounding current. The calculation process includes:

[0017]

[0018]

[0019]

[0020]

[0021] in, , and These are the amplitude threshold, the mutation threshold, and the instantaneous mutation threshold, respectively. , and All are preset coefficients. This is the rated root mean square value. Let m be the instantaneous change value at the m-th sampling point. The number of sampling points in one period;

[0022] The process involves determining the limits based on the root mean square value and the trigger threshold using three over-limit criteria:

[0023] like If so, an alert will be triggered;

[0024] like If so, an alert will be triggered;

[0025] like If so, an alert will be triggered;

[0026] If any over-limit criterion is triggered and a fault is detected, data recording is immediately initiated to input the root mean square value of the grounding current and the grounding current into the fault data packet.

[0027] Optionally, the fault duration is calculated and the fault type is determined based on the root mean square value of the grounding current, including:

[0028] The starting point is defined as the sampling point whose root mean square value corresponding to the current sampling point is higher than the trigger threshold for the first time and whose root mean square value of the previous sampling point is not higher than the trigger threshold.

[0029] The sampling point is defined as the point where the root mean square value of the current sampling point is lower than the trigger threshold for the first time and the root mean square value of the previous sampling point is not lower than the trigger threshold.

[0030] Calculate the fault duration period and the total number of sampling points within the fault duration period based on the times corresponding to the start and end points;

[0031] The initial fault sampling point threshold is calculated based on the device parameters, and the fault type is determined by comparing the total number of sampling points within the fault time period with the initial fault sampling point threshold.

[0032] Optionally, the optimal delay between the ground current and the zero-sequence current is calculated using the ground current value and the zero-sequence current value in the fault data packet, including:

[0033] Obtain the number of complete sampling points N within the fault period, calculate the cross-correlation value corresponding to each delay from 0 to N-1, and iterate to find the delay corresponding to the maximum cross-correlation value as the optimal delay.

[0034] Fault location is achieved by using preset delay thresholds and optimal delay values.

[0035] In a second aspect of this invention, an artificial intelligence-based fault detection device for distribution boxes is proposed, comprising a root mean square (RMS) calculation module, a fault judgment module, a type discrimination module, and a fault location module, wherein:

[0036] The root mean square (RMS) calculation module is used to collect the grounding current signal and zero-sequence current signal of the transmission line in the distribution box within a periodic time period, and calculate the root mean square value of the grounding current based on the grounding current signal.

[0037] The fault judgment module is used to perform fault judgment based on the root mean square value of the ground current and the ground current value at a certain moment. If a fault exists, the ground current value and zero-sequence current value at that moment are stored in the fault data packet.

[0038] The type discrimination module is used to calculate the fault duration and determine the fault type based on the root mean square value of the grounding current; the fault type includes initial fault and permanent fault.

[0039] The fault location module is used to calculate the optimal delay between the ground current and the zero-sequence current using the ground current value and the zero-sequence current value in the fault data packet, and to locate the fault based on the optimal delay.

[0040] Optionally, the root mean square calculation module includes a sampling module and a first calculation module, wherein:

[0041] The sampling module is used to determine the current target sampling point m, and to extract samples forward from m as the endpoint. One continuous sampling point;

[0042] The first calculation module is used to calculate based on The root mean square value of the grounding current corresponding to each m is calculated from the given grounding current values. The process is as follows:

[0043]

[0044] in, This represents the root mean square value of the m-th target sampling point in the j-th line segment. This represents the grounding current value at the nth sampling point of the j-th section of the line.

[0045] Optionally, the fault judgment module includes a trigger threshold calculation module, a criterion judgment module, and a recording module, wherein:

[0046] The trigger threshold calculation module is used to calculate the amplitude threshold, sudden change threshold, and instantaneous sudden change threshold using preset coefficients, grounding current, and the rated root mean square value of grounding current. The calculation process includes:

[0047]

[0048]

[0049]

[0050]

[0051] in, , and These are the amplitude threshold, the mutation threshold, and the instantaneous mutation threshold, respectively. , and All are preset coefficients. This is the rated root mean square value. Let m be the instantaneous change value at the m-th sampling point. The number of sampling points in one period;

[0052] The criterion judgment module is used to perform three over-limit criterion judgments based on the root mean square value and the trigger threshold. The process is as follows:

[0053] like If so, an alert will be triggered;

[0054] like If so, an alert will be triggered;

[0055] like If so, an alert will be triggered;

[0056] The recording module is used to immediately start data recording and input the root mean square value of the grounding current and the grounding current into the fault data packet if any over-limit criterion is triggered and a fault is determined.

[0057] Optionally, the type discrimination module includes a start point module, an end point module, a second calculation module, and a type judgment module, wherein:

[0058] The starting point module is used to define a sampling point as a starting point when the root mean square value of the current sampling point is higher than the trigger threshold for the first time and the root mean square value of the previous sampling point is not higher than the trigger threshold.

[0059] The termination point module is used to define a sampling point as a termination point when the root mean square value of the current sampling point is lower than the trigger threshold for the first time and the root mean square value of the previous sampling point is not lower than the trigger threshold.

[0060] The second calculation module is used to calculate the fault duration period and the total number of sampling points within the fault duration period based on the time corresponding to the start point and the end point;

[0061] The type determination module is used to calculate the initial fault sampling point threshold based on the device parameters, and to determine the fault type based on the total number of sampling points within the fault time period and the initial fault sampling point threshold.

[0062] Optionally, the fault location module includes a delay module and a location module, wherein:

[0063] The delay module is used to obtain the number of complete sampling points N within the fault time period, calculate the cross-correlation value corresponding to each delay from 0 to N-1, and iterate to find the delay corresponding to the maximum cross-correlation value as the optimal delay.

[0064] The positioning module is used to locate faults by using a preset delay threshold and an optimal delay amount.

[0065] The beneficial effects of this invention are:

[0066] This invention proposes an artificial intelligence-based method for detecting faults in distribution boxes. By monitoring the grounding current and zero-sequence current of the power transmission line in the distribution box in real time, the root mean square value of the grounding current is calculated, the fault is identified and the data is integrated. Based on the root mean square value, the duration and type of the fault are determined. The optimal delay is calculated by combining the grounding current and zero-sequence current to accurately locate the fault, improve maintenance efficiency, and reduce the impact of power outages. Attached Figure Description

[0067] The invention will now be further described with reference to the accompanying drawings.

[0068] Figure 1 A flowchart illustrating an artificial intelligence-based fault detection method for distribution boxes, provided as an embodiment of the present invention;

[0069] Figure 2 This is a framework diagram of an artificial intelligence-based power distribution box fault detection device provided in an embodiment of the present invention. Detailed Implementation

[0070] 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. The term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and B can represent: A alone, A and B simultaneously, and B alone. Furthermore, descriptions involving "first," "second," etc., in this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" can explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0071] Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0072] This invention provides an artificial intelligence-based method for detecting faults in distribution boxes. See also... Figure 1 , Figure 1 A flowchart illustrating an artificial intelligence-based fault detection method for distribution boxes, provided as an embodiment of the present invention. The method includes the following steps:

[0073] S101, collects the grounding current signal and zero-sequence current signal of the transmission line in the distribution box within the periodic time, and calculates the root mean square value of the grounding current based on the grounding current signal;

[0074] S102, perform fault judgment based on the root mean square value of the grounding current and the grounding current value at a certain moment. If a fault exists, store the grounding current value and zero-sequence current value at that moment into the fault data packet.

[0075] S103, calculate the fault duration and determine the fault type based on the root mean square value of the grounding current;

[0076] S104 calculates the optimal delay between the ground current and the zero-sequence current using the ground current value and the zero-sequence current value in the fault data packet, and locates the fault based on the optimal delay.

[0077] Fault types include initial faults and permanent faults.

[0078] The present invention provides an artificial intelligence-based method for detecting faults in distribution boxes. By monitoring the grounding current and zero-sequence current of the power transmission line in the distribution box in real time, the root mean square value of the grounding current is calculated, the fault is identified and the data is integrated. Based on the root mean square value, the duration and type of the fault are determined. The optimal delay is calculated by combining the grounding and zero-sequence currents to accurately locate the fault, improve maintenance efficiency, and reduce the impact of power outages.

[0079] In one implementation, the grounding current and the zero-sequence current are collected synchronously. Since the physical characteristics and frequency bands of the grounding current and the zero-sequence current are different, the grounding current is collected through a high-frequency current transformer, while the zero-sequence current is collected through a zero-sequence current transformer.

[0080] In one implementation, the power transmission line in the distribution box is divided into a preset number of sections according to its function. The current characteristics of each section are calculated based on the grounding current and zero-sequence current to detect the fault and more accurately locate the section where the fault occurred.

[0081] In one implementation, the initial fault is a transitional stage before the permanent fault occurs. Both originate from abnormal damage to the insulation layer of the transmission line. The initial fault manifests as instantaneous and repeated arc breakdown of the transmission line core. If it is not detected and intervened in time, this repeated breakdown will continue to damage the insulation layer and may eventually develop into a permanent fault that cannot be self-recovered.

[0082] In one implementation, the initial fault has self-clearing characteristics, a short duration, and the voltage and current will return to normal after the fault occurs. Its grounding current only increases briefly and does not trigger traditional protection actions. Therefore, it needs to be identified by the periodic root mean square value of the grounding current, the instantaneous change, and the cross-correlation characteristics with the zero-sequence current. On the other hand, the permanent fault has no self-clearing capability, a long duration, and the fault current persists and will trigger traditional protection actions. Its current characteristics can be captured by conventional protection devices without the need for complex high-frequency monitoring.

[0083] In one embodiment, calculating the root mean square value of the ground current based on the ground current signal includes:

[0084] Determine the current target sampling point m, and truncate forward from m. One continuous sampling point;

[0085] according to The root mean square value of the grounding current corresponding to each m is calculated from the given grounding current values. The process is as follows:

[0086]

[0087] in, This represents the root mean square value of the m-th target sampling point in the j-th line segment. This represents the grounding current value at the nth sampling point of the j-th section of the line.

[0088] In one implementation, the local features of the grounding current are effectively extracted by calculating the root mean square value of the sliding window of continuous sampling points, which smooths noise, captures abnormal fluctuations, and improves the accuracy of fault diagnosis.

[0089] In one embodiment, fault determination is performed based on the root mean square value of the ground current and the ground current value at a certain moment, including:

[0090] The amplitude threshold, sudden change threshold, and instantaneous sudden change threshold are calculated using preset coefficients, grounding current, and the rated root mean square value of grounding current. The calculation process includes:

[0091]

[0092]

[0093]

[0094]

[0095] in, , and These are the amplitude threshold, the mutation threshold, and the instantaneous mutation threshold, respectively. , and All are preset coefficients. This is the rated root mean square value. Let m be the instantaneous change value at the m-th sampling point. The number of sampling points in one period;

[0096] The process involves determining exceedances based on the root mean square value and the trigger threshold using three criteria:

[0097] like If so, an alert will be triggered;

[0098] like If so, an alert will be triggered;

[0099] like If so, an alert will be triggered;

[0100] If any over-limit criterion is triggered and a fault is detected, data recording is immediately initiated to input the root mean square value of the grounding current and the grounding current into the fault data packet.

[0101] In one implementation, the coefficient is set through a distribution box detection experiment. By calculating three trigger thresholds—amplitude, sudden change, and instantaneous sudden change—the root mean square value of the grounding current is compared in real time. Once the limit is exceeded, the fault data is recorded immediately. This can accurately capture different abnormal characteristics, effectively reduce false alarms and missed alarms, and improve the reliability and real-time performance of fault diagnosis.

[0102] In one embodiment, calculating the fault duration and determining the fault type based on the root mean square value of the ground current includes:

[0103] The starting point is defined as the sampling point whose root mean square value corresponding to the current sampling point is higher than the trigger threshold for the first time and whose root mean square value of the previous sampling point is not higher than the trigger threshold.

[0104] The sampling point is defined as the point where the root mean square value of the current sampling point is lower than the trigger threshold for the first time and the root mean square value of the previous sampling point is not lower than the trigger threshold.

[0105] Calculate the fault duration period and the total number of sampling points within the fault duration period based on the times corresponding to the start and end points;

[0106] The initial fault sampling point threshold is calculated based on the device parameters, and the fault type is determined by comparing the total number of sampling points within the fault time period with the initial fault sampling point threshold.

[0107] In one implementation, the fault duration period and the total number of sampling points within the fault duration period are calculated, and the fault type is determined, including:

[0108]

[0109]

[0110]

[0111] in, This represents the total number of sampling points during the fault duration. The initial fault sampling point threshold, Fault duration period, As the endpoint, Starting point End point The corresponding time, End point The corresponding time, Indicates the sampling frequency of the monitoring device. Let 50Hz be the fundamental frequency of the system.

[0112] like If the transmission line is faulty, then the fault is the initial fault. If so, the transmission line is considered to have a permanent fault.

[0113] In one implementation, by accurately capturing the start and end points, the fault duration is quickly calculated and the number of sampling points is counted. The fault is then classified in combination with the initial fault sampling point threshold, which effectively improves the fault detection accuracy and automation level, and increases the response and processing speed.

[0114] In one embodiment, the optimal delay between the ground current and the zero-sequence current is calculated using the ground current and the zero-sequence current, and fault location is performed based on the optimal delay, including:

[0115] Obtain the number of complete sampling points N within the fault period, calculate the cross-correlation value corresponding to each delay from 0 to N-1, and iterate to find the delay corresponding to the maximum cross-correlation value as the optimal delay.

[0116] Fault location is achieved by using preset delay thresholds and optimal delay values.

[0117] In one implementation, the delay threshold is pre-set in a database based on experimental data, and the cross-correlation value corresponding to each delay value is calculated. The process is as follows:

[0118]

[0119] in, Indicates the delay amount as The cross-correlation value between zero-sequence current and grounding current. This represents the zero-sequence current value at the nth sampling point. This represents the delay of the current signal;

[0120] Traversal The delay corresponding to the maximum cross-correlation value As the optimal delay ,like If so, then the section is determined to be a faulty section.

[0121] In one implementation, the optimal delay amount is determined by calculating the cross-correlation value of each delay amount within the fault time period, and the fault location is achieved by combining it with a preset delay threshold, thereby improving the location accuracy.

[0122] Based on the same inventive concept, this invention also provides an artificial intelligence-based fault detection device for distribution boxes. See also... Figure 2 , Figure 2 This is a framework diagram of an artificial intelligence-based distribution box fault detection device provided in an embodiment of the present invention. The device includes a root mean square (RMS) calculation module, a fault judgment module, a type discrimination module, and a fault location module, wherein:

[0123] The root mean square (RMS) calculation module is used to collect the grounding current signal and zero-sequence current signal of the transmission line in the distribution box within a periodic time period, and calculate the root mean square value of the grounding current based on the grounding current signal.

[0124] The fault judgment module is used to perform fault judgment based on the root mean square value of the ground current and the ground current value at a certain moment. If a fault exists, the ground current value and zero-sequence current value at that moment are stored in the fault data packet.

[0125] The fault type determination module is used to calculate the fault duration and determine the fault type based on the root mean square value of the grounding current; the fault type includes initial fault and permanent fault.

[0126] The fault location module is used to calculate the optimal delay between the ground current and the zero-sequence current using the ground current value and the zero-sequence current value in the fault data packet, and to locate the fault based on the optimal delay.

[0127] The present invention provides an artificial intelligence-based distribution box fault detection device that monitors the grounding current and zero-sequence current of the distribution box transmission line in real time, calculates the root mean square value of the grounding current, judges the fault and integrates the data, determines the fault duration and type based on the root mean square value, and calculates the optimal delay amount by combining the grounding and zero-sequence currents to accurately locate the fault, improve maintenance efficiency and reduce the impact of power outages.

[0128] In one embodiment, the root mean square (RMS) calculation module includes a sampling module and a first calculation module, wherein:

[0129] The sampling module is used to determine the current target sampling point m, and to extract samples forward from m as the endpoint. One continuous sampling point;

[0130] The first calculation module is used to calculate based on... The root mean square value of the grounding current corresponding to each m is calculated from the given grounding current values. The process is as follows:

[0131]

[0132] in, This represents the root mean square value of the m-th target sampling point in the j-th line segment. This represents the grounding current value at the nth sampling point of the j-th section of the line.

[0133] In one embodiment, the fault determination module includes a trigger threshold calculation module, a criterion determination module, and a recording module, wherein:

[0134] The trigger threshold calculation module is used to calculate the amplitude threshold, sudden change threshold, and instantaneous sudden change threshold using preset coefficients, grounding current, and the rated root mean square value of grounding current. The calculation process includes:

[0135]

[0136]

[0137]

[0138]

[0139] in, , and These are the amplitude threshold, the mutation threshold, and the instantaneous mutation threshold, respectively. , and All are preset coefficients. This is the rated root mean square value. Let m be the instantaneous change value at the m-th sampling point. The number of sampling points in one period;

[0140] The criterion judgment module is used to judge three over-limit criteria based on the root mean square value and the trigger threshold. The process is as follows:

[0141] like If so, an alert will be triggered;

[0142] like If so, an alert will be triggered;

[0143] like If so, an alert will be triggered;

[0144] The recording module is used to immediately start data recording and input the root mean square value of the grounding current and the grounding current into the fault data packet if any over-limit criterion is triggered and a fault is determined.

[0145] In one embodiment, the type determination module includes a start point module, an end point module, a second calculation module, and a type judgment module, wherein:

[0146] The starting point module is used to define a sampling point as the starting point when the root mean square value of the current sampling point is higher than the trigger threshold for the first time and the root mean square value of the previous sampling point is not higher than the trigger threshold.

[0147] The termination point module is used to define a sampling point as a termination point when the root mean square value of the current sampling point is lower than the trigger threshold for the first time and the root mean square value of the previous sampling point is not lower than the trigger threshold.

[0148] The second calculation module is used to calculate the fault duration period and the total number of sampling points within the fault duration period based on the time corresponding to the start point and the end point.

[0149] The type determination module is used to calculate the initial fault sampling point threshold based on the device parameters, and to determine the fault type based on the total number of sampling points within the fault time period and the initial fault sampling point threshold.

[0150] In one embodiment, the fault location module includes a delay quantity module and a location module, wherein:

[0151] The delay module is used to obtain the number of complete sampling points N within the fault time period, calculate the cross-correlation value corresponding to each delay from 0 to N-1, and iterate to find the delay corresponding to the maximum cross-correlation value as the optimal delay.

[0152] The location module is used to locate faults by using a preset delay threshold and an optimal delay amount.

[0153] The foregoing has described one embodiment of the present invention in detail, but this content is merely a preferred embodiment and should not be considered as limiting the scope of the present invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the scope of the claims of this invention.

Claims

1. An artificial intelligence-based power distribution box fault detection method, characterized by, The method comprises: Collecting the grounding current signal and the zero sequence current signal of the power transmission line in the distribution box within a collection period, and calculating the grounding current root mean square value according to the grounding current signal; Performing fault judgment according to the grounding current root mean square value and the grounding current value at a certain moment, and if there is a fault, storing the grounding current value and the zero sequence current value at the moment into a fault data packet; Calculating the fault duration according to the grounding current root mean square value and identifying the fault type; the fault type includes initial fault and permanent fault; Calculating the optimal delay amount of the grounding current and the zero sequence current according to the grounding current value and the zero sequence current value in the fault data packet, and performing fault positioning according to the optimal delay amount.

2. The method for detecting faults in an electrical distribution panel based on artificial intelligence as claimed in claim 1, wherein, The grounding current root mean square value is calculated according to the grounding current signal, which comprises: determining a current target sample point m, intercepting forwardly with m as the terminal point one continuous sample point; According to The ground current root mean square value corresponding to each m is calculated according to the ground current value, and the process is: wherein, RMS value of the jth section line at the mth target sampling point, the ground current value of the jth section line at the nth sampling point.

3. The method of claim 1, wherein the method further comprises: The fault judgment is performed according to the grounding current root mean square value and the grounding current value at a certain moment, which comprises: The amplitude threshold value, the mutation threshold value and the instantaneous mutation threshold value are calculated according to the preset coefficient, the grounding current and the rated root mean square value of the grounding current, and the calculation process comprises: wherein, , and are an amplitude threshold value, a sudden change threshold value and an instantaneous sudden change threshold value, respectively, , and are all preset coefficients, is a rated root mean square value, is an instantaneous sudden change value of the mth sampling point, is the number of sampling points in one period; The three over-limit criteria are judged according to the root mean square value and the trigger threshold value, and the process is: If then a warning is triggered; If then a warning is triggered; If then a warning is triggered; If any over-limit criterion is triggered to judge that there is a fault, the grounding current root mean square value and the grounding current are immediately input into the fault data packet by starting data recording.

4. The method of claim 1, wherein the method further comprises: The fault duration is calculated according to the grounding current root mean square value, and the fault type is identified, which comprises: The starting point is defined as the sampling point at which the root mean square value corresponding to the current sampling point is first higher than the trigger threshold value and the root mean square value of the previous sampling point is not higher than the trigger threshold value; The ending point is defined as the sampling point at which the root mean square value corresponding to the current sampling point is first lower than the trigger threshold value and the root mean square value of the previous sampling point is not lower than the trigger threshold value; The fault duration period and the total sampling point number in the fault duration period are calculated according to the time corresponding to the starting point and the ending point; The initial fault sampling point threshold value is calculated according to the device parameters, and the fault type is obtained by judging the total sampling point number in the fault duration period and the initial fault sampling point threshold value.

5. The method of claim 1, wherein the method further comprises: The optimal delay amount of the grounding current and the zero sequence current is calculated according to the grounding current value and the zero sequence current value in the fault data packet, which comprises: The complete sampling point number N in the fault duration period is obtained, the cross-correlation value corresponding to each delay amount is calculated from 0 to N-1, and the delay amount corresponding to the maximum cross-correlation value is obtained as the optimal delay amount by traversing; Fault positioning is performed by the preset delay threshold value and the optimal delay amount.

6. An artificial intelligence based electrical distribution box fault detection device, characterized by, The device comprises a root mean square calculation module, a fault judgment module, a type identification module and a fault positioning module, wherein: The root mean square calculation module is used for collecting the grounding current signal and the zero sequence current signal of the power transmission line in the distribution box within a collection period, and calculating the grounding current root mean square value according to the grounding current signal; The fault judgment module is used for performing fault judgment according to the grounding current root mean square value and the grounding current value at a certain moment, and if there is a fault, storing the grounding current value and the zero sequence current value at the moment into a fault data packet; The type identification module is used for calculating the fault duration according to the grounding current root mean square value and identifying the fault type; the fault type includes initial fault and permanent fault; The fault positioning module is used for calculating the optimal delay amount of the grounding current and the zero sequence current according to the grounding current value and the zero sequence current value in the fault data packet, and performing fault positioning according to the optimal delay amount. The fault positioning module is configured to calculate an optimal delay amount of the ground current and the zero sequence current according to the ground current value and the zero sequence current value in the fault data packet, and perform fault positioning according to the optimal delay amount.

7. The power distribution box fault detection apparatus based on artificial intelligence according to claim 6, wherein, The mean square root calculation module comprises a sampling module and a first calculation module, wherein: The sampling module is configured to determine a current target sampling point m, and intercept a plurality of continuous sampling points with m as a terminal point. a plurality of continuous sampling points. The first calculation module is configured to calculate the ground current root mean square value corresponding to each m according to The ground current root mean square value corresponding to each m is calculated according to the following process: wherein, RMS value of the jth section line at the mth target sampling point, Ground current value of the jth section line at the nth sampling point.

8. The power distribution box fault detection apparatus based on artificial intelligence according to claim 6, wherein, The fault judgment module comprises a trigger threshold calculation module, a criterion judgment module and a recording module, wherein: The trigger threshold calculation module is configured to calculate an amplitude threshold, a mutation threshold and an instantaneous mutation threshold according to a preset coefficient, the ground current and a rated mean square root value of the ground current, and the calculation process comprises: wherein, , and are an amplitude threshold value, a mutation threshold value and an instantaneous mutation threshold value, respectively, , and are preset coefficients, is a rated root mean square value, is an instantaneous mutation value of the mth sampling point, is a number of sampling points in one period. The criterion judgment module is configured to perform three over-limit criterion judgments according to the mean square root value and the trigger threshold, and the process is: If then a warning is triggered; If then a warning is triggered; If then a warning is triggered; The recording module is configured to start data recording to input the ground current mean square root value and the ground current into a fault data packet if any over-limit criterion is triggered to judge that there is a fault.

9. The power distribution box fault detection apparatus based on artificial intelligence according to claim 6, wherein, The type discrimination module comprises a starting point module, a terminal point module, a second calculation module and a type judgment module, wherein: The starting point module is configured to define a sampling point as a starting point if the mean square root value corresponding to the current sampling point is higher than the trigger threshold for the first time and the mean square root value of the previous sampling point is not higher than the trigger threshold; The terminal point module is configured to define a sampling point as a terminal point if the mean square root value corresponding to the current sampling point is lower than the trigger threshold for the first time and the mean square root value of the previous sampling point is not lower than the trigger threshold; The second calculation module is configured to calculate a fault duration period and a total sampling point number in the fault duration period according to the time corresponding to the starting point and the terminal point; The type judgment module is configured to calculate an initial fault sampling point threshold according to device parameters, and perform judgment according to the total sampling point number in the fault duration period and the initial fault sampling point threshold to obtain a fault type.

10. The power distribution box fault detection apparatus based on artificial intelligence according to claim 6, wherein, The fault positioning module comprises a delay amount module and a positioning module, wherein: The delay amount module is configured to obtain a complete sampling point number N in the fault duration period, calculate a cross-correlation value corresponding to each delay amount from 0 to N-1, and traverse the delay amount corresponding to the maximum cross-correlation value as an optimal delay amount; The positioning module is configured to perform fault positioning according to a preset delay threshold and the optimal delay amount.

Images