A high-voltage cable fault diagnosis method and system based on leakage current monitoring

Abstract

The present invention relates to a high-voltage cable fault diagnosis method and system based on leakage current monitoring. The method includes collecting current induction information of the high-voltage cable, determining the total circuit current and the leakage current, and judging whether the system state is an overload fault based on a preset rule; when it is not an overload fault, judging whether the leakage current is within a preset range. If so, determining that the system state is normal; otherwise, performing data classification processing on the leakage current by a multi-class support vector machine (MSVM) to determine that the system fault is a resistive fault and / or a capacitive fault. First, determine whether the system is overloaded by determining the total circuit current and the leakage current through the current induction information, and then perform data classification processing on the leakage current by the multi-class support vector machine (MSVM) to obtain the corresponding resistive leakage current and capacitive leakage current, so as to accurately determine that the system fault is a resistive fault and / or a capacitive fault, realizing accurate classification of the fault and greatly improving the accuracy of high-voltage cable fault diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of high - voltage cable detection, and particularly to a high - voltage cable fault diagnosis method and system based on leakage current monitoring. Background Art

[0002] The condition monitoring and fault diagnosis of electrical equipment are still in the research and development stage. There are a few application cases and solutions, but none of them can effectively solve the problems. For example, some monitoring systems based on ZigBee or WIFI do not consider the key electrical parameters for diagnosing the electrical health status, such as leakage current. The lack of standards for some live tests (such as partial discharge, etc.) leads to different testing equipment using different technologies and diagnostic criteria, and the results are difficult to verify. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a high - voltage cable fault diagnosis method and system based on leakage current monitoring for the deficiencies of the above - mentioned existing technologies.

[0004] The technical solution of the present invention to solve the above - mentioned technical problem is as follows: A high - voltage cable fault diagnosis method based on leakage current monitoring includes the following steps:

[0005] S1: Collect the current induction information of the high - voltage cable through multiple sensors, determine the total circuit current and leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule;

[0006] S2: When the system state is not an overload fault, judge whether the leakage current is within a preset range. If so, determine that the system state is normal; otherwise, enter S3;

[0007] S3: Perform data classification processing on the leakage current based on the multi - classification support vector machine MSVM, calculate the resistive leakage current and capacitive leakage current, and determine that the system fault is a resistive fault and / or capacitive fault according to the resistive leakage current and capacitive leakage current.

[0008] The beneficial effect of the present invention is that the high - voltage cable fault diagnosis method based on leakage current monitoring of the present invention first determines whether the system is overloaded by determining the total circuit current and leakage current through current induction information. On this basis, perform data classification processing on the leakage current based on the multi - classification support vector machine MSVM to obtain the corresponding resistive leakage current and capacitive leakage current, and combine the corresponding preset range, so as to accurately determine that the system fault is a resistive fault and / or capacitive fault, realizing accurate classification of faults and greatly improving the accuracy of high - voltage cable fault diagnosis.

[0009] On the basis of the above - mentioned technical solution, the present invention can also be improved as follows:

[0010] Further: in S1, the specific steps of determining the total current and leakage current in the circuit according to the current sensing data are as follows:

[0011] For N lines, the total power of the system is:

[0012]

[0013] where P i and Q i represent the active power and reactive power of the i-th line;

[0014] Then the total current I T (t) flowing into the load is:

[0015] I T (t) = I z,r (t) + jI X,l,c (t)

[0016] I X,l,c (t) = I X,l (t) - I X,c (t)

[0017] where I X,l (t) and I X,c (t) represent the inductive and capacitive currents of the load respectively; I Z,r (t) = I T (t) cosθ I , represents the resistive current flowing into the circuit, I X,l,c (t) = I T (t) sinθ I , represents the inductive current flowing into the circuit, θ I represents the power angle under normal operating conditions;

[0018] The leakage current is:

[0019] I L (t) = I rl (t) + jI cl (t)

[0020] where I rl (t) = I L (t) cosθ L , represents the resistive leakage current, I cl (t) = I L (t) sinθ L represents the capacitive leakage current; and θ I represents the phase angle between the capacitive leakage current I cl (t) and the resistive leakage current I rl (t).

[0021] The beneficial effect of the above further solution is that the total current in the circuit can be accurately calculated through the inductive current and capacitive current flowing into the circuit, and then, in combination with the capacitive leakage current I cl (t) and the resistive leakage current I rl (t), the values of the capacitive leakage current I cl (t) and the resistive leakage current I rl (t) can be accurately calculated, so as to facilitate judging whether the system state has an overload fault based on a preset rule.

[0022] Further: in S1, the specific steps of judging whether the system state has an overload fault based on a preset rule are as follows:

[0023] S11: If the current sensor does not detect an induced current, the system state is a sensor fault; otherwise, go to S12;

[0024] S12: Judge whether the total current in the circuit is within a preset range. If so, determine that the system state is normal; otherwise, determine that the system state is overloaded.

[0025] The beneficial effect of the above further solution is that it can be judged whether the sensor has a fault by the fact that the current sensor does not detect an induced current, and when the current sensor detects an induced current, judge whether the total current in the current is within a preset range. If it exceeds the preset range, it can be determined that the system state is overloaded; otherwise, it is still necessary to further determine whether there are other faults in the system according to the leakage current, forming a complete fault diagnosis logic with high accuracy.

[0026] Further: in S3, the specific steps of performing data classification processing on the leakage current based on the multi-class support vector machine MSVM and calculating the resistive leakage current and the capacitive leakage current are as follows:

[0027] S31: Convert the leakage current into a scatter plot, and create a hyperplane based on the scatter plot and a preset training sample;

[0028] S32: Extract the point data features corresponding to the leakage current based on the multi-class support vector machine MSVM, and optimize the hyperplane based on the given training sample to obtain an optimal hyperplane; wherein, the optimal hyperplane is used as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data;

[0029] S33: Perform summation calculations on the resistive data and the capacitive data respectively to obtain the resistive leakage current and the capacitive leakage current.

[0030] The beneficial effects of the above further solution are as follows: By converting the leakage current into a scatter plot, constructing a hyperplane, and focusing on the characteristic special zone of the point data corresponding to the leakage current, and then optimizing the hyperplane, the points corresponding to the leakage current can be divided into resistive data and capacitive data, so that the resistive leakage current and the capacitive leakage current can be accurately calculated, facilitating the subsequent determination of whether resistive faults and capacitive faults occur.

[0031] Further: In S31, the steps of converting the leakage current into a scatter plot and creating a hyperplane based on the scatter plot and a preset training sample are specifically as follows:

[0032] For a given training sample {(x i ,y i )}, where y i ∈{1, -1} represents the class label, and create a hyperplane with the expression:

[0033] θ T x i +b = 0

[0034] where θ = [θ1, …, θ n represents the n-dimensional weight vector, x i = [x1, …, x n represents the n-dimensional input vector, b represents the bias unit, and n represents the number of features.

[0035] The beneficial effects of the above further solution are as follows: By using the given training sample to create a hyperplane, it is convenient to divide the point data in the scatter plot into two categories, and after optimizing the hyperplane, the corresponding resistive point data and capacitive point data can be accurately obtained.

[0036] Further: In S32, the steps of extracting the point data features corresponding to the leakage current based on the multi-class support vector machine MSVM and optimizing the hyperplane based on the given training sample to obtain the optimal hyperplane are specifically as follows:

[0037] S321: Construct a non-linear decision function f(x):

[0038]

[0039] where sign() represents the sign function, and α i represents the coefficient of the hyperplane expression corresponding to each sample point;

[0040] S322: Call a preset number of kernel functions k(x i, x) Calculate the maximum value of the non - linear decision function f(x), and determine the weight vector θ corresponding to the leakage current according to the kernel function k(xi, x) corresponding to the maximum value among the maximum values of the non - linear decision function f(x);

[0041] S323: Calculate the sum of the distances from all points corresponding to the leakage current to the hyperplane according to the weight vector θ, and determine the minimum value of the sum of the distances:

[0042]

[0043] Where n represents the number of features. For the measured current sample points, they can be transformed into 2 - D samples including real and imaginary parts through phasor transformation. At this time, n = 2;

[0044] S324: Determine the hyperplane corresponding to the minimum value of the sum of the distances as the optimal hyperplane.

[0045] The beneficial effect of the above - mentioned further solution is: By constructing the non - linear decision function f(x) and calculating the maximum value of the non - linear decision function f(x) used to characterize the confidence level of the calculation result based on a preset plurality of kernel functions k(x i , x), the weight vector θ corresponding to the leakage current can be determined according to the maximum value among the maximum values of all non - linear decision functions f(x), so as to accurately calculate the sum of the distances from all points to the hyperplane, and then accurately determine the optimal hyperplane, realizing the accurate classification of the point data corresponding to the leakage current.

[0046] The present invention also provides a high - voltage cable fault diagnosis system based on leakage current monitoring, including a sensor component, a first calculation and judgment module, and a second calculation and judgment module;

[0047] The sensor component is used to collect the current induction information of the high - voltage cable;

[0048] The first calculation and judgment module is used to determine the total circuit current and the leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule;

[0049] The second calculation and judgment module is used to, when the system state is not an overload fault, judge whether the leakage current is within a preset range. If so, determine that the system state is normal; otherwise, perform data classification processing on the leakage current based on the multi - classification support vector machine MSVM, calculate the resistive leakage current and the capacitive leakage current, and determine that the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and the capacitive leakage current.

[0050] The high-voltage cable fault diagnosis system based on leakage current monitoring of the present invention first determines the total current and leakage current of the circuit through current induction information to determine whether the system is overloaded. On this basis, the leakage current is processed by a multi-class support vector machine (MSVM) for data classification to obtain the corresponding resistive leakage current and capacitive leakage current, and combined with the corresponding preset ranges, so as to accurately determine that the system fault is a resistive fault and / or a capacitive fault, realizing accurate classification of the fault and greatly improving the accuracy of high-voltage cable fault diagnosis.

[0051] On the basis of the above technical solution, the present invention can also be improved as follows:

[0052] Further: The specific implementation of the first calculation and judgment module for judging whether the system state has an overload fault based on a preset rule is:

[0053] If the current sensor does not detect the induced current, the system state is a sensor fault. Otherwise, it is judged whether the total current in the circuit is within the preset range. If so, the system state is determined to be normal. Otherwise, the system state is determined to be overloaded;

[0054] The specific implementation of the second calculation and judgment module for processing the leakage current by a multi-class support vector machine (MSVM) and calculating the resistive leakage current and capacitive leakage current is:

[0055] Convert the leakage current into a scatter plot, and create a hyperplane based on the scatter plot and the preset training samples;

[0056] Extract the corresponding point data features of the leakage current based on the multi-class support vector machine (MSVM), and optimize the hyperplane based on the given training samples to obtain the optimal hyperplane; wherein, the optimal hyperplane is used as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data;

[0057] Perform summation calculations on the resistive data and capacitive data respectively to obtain the resistive leakage current and capacitive leakage current.

[0058] The beneficial effects of the above further solution are as follows: It can be determined whether the sensor fails by the fact that the current sensor does not detect the induced current. When the current sensor detects the induced current, it is judged whether the total current in the current is within the preset range. If it exceeds the preset range, the system state can be determined to be overloaded. Otherwise, it is necessary to further determine whether there are other faults in the system according to the leakage current, forming a complete fault diagnosis logic with high accuracy. By converting the leakage current into a scatter plot, constructing a hyperplane, and performing special processing on the point data features corresponding to the leakage current, and then optimizing the hyperplane, the points corresponding to the leakage current can be divided into resistive data and capacitive data, so that the resistive leakage current and capacitive leakage current can be accurately calculated, which is convenient for subsequent determination of whether resistive faults and capacitive faults occur.

[0059] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the following steps:

[0060] S1: Determine the total circuit current and the leakage current by collecting the current induction information of the high-voltage cable through multiple sensors, and judge whether the system state is an overload fault based on a preset rule;

[0061] S2: When the system state is not an overload fault, judge whether the leakage current is within the preset range. If so, determine that the system state is normal; otherwise, enter S3;

[0062] S3: Perform data classification processing on the leakage current based on the multi-class support vector machine MSVM, calculate the resistive leakage current and the capacitive leakage current, and determine that the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and the capacitive leakage current.

[0063] The present invention also provides a high-voltage cable fault diagnosis device based on leakage current monitoring, including the above storage medium and a processor. When the processor executes the computer program on the storage medium, the following method steps are implemented:

[0064] S1: Determine the total circuit current and the leakage current by collecting the current induction information of the high-voltage cable through multiple sensors, and judge whether the system state is an overload fault based on a preset rule;

[0065] S2: When the system state is not an overload fault, judge whether the leakage current is within the preset range. If so, determine that the system state is normal; otherwise, enter S3;

[0066] S3: Classify the leakage current data based on the multi-class support vector machine (MSVM), calculate the resistive leakage current and capacitive leakage current, and determine whether the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and capacitive leakage current. Description of the Drawings

[0067] Figure 1 Schematic flowchart of a high-voltage cable fault diagnosis method based on leakage current monitoring according to an embodiment of the present invention;

[0068] Figure 2 Schematic architecture diagram of a high-voltage cable fault diagnosis system based on leakage current monitoring according to an embodiment of the present invention;

[0069] Figure 3 Schematic structural diagram of a high-voltage cable fault diagnosis system based on leakage current monitoring according to another embodiment of the present invention. Detailed Embodiments

[0070] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0071] As Figure 1 shown, a high-voltage cable fault diagnosis method based on leakage current monitoring includes the following steps:

[0072] S1: Collect current induction information of a high-voltage cable through multiple sensors, determine the total circuit current and leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule;

[0073] S2: When the system state is not an overload fault, judge whether the leakage current is within a preset range. If so, determine that the system state is normal; otherwise, proceed to S3;

[0074] S3: Classify the leakage current data based on the multi-class support vector machine (MSVM), calculate the resistive leakage current and capacitive leakage current, and determine whether the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and capacitive leakage current.

[0075] The high-voltage cable fault diagnosis method based on leakage current monitoring of the present invention first determines whether the system is overloaded by determining the total circuit current and leakage current through current induction information. On this basis, the leakage current data is classified based on the multi-class support vector machine (MSVM) to obtain the corresponding resistive leakage current and capacitive leakage current, combined with the corresponding preset range, so as to accurately determine that the system fault is a resistive fault and / or a capacitive fault, realize the precise classification of faults, and greatly improve the accuracy of high-voltage cable fault diagnosis.

[0076] In one or more embodiments of the present invention, in S1, the determination of the total current and leakage current in the circuit based on the current sensing data specifically includes the following steps:

[0077] For N lines, the total power of the system is:

[0078]

[0079] where P i and Q i represent the active power and reactive power of the i-th line;

[0080] Then the total current I T (t) flowing into the load is:

[0081] I T (t) = I z,r (t) + jI X,l,c (t)

[0082] I X,l,c (t) = I X,l (t) - I X,c (t)

[0083] where I X,l (t) and I X,c (t) represent the inductive and capacitive currents of the load respectively; I Z,r (t) = I T (t) cosθ I , represents the resistive current flowing into the circuit, I X,l,c (t) = I T (t) sinθ I , represents the inductive current flowing into the circuit, θ I represents the power angle under normal operating conditions;

[0084] The leakage current is:

[0085] I L (t) = I rl (t) + jI cl (t)

[0086] where I rl (t) = I L (t) cosθ L , represents the resistive leakage current, I cl (t) = I L (t) sinθ L represents the capacitive leakage current; and θ I represents the phase angle between the capacitive leakage current I cl (t) and the resistive leakage current I rl (t).

[0087] The total current in the circuit can be accurately calculated through the inductive current and capacitive current flowing into the circuit, and then, in combination with the capacitive leakage current I cl (t) and the resistive leakage current I rl (t), the capacitive leakage current I cl (t) and the resistive leakage current I rl (t) can be accurately calculated, so as to facilitate judging whether the system state has an overload fault based on a preset rule.

[0088] In one or more embodiments of the present invention, in S1, the specific steps of judging whether the system state has an overload fault based on a preset rule include the following steps:

[0089] S11: If the current sensor does not detect an induction current, the system state is a sensor fault; otherwise, go to S12;

[0090] S12: Judge whether the total current in the circuit is within a preset range. If so, determine that the system state is normal; otherwise, determine that the system state is overloaded.

[0091] Whether the sensor is faulty can be judged by the fact that the current sensor does not detect an induction current, and when the current sensor detects an induction current, it is judged whether the total current in the current is within a preset range. If it exceeds the preset range, it can be determined that the system state is overloaded; otherwise, it is still necessary to further determine whether there are other faults in the system according to the leakage current, forming a complete fault diagnosis logic with high accuracy.

[0092] In one or more embodiments of the present invention, in S3, the specific steps of performing data classification processing on the leakage current based on the multi-class support vector machine MSVM and calculating the resistive leakage current and the capacitive leakage current include the following steps:

[0093] S31: Convert the leakage current into a scatter plot, and create a hyperplane based on the scatter plot and a preset training sample;

[0094] S32: Extract the point data features corresponding to the leakage current based on the multi-class support vector machine MSVM, and optimize the hyperplane based on the given training sample to obtain an optimal hyperplane; wherein, the optimal hyperplane is used as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data;

[0095] S33: Perform summation calculations on the resistive data and the capacitive data respectively to obtain the resistive leakage current and the capacitive leakage current.

[0096] By converting the leakage current into a scatter plot, constructing a hyperplane, and focusing on the characteristic special zone of the point data corresponding to the leakage current, and then optimizing the hyperplane, the points corresponding to the leakage current can be divided into resistive data and capacitive data, so that the resistive leakage current and the capacitive leakage current can be accurately calculated, facilitating the subsequent determination of whether resistive faults and capacitive faults occur.

[0097] Specifically, in one or more embodiments of the present invention, in S31, the steps of converting the leakage current into a scatter plot and creating a hyperplane based on the scatter plot and a preset training sample specifically include the following steps:

[0098] For a given training sample where yi ∈ {1, -1} represents the class label, create a hyperplane with the expression:

[0099] θ T x i +b = 0

[0100] where, θ = [θ1, …, θ n represents an n-dimensional weight vector, x i = [x1, …, x n represents an n-dimensional input vector, b represents the bias unit, and n represents the number of features.

[0101] By using the given training sample to create a hyperplane, it is convenient to divide the point data in the scatter plot into two categories, and after optimizing the hyperplane, the corresponding resistive point data and capacitive point data can be accurately obtained.

[0102] Optionally, in one or more embodiments of the present invention, in S32, the steps of extracting the point data features corresponding to the leakage current based on the multi-class support vector machine MSVM and optimizing the hyperplane based on the given training sample to obtain the optimal hyperplane specifically include the following steps:

[0103] S321: Construct a non-linear decision function f(x):

[0104]

[0105] where, sign() represents the sign function, α i represents the coefficient of the hyperplane expression corresponding to each sample point;

[0106] S322: Call a preset plurality of kernel functions k(x i , x) to calculate the maximum value of the non-linear decision function f(x), and determine the weight vector θ corresponding to the leakage current according to the kernel function k(x i , x) corresponding to the maximum value among the maximum values of the non-linear decision function f(x);

[0107] S323: Calculate the sum of the distances from all points corresponding to the leakage current to the hyperplane according to the weight vector θ, and determine that the sum of the distances takes the minimum value:

[0108]

[0109] Where n represents the number of features. For the measured current sample points, they can be transformed into 2D samples including real and imaginary parts through phasor transformation. At this time, n = 2;

[0110] S324: Determine the hyperplane corresponding to the minimum value of the sum of the distances as the optimal hyperplane.

[0111] The beneficial effect of the above further solution is: by constructing the non - linear decision function f(x), and calculating the maximum value of the non - linear decision function f(x) used to characterize the confidence of the calculation result based on a preset plurality of kernel functions k(x i , x), in this way, the weight vector θ corresponding to the leakage current can be determined according to the maximum value among all the maximum values of the non - linear decision function f(x), so as to accurately calculate the sum of the distances from all points to the hyperplane, and then accurately determine the optimal hyperplane, realizing the accurate classification of the point data corresponding to the leakage current.

[0112] As Figure 2 shown, the present invention also provides a high - voltage cable fault diagnosis system based on leakage current monitoring. This system focuses on monitoring cable line faults in any area through timely and reliable monitoring. Figure 2 is the overall framework structure of this system. This process involves the cooperation among a safety monitoring device, a gateway system, a cloud server, a database, a detection algorithm, and visualization. As Figure 2 shown, the safety monitoring device is deployed for data acquisition, calculation, and transmission. Other necessary features are calculated from the data. The data of each sensor is transmitted through multiple channels of the LoRa gateway and uploaded to the cloud server at regular intervals. Then, the proposed algorithm classifies the data according to the preset range of leakage current and the number of active appliances. Due to the increasing number of installed sensors, data from different places are stored and analyzed on the cloud platform. After that, we apply the proposed algorithm to identify abnormal situations of the system. This system can be roughly divided into three components: devices, database, and analysis application. The device part is responsible for data acquisition through the LoRa module and data transmission to the data server. The database part is responsible for obtaining sensor data and storing it in the database. In the analysis part, the relationships between different variables are evaluated to identify abnormal states. According to the load condition corresponding to the line, we determine the acceptable leakage current to classify the abnormal situations of the system. Therefore, the analysis part will display the real - time load curve, leakage current curve, and system status.

[0113] As Figure 3 shown, in one or more embodiments of the present invention, the high-voltage cable fault diagnosis system based on leakage current monitoring includes a sensor assembly, a first calculation and judgment module, and a second calculation and judgment module;

[0114] The sensor assembly is used to collect current induction information of the high-voltage cable;

[0115] The first calculation and judgment module is used to determine the total circuit current and leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule;

[0116] The second calculation and judgment module is used to judge whether the leakage current is within a preset range when the system state is not an overload fault. If so, it is determined that the system state is normal. Otherwise, data classification processing is performed on the leakage current based on the multi-class support vector machine MSVM, and the resistive leakage current and capacitive leakage current are calculated. According to the resistive leakage current and capacitive leakage current, it is determined that the system fault is a resistive fault and / or a capacitive fault.

[0117] The high-voltage cable fault diagnosis system based on leakage current monitoring of the present invention first determines whether the system is overloaded by determining the total circuit current and leakage current through current induction information. On this basis, data classification processing is performed on the leakage current based on the multi-class support vector machine MSVM to obtain the corresponding resistive leakage current and capacitive leakage current, combined with the corresponding preset range, so as to accurately determine that the system fault is a resistive fault and / or a capacitive fault, realize the accurate classification of faults, and greatly improve the accuracy of high-voltage cable fault diagnosis.

[0118] Device part: The device is designed to be applicable to the maximum rated voltage being the amplitude of the single-phase line voltage. The LoRa device integrates multiple sensors to measure electrical parameters such as total current, terminal voltage, and leakage current. For all cases, we calculated additional necessary data based on the measurement data, such as total current, energy consumption, power factor, resistance, capacitive leakage current, and insulation resistance. In addition, our design is such that multi-level warning signals are provided with reference to the total current of the circuit breaker capacity and the allowable range of residual current. The overall calculation and data indexing are performed in the STM32L microcontroller unit (MCU). In the system design, the STM32L MCU is integrated into the LoRa transceiver device to operate under normal conditions and make differential judgments. The LoRa system consists of a terminal device, a gateway, and a network server, forming a star topology structure with the network server as the root, the gateway as the first level, and the terminal device as the leaf. The sensed and measured information is accumulated into each LoRa data packet. Additionally, a dedicated channel is allocated for transmitting LoRa data packets during time intervals, that is, the device remains idle for a period of time under normal conditions to reduce power consumption. Therefore, during the transition time from the normal state to the critical state, the device transmits data at very short intervals. The LoRa module (SX1276) used is connected to the MCU, and an omnidirectional antenna operating at 902 - 928 MHz is used to forward these data packets to the LoRa gateway module with a maximum gain of 2 dBi. LoRa modulation (proprietary chirp spread spectrum modulation) uses different types of physical layer data packets with different time lengths. The LoRa gateway is used to detect fault locations more than one kilometer away due to its proprietary large-area coverage. To store the transmitted data, the interface between the LoRa gateway and the network server (cloud) is provided by the cellular Internet protocol using the standard Transmission Control Protocol (TCP).

[0119] In one or more embodiments of the present invention, the specific implementation of the first calculation and judgment module for determining the total circuit current and leakage current based on the current sensing information is as follows:

[0120] When collecting data, the dynamic characteristics of the load must be considered because the leakage current itself is related to the volatility of the load.

[0121] For N lines, the total power of the system is:

[0122]

[0123] where P i and Q i represent the active power and reactive power of the i-th line;

[0124] Then the total current I T (t) flowing into the load is:

[0125] I TI(t) = I z,r (t) + jI X,l,c (t)

[0126] I X,l,c I(t) = I X,l (t) - I X,c (t)

[0127] where I X,l (t) and I X,c (t) represent the inductive and capacitive load currents respectively; I Z,r I(t) = I T (t) cosθ I represents the resistive current flowing into the circuit, and I X,l,c I(t) = I T (t) sinθ I represents the inductive current flowing into the circuit, and θ I represents the power angle under normal operating conditions;

[0128] The leakage current is:

[0129] I L I(t) = I rl (t) + jI cl (t)

[0130] where I rl I(t) = I L (t) cosθ L represents the resistive leakage current, and I cl I(t) = I L (t) sinθ L represents the capacitive leakage current; and θ I represents the phase angle between the capacitive leakage current I cl (t) and the resistive leakage current I rl (t).

[0131] In addition, considering the residual current I L,T (t), the total current flowing out of the circuit is expressed by the following formula:

[0132] I L I(t) = I T (t) - I L,T (t)

[0133] Therefore, the total insulation impedance can be calculated by dividing the bus voltage V(t) by the leakage current, and the formula is as follows:

[0134] Z in (t) = V(t) / I L (t)

[0135] In one or more embodiments of the present invention, the specific implementation of the first calculation and judgment module for judging whether the system state has an overload fault based on a preset rule is as follows:

[0136] If the current sensor does not detect an induced current, the system state is a sensor fault. Otherwise, it is judged whether the total current in the circuit is within a preset range. If so, the system state is determined to be normal. Otherwise, the system state is determined to be overloaded.

[0137] Whether the sensor has a fault can be judged by the fact that the current sensor does not detect an induced current. And when the current sensor detects an induced current, it is judged whether the total current in the current is within a preset range. If it exceeds the preset range, the system state can be determined to be overloaded. Otherwise, it is still necessary to further determine whether there are other faults in the system according to the leakage current, forming a complete fault diagnosis logic with high accuracy.

[0138] In one or more embodiments of the present invention, when the system state is not an overload fault, it is judged whether the leakage current is within a preset range. If so, the system state is determined to be normal.

[0139] During normal operation, the load steady-state current is much larger than the leakage current, and it is difficult to judge the actual state of the line by the consumption of energy. However, the system security can be ensured by means of three (multiple) warning types. Therefore, the overcurrent protection warning of the power system is designed according to the capacity of the deployed circuit breaker, and the multi-level warning is for leakage current protection by distinguishing resistive and capacitive residual currents.

[0140] The operating state of the system can be continuously classified by considering the operating conditions of the system, as shown in the following formula:

[0141]

[0142] According to different threshold ranges, the state is defined as SoS ∈ {SoS IT , SoS IL , SoS rl , SoS Icl}. The dynamic states in terms of the total current and the leakage current are defined as SoS IT ∈ {SoS IT N , SoS IT W , SoS IT C}, SoS IL ∈ {SoS IL N , SoS IL W , SoS ILC}. Similarly, SoS rl ∈{SoS rl N ,SoS rl W ,SoS rl C}, SoS Icl ∈{SoS Icl N ,SoS Icl W ,SoS Icl C}. Since the total current is limited by the number of lines and the rated power, the threshold range can be determined accordingly. The cut-off value of the uninterrupted and healthy system can be defined as Th N ∈{Th IT N ,Th IL N ,Th rl N ,Th Icl N}. In the present invention, we have considered the intermediate state between the safety and interruption conditions. The range set of the system's temporary warning situation is represented as Th W ∈{Th IT W ,Th IL W ,Th rl W ,Th Icl W}; The set of abnormal states is Th C ∈{Th IT C ,Th IL C ,Th rl C ,Th Icl C}. Therefore, the distinguishable constraints on I T are as follows:

[0143]

[0144] Among them, the subscript s represents the lower limit value of the normal operating current in the healthy state, and the subscript e represents the upper limit value of the normal operating current in the healthy state. However, the leakage current is not necessarily directly related to overloading. Therefore, it is necessary to describe whether the system is safe through leakage current detection. Similarly, the state of the leakage current will be determined according to the following constraints:

[0145]

[0146] Among them, the subscript s represents the lower limit of the normal operating current in a healthy state, and the subscript e represents the upper limit of the normal operating current in a healthy state. Due to the similarity of the special environment of cable lines (multiple cables in the same trench, etc.), the possibility of leakage current problems occurring simultaneously in multiple lines is relatively high. Therefore, distinguishing between resistive and capacitive leakage currents can speed up the process of identifying lines in abnormal states. Therefore, it is necessary to know the acceptable leakage current range for abnormal conditions in advance. In addition, the permissible limit of leakage current varies depending on the line type, environment, and conditions. Therefore, the constraints for a reliable and healthy system are defined as follows:

[0147]

[0148] in, By given conditions, the state of total current and leakage current can be determined. Next, the state of resistive leakage in the system is identified (this is the detection scheme of resistive leakage current). Since the boundaries of the clusters are very close to each other, the classification algorithm may provide lower accuracy. Therefore, the features can be scaled and rescaled according to the following formula.

[0149]

[0150] Among them, C k , x i , F k represents the kth cluster, the i-th original data and scaled features, which are used to form the boundaries of the cluster, E k represents the expected value of the k-th cluster and the i-th original data.

[0151] In one or more embodiments of the present invention, the second calculation and judgment module performs data classification processing on the leakage current based on a multi-classification support vector machine MSVM, and calculates the resistive leakage current and the capacitive leakage current in a specific implementation as follows:

[0152] Converting the leakage current into a scatter plot, and creating a hyperplane based on the scatter plot and a preset training sample;

[0153] Extracting the corresponding point data features of the leakage current based on a multi-classification support vector machine (MSVM), and optimizing the hyperplane based on a given training sample to obtain an optimal hyperplane; wherein the optimal hyperplane serves as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data;

[0154] The resistive leakage current and the capacitive leakage current are obtained by summing up the resistive data and the capacitive data respectively.

[0155] By converting the leakage current into a scatter plot, constructing a hyperplane, and performing special processing on the point data features corresponding to the leakage current, and then optimizing the hyperplane, the points corresponding to the leakage current can be divided into resistive data and capacitive data, so that the resistive leakage current and the capacitive leakage current can be accurately calculated, facilitating the subsequent determination of whether resistive faults and capacitive faults occur.

[0156] In one or more embodiments of the present invention, the specific implementation of converting the leakage current into a scatter plot and creating a hyperplane based on the scatter plot and a preset training sample is as follows:

[0157] For a given training sample where yi ∈ {1, -1} represents the class label, create a hyperplane with the expression:

[0158] θ T x i + b = 0

[0159] where θ = [θ1, …, θ n represents an n-dimensional weight vector, x i = [x1, …, x n represents an n-dimensional input vector, b represents the bias unit, and n represents the number of features.

[0160] By using the given training sample to create a hyperplane, it is convenient to divide the point data in the scatter plot into two categories, and after optimizing the hyperplane, the corresponding resistive point data and capacitive point data can be accurately obtained.

[0161] In one or more embodiments of the present invention, the specific implementation of extracting the point data features corresponding to the leakage current based on the multi-class support vector machine MSVM and optimizing the hyperplane based on the given training sample to obtain the optimal hyperplane is as follows:

[0162] Construct a non-linear decision function f(x):

[0163]

[0164] where sign() represents the sign function, α i represents the coefficient of the hyperplane expression corresponding to each sample point;

[0165] Call a preset multiple kernel functions k(x i , x) to calculate the maximum value of the non-linear decision function f(x), and determine the weight vector θ corresponding to the leakage current according to the kernel function k(x i , x) corresponding to the maximum value among the maximum values of the non-linear decision function f(x); the multiple kernel functions are shown in Table 1 below:

[0166] Table 1

[0167]

[0168] Calculate the sum of the distances from all points corresponding to the leakage current to the hyperplane according to the weight vector θ, and determine the minimum value of the sum of the distances:

[0169]

[0170] Where n represents the number of features. For the measured current sample points, they can be transformed into 2D samples including real and imaginary parts through phasor transformation. At this time, n = 2.

[0171] Determine the hyperplane corresponding to the minimum value of the sum of the distances as the optimal hyperplane.

[0172] By constructing a non - linear decision function f(x) and calculating the maximum value of the non - linear decision function f(x) used to characterize the confidence of the calculation result based on a preset plurality of kernel functions k(x i , x), the weight vector θ corresponding to the leakage current can be determined according to the maximum value among the maximum values of all non - linear decision functions f(x), so as to accurately calculate the sum of the distances from all points to the hyperplane, and then accurately determine the optimal hyperplane, realizing the accurate classification of the point data corresponding to the leakage current.

[0173] The present invention also provides a computer - readable storage medium storing a computer program. When the computer program is executed by a processor, the following steps are implemented:

[0174] S1: Determine the total circuit current and leakage current by collecting current induction information of the high - voltage cable through multiple sensors, and judge whether the system state is an overload fault based on a preset rule;

[0175] S2: When the system state is not an overload fault, judge whether the leakage current is within a preset range. If so, determine that the system state is normal; otherwise, enter S3;

[0176] S3: Perform data classification processing on the leakage current based on the multi - classification support vector machine MSVM, calculate the resistive leakage current and capacitive leakage current, and determine that the system fault is a resistive fault and / or capacitive fault according to the resistive leakage current and capacitive leakage current.

[0177] The present invention also provides a high - voltage cable fault diagnosis device based on leakage current monitoring, including the above - mentioned storage medium and a processor. When the processor executes the computer program on the storage medium, the following method steps are implemented:

[0178] S1: Determine the total circuit current and leakage current by collecting the current induction information of the high-voltage cable through multiple sensors, and judge whether the system state is an overload fault based on a preset rule;

[0179] S2: When the system state is not an overload fault, judge whether the leakage current is within a preset range. If so, determine that the system state is normal; otherwise, enter S3;

[0180] S3: Perform data classification processing on the leakage current based on the multi-class support vector machine MSVM, calculate the resistive leakage current and capacitive leakage current, and determine that the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and capacitive leakage current.

[0181] The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A high-voltage cable fault diagnosis method based on leakage current monitoring, characterized in that: It includes the following steps: S1: Collect the current induction information of the high-voltage cable through multiple sensors, determine the total circuit current and leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule; In the above S1, the specific steps of judging whether the system state has an overload fault based on the preset rule include the following steps: S11: If the current sensor does not detect the induction current, the system state is a sensor fault; otherwise, go to S12; S12: Judge whether the total current in the circuit is within the preset range. If so, determine that the system state is normal; otherwise, determine that the system state is overloaded; S2: When the system state is not an overload fault, judge whether the leakage current is within the preset range. If so, determine that the system state is normal; otherwise, go to S3; S3: Perform data classification processing on the leakage current based on the multi-class support vector machine MSVM, calculate the resistive leakage current and capacitive leakage current, and determine that the system fault is a resistive fault and / or a capacitive fault according to the resistive leakage current and capacitive leakage current; In the above S3, the specific steps of performing data classification processing on the leakage current based on the multi-class support vector machine MSVM and calculating the resistive leakage current and capacitive leakage current include the following steps: S31: Convert the leakage current into a scatter plot, and create a hyperplane based on the scatter plot and a preset training sample; S32: Extract the corresponding point data features of the leakage current based on the multi-class support vector machine MSVM, and optimize the hyperplane based on the given training sample to obtain the optimal hyperplane; among them, the optimal hyperplane is used as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data; S33: Perform summation calculations on the resistive data and capacitive data respectively to obtain the resistive leakage current and capacitive leakage current.

2. The high-voltage cable fault diagnosis method based on leakage current monitoring according to claim 1, wherein: In the above S1, the specific steps of determining the total current and leakage current in the circuit according to the current sensing data include the following steps: For N lines, the total power of the system is: Among them, P i and Q i represent the active power and reactive power of the i-th line; Then the total current I T (t) flowing into the load is: I T i(t) = I z,r (t) + jI X,l,c (t) I X,l,c I(t) = I X,l I(t) - I X,c I(t) Among them, I X,l (t) and I X,c (t) respectively represent the inductive and capacitive currents of the load; I Z,r (t) = I T (t) cosθ I , represents the resistive current flowing into the circuit, I X,l,c (t) = I T (t) sinθ I , represents the inductive current flowing into the circuit, θ I represents the power angle under normal operating conditions; The leakage current is: I L I(t) = rl I(t)+jI cl (t) Among them, I rl (t) = I L (t) cosθ L , represents the resistive leakage current, I cl (t) = I L (t) sinθ L represents the capacitive leakage current; and θ I represents the phase angle between the capacitive leakage current I cl (t) and the resistive leakage current I rl (t).

3. The high-voltage cable fault diagnosis method based on leakage current monitoring according to claim 1, wherein: In the above S31, the specific steps of converting the leakage current into a scatter plot and creating a hyperplane based on the scatter plot and a preset training sample include the following steps: For a given set of training samples \(\{(x i ,y i )\}\), where \(y i \in\{1, - 1\}\) represents the class label, create a hyperplane with the expression: θ T x i +b = 0 Among them, θ = [θ1, …, θ n represents an n-dimensional weight vector, x i = [x1, …, x n represents an n-dimensional input vector, b represents a bias unit, and n represents the number of features.

4. The high-voltage cable fault diagnosis method based on leakage current monitoring according to claim 3, characterized in that: In the above S32, the specific steps of extracting the corresponding point data features of the leakage current based on the multi-class support vector machine MSVM and optimizing the hyperplane based on the given training sample to obtain the optimal hyperplane include the following steps: S321: Construct a non-linear decision function f(x): where sign() represents the sign function, and α i represents the coefficients of the hyperplane expression corresponding to each sample point; S322: Call multiple preset kernel functions k(x i , x) to calculate the maximum value of the non-linear decision function f(x), and determine the weight vector θ corresponding to the leakage current according to the kernel function k(x i , x) with the largest value among the maximum values of the non-linear decision function f(x); S323: Calculate the sum of the distances between all points corresponding to the leakage current and the hyperplane according to the weight vector θ, and determine that the sum of the distances takes the minimum value: Among them, n represents the number of features. For the measured current sample points, they can be converted into 2D samples including real and imaginary parts through phasor transformation. At this time, n = 2; S324: Determine the hyperplane corresponding to the minimum value of the sum of the distances as the optimal hyperplane.

5. A high-voltage cable fault diagnosis system based on leakage current monitoring, characterized in that: It includes a sensor component, a first calculation and judgment module, and a second calculation and judgment module; The sensor component is used to collect the current induction information of the high-voltage cable; The first calculation and judgment module is configured to determine the total circuit current and the leakage current according to the current induction information, and judge whether the system state is an overload fault based on a preset rule; The specific implementation of the first calculation and judgment module for judging whether the system state has an overload fault based on a preset rule is as follows: If the current sensor does not detect the induced current, the system state is a sensor fault. Otherwise, it is judged whether the total current in the circuit is within a preset range. If so, the system state is determined to be normal. Otherwise, the system state is determined to be overloaded; The second calculation and judgment module is configured to judge whether the leakage current is within a preset range when the system state is not an overload fault. If so, the system state is determined to be normal. Otherwise, data classification processing is performed on the leakage current based on the multi-class support vector machine (MSVM), and the resistive leakage current and the capacitive leakage current are calculated. The system fault is determined to be a resistive fault and / or a capacitive fault according to the resistive leakage current and the capacitive leakage current; The specific implementation of the second calculation and judgment module for performing data classification processing on the leakage current based on the multi-class support vector machine (MSVM) and calculating the resistive leakage current and the capacitive leakage current is as follows: Convert the leakage current into a scatter plot, and create a hyperplane based on the scatter plot and a preset training sample; Extract the corresponding point data features of the leakage current based on the multi-class support vector machine (MSVM), and optimize the hyperplane based on the given training sample to obtain an optimal hyperplane. Among them, the optimal hyperplane is used as a decision boundary to divide the points corresponding to the leakage current into resistive data and capacitive data; Perform a summation calculation according to the resistive data and the capacitive data respectively to obtain the resistive leakage current and the capacitive leakage current.

6. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the high-voltage cable fault diagnosis method based on leakage current monitoring according to any one of claims 1-4 are implemented.

7. A high-voltage cable fault diagnosis device based on leakage current monitoring, characterized in that: Including the storage medium and the processor according to claim 6, when the processor executes the computer program on the storage medium, the steps of the high-voltage cable fault diagnosis method based on leakage current monitoring according to any one of claims 1-4 are implemented.

Images