Wireless sensor network grounding grid fault diagnosis method and system

Abstract

The invention discloses a wireless sensor network grounding grid fault diagnosis method and system, and the method comprises the steps: arranging a plurality of wireless sensors in a grounding grid range to form a wireless sensor network, and carrying out the magnetic field excitation of the grounding grid through excitation current; each wireless sensor in the wireless sensor network detects the magnetic field intensity of a magnetic field excited by the grounding grid to obtain magnetic field intensity distribution data; a fault diagnosis model is constructed, the fault diagnosis model is a least square support vector machine, the magnetic field intensity distribution data are input into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs a grounding grid fault result. According to the method, the wireless sensor network is constructed, the magnetic field intensity of the grounding grid is detected by using the wireless sensor, and the fault of the grounding grid is diagnosed according to the magnetic field intensity by constructing the least square support vector machine, so that the problems that the grounding grid is not detected in time and the detection efficiency is low in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of grounding grid detection, and particularly to a method and system for diagnosing grounding grid faults in a wireless sensor network. Background Art

[0002] A grounding grid is the general term for a grounding body composed of multiple metal grounding electrodes buried at a certain depth underground and a mesh structure formed by connecting these grounding electrodes with conductors. It is widely used in many industries such as power, construction, computers, industrial and mining enterprises, and communications, playing roles such as safety protection and shielding.

[0003] Current grounding grid fault monitoring methods mainly rely on regular inspections and off-line detections, and these two methods have the following deficiencies: 1. Relying on manual inspections: Traditional grounding grid monitoring methods usually require engineers to conduct regular inspections and tests, which have problems such as high labor costs and low work efficiency. In addition, the inspection frequency may be insufficient to detect problems in a timely manner. 2. Off-line detection: It cannot respond to faults in real time because it only screens and checks based on the provided historical data and cannot provide immediate alarms and diagnoses. Moreover, these two technologies have a common drawback: they generate a large amount of data and require complex data processing and analysis processes, which may lead to information overload and the risk of missing key information; many grounding grid faults require real-time response and processing to prevent the instability of the power system, and the above two methods cannot provide immediate alarms and diagnoses. Therefore, there is a need for a detection method that can autonomously detect the grounding grid and is simple and convenient. Summary of the Invention

[0004] To solve the above problems, the present invention provides a method and system for diagnosing grounding grid faults in a wireless sensor network. By constructing a wireless sensor network, using wireless sensors to detect the magnetic field intensity of the grounding grid, and constructing a least squares support vector machine to diagnose grounding grid faults based on the magnetic field intensity, the problems of untimely detection and low detection efficiency of the grounding grid in the prior art are solved.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A method for diagnosing grounding grid faults in a wireless sensor network includes the following steps:

[0007] S1. Set a number of wireless sensors within the grounding grid range to form a wireless sensor network, and excite the magnetic field of the grounding grid through an excitation current;

[0008] S2. Each wireless sensor in the wireless sensor network detects the magnetic field intensity of the magnetic field excited by the grounding grid to obtain magnetic field intensity distribution data;

[0009] S3. Construct a fault diagnosis model, where the fault diagnosis model is a least squares support vector machine. Input the magnetic field intensity distribution data into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs the grounding grid fault result.

[0010] Further, the magnetic field excitation of the grounding grid by the excitation current is specifically implemented as follows: The grounding grid includes an upper conductor, and a multi-frequency mixed sine wave current is input into the grounding grid along the upper conductor.

[0011] Further, the wireless sensor network includes a coordinator, routers, and wireless sensors. The coordinator is used to control and schedule the work of each wireless sensor. Each wireless sensor includes a sensor module, a processor module, and a wireless communication module.

[0012] Further, the sensor module is a giant magnetoresistive sensor.

[0013] Further, the wireless sensor network is a ZigBee wireless network. The wireless sensor network has a star structure, with the coordinator as the central node and each wireless sensor as a sub-node.

[0014] Further, in step S2, each wireless sensor in the wireless sensor network detects the magnetic field intensity of the magnetic field excited by the grounding grid. The specific implementation method includes:

[0015] S21. The coordinator creates a ZigBee network and waits for each wireless sensor to join.

[0016] S22. When each wireless sensor is powered on, the wireless sensor automatically searches for and joins the ZigBee network created by the coordinator, and returns the physical address of the wireless sensor to the coordinator.

[0017] S23. When the detection data of a certain physical address is required, the coordinator sends a detection instruction to the wireless sensor corresponding to the physical address according to the requirement. The wireless sensor performs detection according to the detection instruction and returns the detection data to the coordinator.

[0018] Further, in step S3, the input of the magnetic field intensity distribution data into the fault diagnosis model is specifically implemented as follows: Feature extraction is performed on the magnetic field intensity distribution data to obtain current signal features, and the current signal features are input into the fault diagnosis model.

[0019] Further, the feature extraction based on the magnetic field intensity distribution data is specifically implemented as follows: Each wireless sensor corresponds to a grounding grid node, and the magnetic field intensity distribution data includes the magnetic field intensity detected by each wireless sensor. According to Faraday's law of electromagnetic induction and Ohm's law, the current signal of each grounding grid node is calculated using the magnetic field intensity distribution data;

[0020] Feature extraction is performed on each current signal based on the box dimension. For the current signal I = {x1, x2, …, xi, ..., xn}, where n is the number of grounding grid nodes, a non-empty bounded subset is formed Cover it with M(X,a) w-dimensional hypercubes with a grid basis of a, where M(X,a) represents the number of coverages of X when covering the current signal I with grid hypercubes with a grid basis of a, and the box dimension is obtained as:

[0021]

[0022] Solve the box dimension using an approximation. Gradually magnify the grid basis a to ξa and denote:

[0023]

[0024] where P(ξa) is an intermediate variable, ξ is the scaling scale, k = 1, 2, …, n / ξ, ξ = 1, 2, …, Q, Q < n, and both n and Q are positive integers, x ξ(k-1)+1 ,x ξ(k-1)+2 ,....,x ξ(k-1)+ξ+1 represents dividing the current signal into several consecutive windows, each window containing ξ data points, where ξ is a fixed interval length, indicating that the number of elements in each group of data is ξ, max() represents taking the maximum value, min() represents taking the minimum value, and the grid count Bξa is defined as:

[0025]

[0026] Select a section (ξ1, ξ2) of the curve lnξa - lnBξa as the scale-free region, then there is:

[0027] lnBξa = clnξa + d, ξ1 ≤ ξ ≤ ξ2

[0028] where d is a constant term. Solve the slope c using the least squares method. The slope c represents the slope between ξ1 and ξ2 in lnξa - lnBξa. The obtained slope c is the box dimension D0 of the current signal. The current signal feature vector is obtained according to the change of the box dimension D0.

[0029] Further, the constraint formula of the fault diagnosis model is:

[0030]

[0031] Among them, the diagnostic result of the i-th grounding grid node U is the weight vector, G is the marginal coefficient, n is the number of grounding grid nodes, and ξ i is the error of the i-th grounding grid node, q is the bias term, and y i is the current signal feature vector of the i-th grounding grid node, and f(y i ) is the non-linear conversion function.

[0032] Through the above technical solution, the present invention has the following beneficial effects: real-time and automatically collect data through a wireless sensor network, and transmit it through wireless communication, and use a series of technologies such as data analysis and machine learning to perform real-time analysis on the data detected by the sensor. The present invention can help identify abnormal situations and predict potential fault signs, improve the reliability and accuracy of fault diagnosis, so as to timely notify relevant personnel of the faults or abnormal situations of the grounding grid, contribute to early maintenance, reduce power outage time, prevent the situation from expanding, and reduce maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a schematic diagram of the overall flow of a method for diagnosing grounding grid faults in a wireless sensor network according to the present invention.

[0034] Figure 2 is a schematic diagram of the structure of a grounding grid fault diagnosis system in a wireless sensor network according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] Example 1

[0038] See Figure 1 , a method for diagnosing grounding grid faults in a wireless sensor network, including the following steps:

[0039] S1. Set up a number of wireless sensors within the grounding grid range to form a wireless sensor network, and perform magnetic field excitation on the grounding grid through an excitation current;

[0040] S2. Each wireless sensor in the wireless sensor network detects the magnetic field intensity of the magnetic field excited by the grounding grid, and obtains the magnetic field intensity distribution data;

[0041] S3. A fault diagnosis model is constructed. The fault diagnosis model is a least squares support vector machine. The magnetic field intensity distribution data is input into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs the grounding grid fault result.

[0042] Specifically, this embodiment includes: wireless sensors, gateways or data collectors, central processing units or microcontrollers, communication modules, storage devices, displays and user interfaces, software platforms, security modules, backup and recovery systems. The wireless sensor network therein needs to detect the following data: 1. Detection of the polarization resistance of the grounding grid and measurement of the fault current distribution: current sensors; 2. Monitoring of the surface potential and evaluation of the step voltage: voltage sensors; 3. Measurement of the touch voltage: touch voltage sensors; 4. Monitoring of thermal stability: temperature sensors.

[0043] In an alternative embodiment, the magnetic field excitation of the grounding grid by the excitation current is specifically implemented as follows: The grounding grid includes an upper conductor, and a multi-frequency mixed sine wave current is input into the grounding grid along the upper conductor.

[0044] Specifically, the grounding grid includes two upper conductors, and a multi-frequency mixed sine wave current is injected along the two upper conductors of the grounding grid. The magnetic field generated by the conductor passing through the multi-frequency mixed sine wave current can be calculated according to the Biot-Savart law. The expression of the magnetic induction intensity generated by a long straight conductor L at point P is:

[0045]

[0046] Among them, B is the electromagnetic induction intensity generated by the grounding conductor line unit at the surface field point, I e is the value of the axial current in the grounding conductor unit, dl is the infinitesimal line element of the current, r is the distance between the grounding grid unit and the surface field point, and μ0 is the magnetic permeability in vacuum. The grounding grid can be regarded as a straight conductor. For a straight conductor, when the relative position between the detection point and the current unit remains unchanged, the magnetic field intensity is proportional to the current magnitude, and the current in the conductor is inversely proportional to the impedance of the conductor, and the impedance of the conductor is inversely proportional to the cross-sectional area. Therefore, the measured value of the magnetic field intensity can reflect the change in the cross-sectional area of the conductor caused by corrosion and other reasons.

[0047] In an alternative embodiment, the wireless sensor network includes a coordinator, routers, and wireless sensors. The coordinator is used to control and schedule the work of each wireless sensor. Each wireless sensor includes a sensor module, a processor module, and a wireless communication module.

[0048] In an alternative embodiment, the sensor module is a giant magnetoresistive sensor.

[0049] The sensor module uses a giant magnetoresistive (GMR) sensor to detect the electromagnetic induction intensity on the ground surface caused by the excitation current in the grounding grid, and can accurately analyze and diagnose the corrosion state of the grounding grid conductor material. The processor module and the wireless communication module use a CC2530 chip plus a low-power radio frequency front-end CC2591 to amplify the output power, greatly simplifying the design of the radio frequency circuit; the energy supply module uses 2 rechargeable dry batteries to supply energy to the system.

[0050] In an alternative embodiment, the wireless sensor network is a ZigBee wireless network, the wireless sensor network is in a star structure, the coordinator is the central node, and each wireless sensor is a sub-node.

[0051] The coordinator is responsible for scheduling the work of each sensor node, and its operation directly affects the stability of the system. The coordinator uses a CC2530F256 chip, which has 256KB of programmable flash memory, and is equipped with a serial port module, an OLED display module, an LED indicator, a crystal oscillator module, a power supply module, and a CC2591 module. The OLED module is the interaction interface between the user and the sensor network, used to display information about nodes joining and leaving the network and data in the monitoring area. The LED module is used to display the status of the network connection, and the CC2591 low-power radio frequency front-end is used to increase the output power. The communication interface uses a standard RS232 to upload the collected data to the computer through the serial port, and a MAX3232 is used to convert the RS232 level and the TTL level. When data is uploaded from the node to the computer, the MAX3232 converts the TTL level to the RS232 level for the computer to read; when the computer sends data to the node, the MAX3232 converts the RS232 level to the TTL level.

[0052] The sensor consists of a CC2530F64 chip, an external power amplifier module of the CC2591 low-power radio frequency front-end, a power supply module, a clock module, and an LED module. The CC2530 processor first performs analog-to-digital conversion on the collected data signal, then processes it, and sends it to the coordinator through the CC2591 power amplifier chip. The HGM, EN, and PA_EN pins of the CC2591 are all connected to the I / O ports P1_1, P1_4, and P0_7 of the CC2530 processor module and the wireless communication module, and are controlled by the single-chip microcomputer. When HGM is at a high level, it means that when the CC2591 power amplifier module receives data, the LNA is in a high-gain mode; when HGM is at a low level, it means that when the CC2591 power amplifier module receives data, the LNA is in a low-gain mode. The EN pin and the PA_EN pin are set to a high level when the CC2591 is working normally, and are set to a low level when it enters the low-power mode, which can reduce power consumption.

[0053] In an optional embodiment, in step S2, each wireless sensor in the wireless sensor network detects the magnetic field intensity of the magnetic field excited by the grounding grid, and the specific implementation method includes:

[0054] S21. The coordinator creates a ZigBee network and waits for each wireless sensor to join;

[0055] S22. After each wireless sensor is powered on, the wireless sensor automatically searches for and joins the ZigBee network created by the coordinator, and returns the physical address of the wireless sensor to the coordinator;

[0056] S23. When the detection data of a certain physical address is required, the coordinator sends a detection instruction to the wireless sensor corresponding to the physical address according to the requirement, and the wireless sensor performs detection according to the detection instruction and returns the detection data to the coordinator.

[0057] The ZigBee wireless network supports star, tree, and mesh topologies. The star structure connection method is relatively simple and can only form a wireless network with fewer nodes. Each sensor node realizes network connection through the coordinator.

[0058] In an optional embodiment, in step S3, inputting the magnetic field intensity distribution data into the fault diagnosis model, the specific implementation method further includes: extracting features according to the magnetic field intensity distribution data to obtain current signal features, and inputting the current signal features into the fault diagnosis model.

[0059] In an optional embodiment, the specific implementation method for extracting features according to the magnetic field intensity distribution data is: each wireless sensor corresponds to a grounding grid node, the magnetic field intensity distribution data includes the magnetic field intensity detected by each wireless sensor, and according to Faraday's law of electromagnetic induction and Ohm's law, the current signal of each grounding grid node is calculated by using the magnetic field intensity distribution data;

[0060] From the known strength of the electromagnetic induction intensity, through Faraday's law of electromagnetic induction, the expression is:

[0061]

[0062] The relationship between the induced electromotive force ε and the magnitude of the current I is determined by Ohm's law:

[0063]

[0064] Feature extraction is performed on each current signal based on the box dimension. For the current signal I =

[0065] {x1,x2,…,x i ,...,xn}, where n is the number of nodes in the grounding grid, forming a non-empty bounded subset Cover with M(X,a) w-dimensional hypercubes with a grid reference of a, where M(X,a) represents the number of covers of X when covering the current signal I with grid hypercubes with a grid reference of a, and the box dimension is obtained as follows:

[0066]

[0067] Solve for the box dimension using approximation. Gradually increase the grid reference a to ξa and denote:

[0068]

[0069] where P(ξa) is an intermediate variable, ξ is the scaling factor, k = 1, 2,..., n / ξ, ξ = 1, 2,..., Q, Q < n, and n and Q are both positive integers, x ξ(k-1)+1 , x ξ(k-1)+2 ,...., x ξ(k-1)+ξ+1 represents dividing the current signal into several consecutive windows, each window containing ξ data points, where ξ is a fixed interval length, indicating that the number of elements in each group of data is ξ. max() represents taking the maximum value, and min() represents taking the minimum value. In the formula, the maximum and minimum values calculate the data range of each window, and the final summation accumulates the fluctuations of all window data to reflect the overall change situation, where ξ is a fixed interval length, indicating that the number of elements in each group of data is ξ. Then define the grid count Bξa as:

[0070]

[0071] B is a normalized network count index used to characterize the distribution of points in a certain area; it is an intermediate quantity in the calculation of the box dimension and is used to further calculate the D0 box dimension.

[0072] Select a section (ξ1, ξ2) of the curve lnξa - lnBξa as the scale-free region, then there is:

[0073] lnBξa = clnξa + d, ξ1 ≤ ξ ≤ ξ2

[0074] where '-' in lnξa - lnBξa is the subtraction sign, d is a constant term, and use the least squares method to solve for the slope c. The slope c represents the slope between ξ1 and ξ2 in lnξa - lnBξa. The scale-free region refers to an interval that satisfies the linear power-law relationship within a certain scale range (ξ1, ξ2) and is used to calculate the fractal dimension.

[0075] In an alternative embodiment, the constraint formula of the fault diagnosis model is:

[0076]

[0077] Among them, the diagnostic result of the $i$-th grounding grid node $U$ is the weight vector, $G$ is the marginal coefficient, $n$ is the number of grounding grid nodes, and $\xi$ i is the error of the $i$-th grounding grid node, $q$ is the bias term, and $y$ i is the current signal feature vector of the $i$-th grounding grid node, and $f(y$ i ) is the non-linear transformation function.

[0078] Specifically, the current feature vector $y$ constructed by the fractal dimension is classified using LSSVM. Through the non-linear transformation function $f(y)$, $y$ is mapped from the original observation space to a high-dimensional feature space, and the optimal classification surface for different states of the line is found in the high-dimensional space.

[0079] To solve the above constrained optimization problem, the Lagrange multiplier $\beta$ i is introduced, and it is converted into an unconstrained objective function:

[0080]

[0081] According to the Karush Kuhn Tucker (KKT) conditions, let the partial derivative values of $U$, $\xi$ i , $\beta$ i and $q$ be equal to 0 respectively, and we can obtain:

[0082]

[0083] After organizing the above formula, we get:

[0084]

[0085] According to the above derivation, it can be known that this optimization problem can be solved by solving a system of linear equations. Its matrix form is shown above, and the parameters $\beta$ and $q$ can be solved by the least squares method.

[0086] The fault diagnosis result is:

[0087]

[0088] Among them, $z$ * $(y)$ is the predicted label calculated by LSSVM, representing the result of the model classifying the sample $y$, and $z$ i is the true label; $z$ * $(y)$ is the prediction result of the model, and $z$ i is the true category of the sample. During the training process, we hope that $z$ * $(y)$ and $z$ i are as close as possible, that is, the prediction is accurate; $z$* (y) generically refers to the prediction result, while refers to the predicted value of a certain sample; K(y, y i ) is a kernel function that calculates the similarity between the input sample y and the training sample y i .

[0089] Compare the LSSVM diagnosis result with the true value z in the test set i to obtain the result. Since the prediction result of the machine learning model may not be completely correct, it is necessary to compare it with the true value zi to ensure the reliability of the model.

[0090] Example 2

[0091] See Figure 2 , a ground grid fault diagnosis system for a wireless sensor network, including:

[0092] A magnetic field excitation module for setting a number of wireless sensors within the ground grid to form a wireless sensor network and exciting the ground grid with an excitation current;

[0093] A magnetic field detection module for using each wireless sensor in the wireless sensor network to detect the magnetic field intensity of the magnetic field excited by the ground grid to obtain magnetic field intensity distribution data;

[0094] A fault detection module for constructing a fault diagnosis model, where the fault diagnosis model is a least squares support vector machine, inputting the magnetic field intensity distribution data into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs the ground grid fault result.

[0095] The embodiments disclosed in this specification are only an illustration of the unilateral features of the present invention. The protection scope of the present invention is not limited to this embodiment, and any other functionally equivalent embodiments fall within the protection scope of the present invention. For those skilled in the art, various corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. A method for diagnosing faults in a grounding grid of a wireless sensor network, characterized in that, It includes the following steps: S1. Set several wireless sensors within the grounding grid to form a wireless sensor network, and excite the magnetic field of the grounding grid through an exciting current; S2. Each wireless sensor in the wireless sensor network detects the magnetic field intensity of the magnetic field excited by the grounding grid to obtain magnetic field intensity distribution data; S3. Construct a fault diagnosis model, where the fault diagnosis model is a least squares support vector machine. Input the magnetic field intensity distribution data into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs the grounding grid fault result.

2. The method for diagnosing the grounding grid fault of a wireless sensor network according to claim 1, characterized in that The specific implementation method of exciting the magnetic field of the grounding grid through the exciting current is as follows: The grounding grid includes an upper conductor, and a multi-frequency mixed sine wave current is input into the grounding grid along the upper conductor.

3. The method for diagnosing the grounding grid fault of a wireless sensor network according to claim 1, wherein, The wireless sensor network includes a coordinator, routers, and wireless sensors. The coordinator is used to control and schedule the work of each wireless sensor. Each wireless sensor includes a sensor module, a processor module, and a wireless communication module.

4. A method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 3, characterized in that, The sensor module is a giant magnetoresistive sensor.

5. The method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 3, characterized in that, The wireless sensor network is a ZigBee wireless network. The wireless sensor network is in a star structure. The coordinator is the central node, and each wireless sensor is a sub-node.

6. A method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 5, characterized in that, In step S2, the specific implementation method for each wireless sensor in the wireless sensor network to detect the magnetic field intensity of the magnetic field excited by the grounding grid includes: S21. The coordinator creates a ZigBee network and waits for each wireless sensor to join; S22. When each wireless sensor is powered on, the wireless sensor automatically searches for and joins the ZigBee network created by the coordinator, and returns the physical address of the wireless sensor to the coordinator; S23. When the detection data of a certain physical address is required, the coordinator sends a detection instruction to the wireless sensor corresponding to the physical address according to the requirement. The wireless sensor performs detection according to the detection instruction and returns the detection data to the coordinator.

7. A method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 1, characterized in that, In step S3, the specific implementation method of inputting the magnetic field intensity distribution data into the fault diagnosis model further includes: extracting features from the magnetic field intensity distribution data to obtain current signal features, and inputting the current signal features into the fault diagnosis model.

8. A method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 7, characterized in that, The specific implementation method of extracting features from the magnetic field intensity distribution data is as follows: Each wireless sensor corresponds to a grounding grid node. The magnetic field intensity distribution data includes the magnetic field intensity detected by each wireless sensor. According to Faraday's law of electromagnetic induction and Ohm's law, the current signal of each grounding grid node is calculated using the magnetic field intensity distribution data; Feature extraction is performed on each current signal based on the box dimension. For the current signal I = {x1, x2, …, xi, …, xn}, where n is the number of nodes in the grounding grid, a non-empty bounded subset is formed Cover with w-dimensional hypercubes with M(X,a) grid bases of a. Among them, M(X,a) represents the number of covers of X when the current signal I is covered by grid hypercubes with a grid base of a. The box dimension is obtained as follows: Use the approximation to solve the box dimension. Gradually magnify the grid reference a to ξa, and record: Among them, P(ξa) is an intermediate variable, ξ is a scaling scale, k = 1, 2, …, n / ξ, ξ = 1, 2, …, Q, Q < n, both n and Q are positive integers, x ξ(k-1)+1 , x ξ(k-1)+2 ,...., x ξ(k-1)+ξ+1 represents dividing the current signal into several consecutive windows, each window containing ξ data points, where ξ is a fixed interval length, indicating that the number of elements in each group of data is ξ, max() represents taking the maximum value, min() represents taking the minimum value, then the grid count Bξa is defined as: Select a section (ξ1, ξ2) in the curve lnξa - lnBξa as the scale-free region, then there is: lnBξa = clnξa + d, ξ1 ≤ ξ ≤ ξ2 where d is a constant term. Use the least squares method to solve the slope c. The slope c represents the slope between ξ1 and ξ2 in lnξa - lnBξa. The obtained slope c is the box dimension D0 of the current signal. The current signal feature vector is obtained according to the change of the box dimension D0.

9. A method for diagnosing faults in a grounding grid of a wireless sensor network according to claim 8, characterized in that, The constraint formula of the fault diagnosis model is as follows: Among them, the diagnostic result of the i-th grounding grid node U is the weight vector, G is the marginal coefficient, n is the number of grounding grid nodes, ξ i is the error of the i-th grounding grid node, q is the bias term, y i is the current signal feature vector of the i-th grounding grid node, f(y i ) is the non-linear conversion function.

10. A grounding grid fault diagnosis system for a wireless sensor network, characterized in that, Including: A magnetic field excitation module, configured to set a plurality of wireless sensors within the grounding grid to form a wireless sensor network, and perform magnetic field excitation on the grounding grid through an excitation current; A magnetic field detection module, configured to use each wireless sensor in the wireless sensor network to detect the magnetic field intensity of the magnetic field excited by the grounding grid, and obtain magnetic field intensity distribution data; A fault detection module, configured to construct a fault diagnosis model, where the fault diagnosis model is a least squares support vector machine, input the magnetic field intensity distribution data into the fault diagnosis model, and the fault diagnosis model calculates the magnetic field intensity distribution data and outputs the grounding grid fault result.

Images