Power transmission line fault positioning method and system based on Beidou time service and positioning technology

Through Beidou time and positioning technology, combined with traceless Kalman filtering and graph neural module, the transmission line fault positioning method is optimized, solving the problem of low accuracy in traditional methods, and achieving higher positioning accuracy and system response capabilities.

CN118914747AActive Publication Date: 2024-11-08MAANSHAN POWER SUPPLY COMPANY STATE GRID ANHUI ELECTRIC POWER +1

Patent Information

Application Number
CN202410980041.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-22
Publication Date
2024-11-08
Estimated Expiration
2044-07-22

AI Technical Summary

Technical Problem

The traditional transmission line fault positioning method has low accuracy and cumbersome steps, which is easily affected by traveling wave superposition and interference, resulting in insufficient positioning accuracy.

Method used

Using a method based on Beidou timer and positioning technology, the identification of fault point location and type is optimized by building a state space model, using traceless Kalman filtering algorithm and graph neural module, and combining the instantaneous switching model.

Benefits of technology

It improves the accuracy and accuracy of power line fault positioning, reduces errors, enhances the system's response capabilities, and can effectively combine model prediction and measurement information.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914747A_ABST
    Figure CN118914747A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of power transmission line fault positioning, and discloses a power transmission line fault positioning method and system based on Beidou time service and positioning technology, and the method comprises the steps: building a state space model based on the parameters of a monitored power transmission line; if the value of the dimension, representing the position of the fault point on the power transmission line, of the posterior extended state estimation value obtained through the unscented Kalman filtering algorithm is a non-zero value, entering the next step; inputting the covariance matrix at the first moment and the parameters of the power transmission line at the first moment into an instantaneous switching model, and outputting an instantaneous covariance matrix by the instantaneous switching model; replacing the covariance matrix at the first moment with the instantaneous covariance matrix, continuing to run the unscented Kalman filtering algorithm, then obtaining a posterior extended state estimation value at a third moment, and decoding to obtain the position of a fault point; according to the invention, artificial intelligence is combined with unscented Kalman filtering to optimize the defects of a traditional power transmission line fault positioning algorithm.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of transmission line fault location, and more specifically, to a transmission line fault location method and system based on Beidou timing and positioning technology. Background Art

[0002] Traditional transmission line fault location requires current transient traveling waves to identify the fault type. Different fault types have different characteristics of current transient traveling waves. Next, the current transient traveling waves need to be analyzed according to the fault type to find the peaks of the primary echo and the secondary echo, and then the distance is calculated by time difference to locate the fault. The steps are relatively cumbersome, and there is superposition and interference of traveling waves. Incorrect fault type judgment will also affect the accuracy of fault location. Overall, the positioning accuracy of this method is basically hundreds of meters. Summary of the invention

[0003] The present invention provides a transmission line fault locating method and system based on Beidou timing and positioning technology, which solves the technical problem of low accuracy of traditional transmission line fault locating in related technologies.

[0004] The present invention provides a transmission line fault location method based on Beidou timing and positioning technology, comprising:

[0005] Step 101, constructing a state space model based on the parameters of the monitored transmission line, the state space model including a state vector, a state equation, a measurement equation and a measurement vector;

[0006] Step 102, increasing the dimension of the state vector representing the position of the fault point on the transmission line to obtain an extended state vector, and updating the state equation and the measurement equation based on the extended state vector;

[0007] Step 103, obtaining the posterior extended state estimate value at each monitoring moment by using the unscented Kalman filter algorithm, if the value of the dimension of the posterior extended state estimate value representing the position of the fault point on the transmission line is non-zero, proceeding to the next step, and defining this monitoring moment as the first moment;

[0008] Step 104, inputting the covariance matrix at the first moment and the parameters of the power transmission line at the first moment into the instantaneous switching model, and the instantaneous switching model outputs the instantaneous covariance matrix;

[0009] Step 105, replacing the covariance matrix at the first moment with the instantaneous covariance matrix, continuing to run the unscented Kalman filter algorithm, and then obtaining the posterior extended state estimate at the third moment, decoding the value of the dimension of the extended state estimate representing the position of the fault point on the transmission line to obtain the position of the fault point;

[0010] There is a second moment between the first moment and the third moment. The first moment, the second moment and the third moment are three adjacent moments and are arranged from front to back in time.

[0011] Furthermore, the instantaneous switching model includes:

[0012] A graph neural module, a first convolution module, an upsampling layer, a tensor layer, and a fully connected layer, wherein the graph neural module inputs graph structure data constructed based on the parameters of the transmission line at the first moment, and outputs a first eigenvector. The first convolution module convolves the input covariance matrix to obtain a first feature map. The upsampling layer upsamples the first feature map to generate an instantaneous covariance matrix. The instantaneous covariance matrix is ​​input into the tensor layer to obtain a second eigenvector. The first eigenvector and the second eigenvector are concatenated and input into the fully connected layer. The fully connected layer outputs a classification result representing the fault type.

[0013] Furthermore, the method for constructing graph structure data based on the parameters of the power transmission line at the first moment includes:

[0014] The graph structure data includes nodes and edges connecting the nodes, wherein the nodes include a first node representing a line segment and a second node representing a global transmission line;

[0015] The node feature of the first node is obtained by encoding the operation parameters of the corresponding line segment;

[0016] The node characteristics of the second node are obtained by encoding the global operation parameters of the transmission line;

[0017] There are edges between the second node and all the first nodes;

[0018] There is an edge between the first nodes connected by the line segments represented;

[0019] The calculation formula of the graph neural module is as follows:

[0020] Messaging:

[0021]

[0022] Where m i←j is the message sent by node j to node i, h i and h j are the current representations of nodes i and j respectively, is the message function;

[0023] Message aggregation:

[0024] m i =ρ ← ({m i←j ∣j∈N(i)})

[0025] Where mi is the aggregation of all messages received by node i, N(i) is the set of neighbor nodes of node i, ρ ← is the message aggregation function;

[0026] Node Update:

[0027]

[0028] where h i ' is node i according to message m i The updated new representation, is the node update function;

[0029] Repeat K iterations to obtain a higher-order node representation. The second node representation after iteration is as the first eigenvector.

[0030] Furthermore, the fault types include: lightning strike, lightning strike back, tree obstacles, forest fires, hanging objects, ice cover, wind deviation, and construction equipment touching the line.

[0031] Furthermore, the state space model of a uniform transmission line with a length of l is expressed as:

[0032] Equation of state:

[0033] x k =f(x k-1 ,u k-1 )+w k-1

[0034] Measurement equation:

[0035] z k =h(x k ,u k )+v k

[0036] In the state equation and measurement equation:

[0037] is the state vector at time k, which includes multiple dimensions;

[0038] is the control input vector at time k;

[0039] is the measurement vector at time k;

[0040] is a nonlinear state transfer function;

[0041] is a nonlinear measurement function;

[0042] is process noise, with mean 0 and covariance Q k Gaussian distribution of

[0043] is the measurement noise, with mean 0 and covariance R k Gaussian distribution of; the extended state vector is:

[0044]

[0045] Accordingly, the state equation and measurement equation are modified as follows:

[0046]

[0047] z k =h(y k ,u k )+v k

[0048] The fault distance ratio is:

[0049]

[0050] in:

[0051] α is the fault distance ratio, ranging from [0,1];

[0052] d f Indicates the distance from the fault point to the line terminal;

[0053] l represents the total length of the entire transmission line.

[0054] Furthermore, the state vector includes the measured state variables such as voltage and current on all line segments.

[0055] Further, the measurement vector includes voltage and current measurements at both ends of all line segments.

[0056] Further, the steps of the unscented Kalman filter algorithm are as follows:

[0057] initialization:

[0058] Initial state mean: y0 = E[y0]

[0059] Initial state covariance: P0 = E[(y0-y0)(y0-y0) T ]

[0060] For k = 1, 2, .., perform the following steps:

[0061] a. Calculate 2n+1 sigma points:

[0062]

[0063] in is the scaling parameter, α determines the distribution of sigma points around y(k-1), and usually takes a very small positive value; k is a non-negative second-order moment parameter, usually 3-n.

[0064] Time update:

[0065] Status prediction:

[0066]

[0067] State prediction covariance:

[0068]

[0069] in and is the weight, satisfying

[0070] The commonly used weight calculation methods are as follows:

[0071]

[0072] Here β is used to incorporate high-order information, and β=2 is optimal for Gaussian distribution.

[0073] Measurement Update:

[0074] Measuring predictions:

[0075]

[0076] Measure the prediction covariance:

[0077]

[0078] Cross covariance:

[0079]

[0080] Kalman gain:

[0081]

[0082] State and covariance updates:

[0083]

[0084] P k Represents the updated covariance at time k.

[0085] The present invention provides a transmission line fault location system based on Beidou timing and positioning technology, comprising: a signal detection terminal, a master station system and a mobile handheld terminal;

[0086] The signal detection terminal is installed on the transmission line conductor or tower, with one set installed every 20 to 30 km. It detects the traveling wave signal on the conductor in real time and sends it to the main station system via GPRS.

[0087] Master station system:

[0088] Software system: used to execute the transmission line fault location method based on Beidou timing and positioning technology as described in any one of claims 1 to 8;

[0089] Hardware system: application server, data server, data network communication equipment, and human-computer interaction equipment

[0090] A computer storage medium, characterized in that it stores computer-readable instructions, and when a computer reads the computer-readable instructions, it can execute the above-mentioned transmission line fault location method based on Beidou timing and positioning technology

[0091] The beneficial effects of the present invention are:

[0092] The present invention adopts a lightweight system based on the Kalman filter algorithm to ensure timely response of the system, and takes into account that the Kalman filter relies on continuous state transfer in time series, while the fault of the transmission line is non-frequent, that is to say, in the signal sequence within a period of time, the abnormal signal caused by the fault accounts for a very small proportion. Ideally, when the covariance matrix of the Kalman filter converges to a stable value, the Kalman gain will also tend to be stable, which means that the filter has found an optimal balance point and can effectively combine model prediction and measurement information. However, the abnormal signal of the fault will break the balance of the Kalman filter that has converged after a long period of fault-free signal, and the error of the Kalman filter in locating the fault is difficult to further reduce. The present invention optimizes the defects of the traditional transmission line fault location algorithm by combining artificial intelligence with unscented Kalman filtering. BRIEF DESCRIPTION OF THE DRAWINGS

[0093] Figure 1 It is a flow chart of the transmission line fault location method based on Beidou timing and positioning technology of the present invention;

[0094] Figure 2 is a schematic diagram of the instantaneous switching model of the present invention;

[0095] Figure 3 It is an architecture diagram of the transmission line fault location system based on Beidou timing and positioning technology of the present invention;

[0096] Figure 4 It is a physical picture of the signal detection terminal of the present invention. DETAILED DESCRIPTION

[0097] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Various examples may omit, replace or add various processes or components as needed. In addition, the features described in some examples may also be combined in other examples.

[0098] At least one embodiment of the present invention discloses a method for locating a transmission line fault based on Beidou timing and positioning technology, such as Figure 1 As shown, including:

[0099] Step 101, constructing a state space model based on the parameters of the monitored transmission line, the state space model including a state vector, a state equation, a measurement equation and a measurement vector;

[0100] Step 102, increasing the dimension of the state vector representing the position of the fault point on the transmission line to obtain an extended state vector, and updating the state equation and the measurement equation based on the extended state vector;

[0101] Step 103, obtaining the posterior extended state estimate value at each monitoring moment by using the unscented Kalman filter algorithm, if the value of the dimension of the posterior extended state estimate value representing the position of the fault point on the transmission line is non-zero, proceeding to the next step, and defining this monitoring moment as the first moment;

[0102] Time is a continuous value and needs to be discretized for analysis. The interval between two adjacent monitoring moments is generally the sampling period of the measurement unit.

[0103] Step 104, inputting the covariance matrix at the first moment and the parameters of the power transmission line at the first moment into the instantaneous switching model, and the instantaneous switching model outputs the instantaneous covariance matrix;

[0104] Step 105, replacing the covariance matrix at the first moment with the instantaneous covariance matrix, continuing to run the unscented Kalman filter algorithm, and then obtaining the posterior extended state estimate at the third moment, decoding the value of the dimension of the extended state estimate representing the position of the fault point on the transmission line to obtain the position of the fault point;

[0105] There is a second moment between the first moment and the third moment. The first moment, the second moment and the third moment are three adjacent moments and are arranged from front to back in time.

[0106] This embodiment provides a lightweight online monitoring technology for power transmission lines that uses an unscented Kalman filter algorithm as the main body for dynamic monitoring, and combines deep learning technology to apply to system tuning of fault transients to improve positioning accuracy.

[0107] like Figure 2 As shown, in one embodiment of the present invention, the instantaneous switching model includes:

[0108] A graph neural module, a first convolution module, an upsampling layer, a tensor layer, and a fully connected layer, wherein the graph neural module inputs graph structure data constructed based on the parameters of the transmission line at the first moment, and outputs a first eigenvector. The first convolution module convolves the input covariance matrix to obtain a first feature map. The upsampling layer upsamples the first feature map to generate an instantaneous covariance matrix. The instantaneous covariance matrix is ​​input into the tensor layer to obtain a second eigenvector. The first eigenvector and the second eigenvector are concatenated and input into the fully connected layer. The fully connected layer outputs a classification result representing the fault type.

[0109] The aforementioned method for constructing graph structure data based on the parameters of the power transmission line at the first moment includes:

[0110] The graph structure data includes nodes and edges connecting the nodes, wherein the nodes include a first node representing a line segment and a second node representing a global transmission line;

[0111] The node feature of the first node is obtained by encoding the operation parameters of the corresponding line segment;

[0112] The node characteristics of the second node are obtained by encoding the global operation parameters of the transmission line;

[0113] There are edges between the second node and all the first nodes;

[0114] There is an edge between the first nodes connected by the line segments represented;

[0115] The calculation formula of the graph neural module is as follows:

[0116] Messaging:

[0117]

[0118] Where m i←j is the message sent by node j to node i, h i and h j are the current representations of nodes i and j respectively, is the message function;

[0119] Message aggregation:

[0120] m i =ρ ← ({m i←j∣j∈N(i)})

[0121] Where m i is the aggregation of all messages received by node i, N(i) is the set of neighbor nodes of node i, ρ ← is the message aggregation function;

[0122] Node Update:

[0123]

[0124] where h i ' is node i according to message m i The updated new representation, is the node update function;

[0125] Repeat K iterations to obtain a higher-order node representation. The second node representation after iteration is as the first eigenvector.

[0126] As mentioned above, the tensorization method of the tensor layer is to concatenate the row vectors of the matrix in order.

[0127] As mentioned above, upsampling may be performed by methods such as deconvolution.

[0128] The aforementioned fault types include: lightning strike, lightning strike back, tree obstacles, forest fires, hanging objects, ice cover, wind deviation, construction equipment touching the line, etc.

[0129] During training, the loss function, such as the cross entropy loss function, is calculated using the results of fault type prediction.

[0130] Furthermore, the loss of the fault point location prediction is used as the local loss function of the upsampling layer and the first convolution module, which can be used for transfer learning of the pre-trained instantaneous switching model.

[0131] As mentioned above, the first convolution module can adopt a CNN model such as LeNet or VGGNet.

[0132] In one embodiment of the present invention, a system for locating a transmission line fault using the aforementioned Beidou timing and positioning technology is provided, comprising: a traveling wave signal detection terminal (hereinafter referred to as a 'signal detection terminal'), a master station system, and a mobile handheld terminal;

[0133] Signal detection terminal:

[0134] The signal detection terminal is installed on the transmission line conductor or tower, with one set installed every 20 to 30 km. It detects the traveling wave signal on the conductor in real time and sends it to the main station system via GPRS.

[0135] Master station system:

[0136] -Software system: fault analysis and location application software, database software, operating system, and other application software;

[0137] -Hardware system: application server and its peripherals, data server, data network communication equipment, and human-computer interaction equipment, etc.;

[0138] Mobile handheld terminal:

[0139] -3G mobile handheld terminal: intelligent embedded operating system handheld terminal;

[0140] It also includes the necessary Beidou timing module, which performs clock synchronization of all signal detection devices based on Beidou satellite timing.

[0141] In one embodiment of the present invention, the remote communication interface between the signal detection device and the system master station adopts IEC61850 and IEC104 communication protocols, which can also meet the needs of accessing the current mainstream integrated automation and dispatching automation systems.

[0142] In one embodiment of the present invention, various types of signal detection terminal devices are designed and produced to meet the application requirements of AC and DC power transmission lines of various voltage levels and different line types.

[0143] These signal detection terminals are installed on transmission line conductors or towers to detect traveling wave signals generated by the following faults:

[0144] -Lightning fault: monitor, record and upload lightning fault current waveform in transmission lines;

[0145] - (Non-lightning fault) Single-phase and phase-to-phase fault: monitor, record and upload transient current traveling wave waveforms in power lines.

[0146] The configuration and selection of the signal detection terminal are shown in Table 3-1.

[0147] Table 3-1 Signal detection terminal configuration selection table

[0148]

[0149] In one embodiment of the present invention,

[0150] By arranging several on-site monitoring terminals on the transmission line, the transmission line is divided into several sections (each section length ≤ 30km), and the fault data and signals of each section are monitored and recorded respectively. This monitoring method of decentralized layout in the line rather than centralized layout at the substations at both ends not only discretizes the inherent parameters that affect the monitoring error, such as the length of the transmission line and the sag of the conductor, but also effectively reduces the influence of factors such as the change of the speed of small traveling waves and attenuation on the precise location of faults due to the shortened monitoring section of each monitoring device, thereby greatly improving the accuracy of transmission line fault location.

[0151] Consider a uniform transmission line with a length of l, whose state space model can be expressed as:

[0152] Equation of state:

[0153] x k =f(x k-1 ,u k-1 )+w k-1

[0154] Measurement equation:

[0155] z k =h(x k ,u k )+v k

[0156] The state equation describes the evolution of the system state over time, which can usually be modeled by the system's differential equation or difference equation. Taking a transmission line as an example, the state vector may include variables such as voltage and current at each node on the line.

[0157] Then the state equation x k =f(x k-1 ,u k-1 )+w k-1 Kirchhoff's laws, circuit theory, etc. can be used to establish differential or difference equations, and the process noise w can be added k-1 To describe the uncertainty of the system.

[0158] The measurement equation describes how to obtain the observed value from the system state, which depends on the actual measurement equipment and method. Taking the transmission line as an example, the measured value may be the voltage, current and other telemetry values ​​at both ends of the line. Then the measurement equation z k =h(x k ,u k )+v k It can be modeled by the measurement principle, such as the working principle of voltage and current sensors, and the measurement noise v is added k to describe the uncertainty of the measurement.

[0159] The measurement equation and state equation described in this embodiment can be referenced by using existing nonlinear equations for transmission line fault location, or can even be directly referenced by using existing

[0160] Since the present invention does not optimize the state equation and the measurement equation, and the innovation of the present invention is to optimize the defects of the traditional transmission line fault location algorithm through artificial intelligence combined with unscented Kalman filtering, it is also necessary for the present invention to consider not optimizing the state equation and measurement equation derived from the existing physical model. Of course, it does not mean that the choice of state equation and measurement equation has no effect on the results. It must be pointed out that modeling can better express the equations of the transmission line system and can bring better positioning results.

[0161] In the state equation and measurement equation:

[0162] is the state vector at time k, which includes multiple dimensions;

[0163] In one embodiment of the present invention, it is necessary to take into account the hardware system on which the present invention relies, and a data acquisition unit is set on each line segment of the transmission line, so the state vector includes state variables such as voltage and current on all measured line segments.

[0164] is the control input vector at time k, including line parameters, load conditions, etc.

[0165] is the measurement vector at time k, including the voltage and current measurements at both ends of all line segments.

[0166] is a nonlinear state transfer function.

[0167] is a nonlinear measurement function.

[0168] is process noise, with mean 0 and covariance Q k Gaussian distribution of

[0169] is the measurement noise, with mean 0 and covariance R k Gaussian distribution of

[0170] In order to introduce the fault location into the state space model, the fault distance ratio α∈[0,1] is defined, which represents the ratio of the distance from the fault point to the terminal to the total length of the line. The extended state vector is:

[0171]

[0172] Accordingly, the state equation and measurement equation are modified as follows:

[0173]

[0174] The fault distance ratio is:

[0175]

[0176] in:

[0177] α is the fault distance ratio, ranging from [0,1];

[0178] d f Indicates the distance from the fault point to the line terminal;

[0179] l represents the total length of the entire transmission line;

[0180] Considering simultaneous concurrent failures, although this situation is relatively rare, the present invention provides an embodiment applicable to this situation, configuring more than two dimensions representing the location of the fault point, so that concurrent failures can be resolved.

[0181] The main steps of the UKF algorithm are as follows:

[0182] (1) Initialization:

[0183] Initial state mean: y0 = E[y0]

[0184] Initial state covariance: P0 = E[(y0-y0)(y0-y0) T ]

[0185] (2) For k = 1, 2, ..., perform the following steps:

[0186] a. Calculate 2n+1 sigma points:

[0187]

[0188] in is the scaling parameter, α determines the distribution of sigma points around y(k-1), and usually takes a very small positive value; k is a non-negative second-order moment parameter, usually 3-n.

[0189] b. Time update (forecast):

[0190] Status prediction:

[0191]

[0192] State prediction covariance:

[0193]

[0194] in and is the weight, satisfying

[0195] The commonly used weight calculation methods are as follows:

[0196]

[0197] Here β is used to incorporate high-order information, and β=2 is optimal for Gaussian distribution.

[0198] c. Measurement update (correction):

[0199] Measuring predictions:

[0200]

[0201] Measure the prediction covariance:

[0202]

[0203] Cross covariance:

[0204]

[0205] Kalman gain:

[0206]

[0207] State and covariance updates:

[0208]

[0209] P k Represents the updated covariance at time k;

[0210] After the above UKF algorithm, the posterior extended state estimate at each time k can be obtained: This includes the fault distance ratio α k The estimated results.

[0211] The important fault type identification criteria are as follows:

[0212] (1) Three-phase short circuit fault:

[0213] (2) The three-phase current amplitudes change suddenly and are nearly equal, and the phase difference is close to 120°.

[0214] (2) Two-phase short circuit fault:

[0215] The amplitudes of the two-phase currents change suddenly and are nearly equal, and are much larger than the third-phase current, and the phase difference between the two-phase currents is close to 180°.

[0216] (3) Single-phase grounding fault:

[0217] The current amplitude of the fault phase increases significantly and is significantly greater than the current amplitudes of the other two phases.

[0218] (4) Two-phase grounding fault:

[0219] The amplitudes of the two-phase currents suddenly change and become nearly equal, and are significantly larger than the third-phase current. The phase difference between the two-phase currents may deviate from 180°.

[0220] At least one embodiment of the present invention provides a computer storage medium that stores computer-readable instructions. When a computer reads the computer-readable instructions, it can execute the aforementioned transmission line fault location method based on Beidou timing and positioning technology.

[0221] The above describes an embodiment of the present invention, but this embodiment is not limited to the above-mentioned specific implementation mode. The above-mentioned specific implementation mode is merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make more forms of equivalent embodiments, all of which are within the protection of this embodiment.

Claims

1. A transmission line fault location method based on Beidou timing and positioning technology, characterized in that: include: Step 101, constructing a state space model based on the parameters of the monitored transmission line, the state space model including a state vector, a state equation, a measurement equation and a measurement vector; Step 102, increasing the dimension of the state vector representing the position of the fault point on the transmission line to obtain an extended state vector, and updating the state equation and the measurement equation based on the extended state vector; Step 103, obtaining the posterior extended state estimate value at each monitoring moment by using the unscented Kalman filter algorithm, if the value of the dimension of the posterior extended state estimate value representing the position of the fault point on the transmission line is non-zero, proceeding to the next step, and defining this monitoring moment as the first moment; Step 104, inputting the covariance matrix at the first moment and the parameters of the power transmission line at the first moment into the instantaneous switching model, and the instantaneous switching model outputs the instantaneous covariance matrix; Step 105, replacing the covariance matrix at the first moment with the instantaneous covariance matrix, continuing to run the unscented Kalman filter algorithm, and then obtaining the posterior extended state estimate at the third moment, decoding the value of the dimension of the extended state estimate representing the position of the fault point on the transmission line to obtain the position of the fault point; There is a second moment between the first moment and the third moment. The first moment, the second moment and the third moment are three adjacent moments and are arranged from front to back in time.

2. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 1, characterized in that: The instantaneous switching model includes: A graph neural module, a first convolution module, an upsampling layer, a tensor layer, and a fully connected layer, wherein the graph neural module inputs graph structure data constructed based on the parameters of the transmission line at the first moment, and outputs a first eigenvector. The first convolution module convolves the input covariance matrix to obtain a first feature map. The upsampling layer upsamples the first feature map to generate an instantaneous covariance matrix. The instantaneous covariance matrix is ​​input into the tensor layer to obtain a second eigenvector. The first eigenvector and the second eigenvector are concatenated and input into the fully connected layer. The fully connected layer outputs a classification result representing the fault type.

3. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 2 is characterized in that: The method for constructing graph structure data based on the parameters of the power transmission line at the first moment includes: The graph structure data includes nodes and edges connecting the nodes, wherein the nodes include a first node representing a line segment and a second node representing a global transmission line; The node feature of the first node is obtained by encoding the operation parameters of the corresponding line segment; The node characteristics of the second node are obtained by encoding the global operation parameters of the transmission line; There are edges between the second node and all the first nodes; There is an edge between the first nodes connected by the line segments represented; The calculation formula of the graph neural module is as follows: Messaging: Where m i←j is the message sent by node j to node i, h i and h j are the current representations of nodes i and j respectively, is the message function; Message aggregation: m i =ρ ← ({m i←j ∣j∈N(i)}) Where m i is the aggregation of all messages received by node i, N(i) is the set of neighbor nodes of node i, ρ ← is the message aggregation function; Node Update: where h i ' is node i according to message m i The updated new representation, is the node update function; Repeat K iterations to obtain a higher-order node representation. The second node representation after iteration is as the first eigenvector.

4. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 2, characterized in that: Fault types include: lightning strike, lightning strike back, tree obstacles, forest fires, hanging objects, ice cover, wind deviation, and construction equipment touching the line.

5. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 1, characterized in that: The state space model of a uniform transmission line with a length of l is expressed as: Equation of state: x k =f(x k-1 ,u k-1 )+w k-1 Measurement equation: z k =h(x k ,u k )+v k In the state equation and measurement equation: is the state vector at time k, which includes multiple dimensions; is the control input vector at time k; is the measurement vector at time k; is a nonlinear state transfer function; is a nonlinear measurement function; is process noise, which obeys a Gaussian distribution with a mean of 0 and a covariance of Qk; is the measurement noise, which follows a Gaussian distribution with a mean of 0 and a covariance of Rk; The extended state vector is: Accordingly, the state equation and measurement equation are modified as follows: z k =h(y k ,u k )+v k The fault distance ratio is: in: α is the fault distance ratio, ranging from [0, 1]; a f Indicates the distance from the fault point to the line terminal; l represents the total length of the entire transmission line.

6. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 5, characterized in that: The state vector includes the measured state variables such as voltage and current on all line segments.

7. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 5, characterized in that: The measurement vector includes the voltage and current measurements at both ends of all line segments.

8. The method for locating a transmission line fault based on Beidou timing and positioning technology according to claim 5, characterized in that: The steps of the unscented Kalman filter algorithm are as follows: initialization: Initial state mean: y0 = E[y0] Initial state covariance: P0 = E[(y0-y0)(y0-y0) T ] For k=1, 2, .., perform the following steps: a. Calculate 2n+1 sigma points: in is the scaling parameter; Time update: Status prediction: State prediction covariance: Where W i (m) and W i (c) is the weight, satisfying ∑W i =1,W i (c) ≥0; Measurement Update: Measuring predictions: Measure the prediction covariance: Cross covariance: Kalman gain: State and covariance updates: P k Represents the updated covariance at time k.

9. The transmission line fault location system based on Beidou timing and positioning technology is characterized by: include: Signal detection terminal, master station system and mobile handheld terminal; The signal detection terminal is installed on the transmission line conductor or tower, with one set installed every 20 to 30 km. It detects the traveling wave signal on the conductor in real time and sends it to the main station system via GPRS. Master station system: Software system: used to execute the transmission line fault location method based on Beidou timing and positioning technology as described in any one of claims 1 to 8; Hardware system: application server, data server, data network communication equipment, and human-computer interaction equipment; It also includes a Beidou timing module, which performs clock synchronization of all signal detection devices based on Beidou satellite timing.

10. A computer storage medium, characterized in that: It stores computer-readable instructions, and when a computer reads the computer-readable instructions, it can execute the transmission line fault locating method based on Beidou timing and positioning technology as described in any one of claims 1-8.

Citation Information

Patent Citations

  • Power transmission line fault positioning method based on Kalman filter

    CN110749835A

  • Microgrid group fault diagnosis method and device based on multi-source data fusion

    CN116990631A

  • Volume Kalman filtering cooperative positioning method based on robust message passing mechanism

    CN117288175A

  • Positioning method and positioning apparatus

    WO2023023936A1

Cited By

  • Distribution line fault positioning method and system based on Beidou system

    CN119482269A

  • Traveling wave fault positioning system and method based on Beidou synchronization and intelligent filtering

    CN120294800A

  • Power transmission line fault positioning method based on Beidou time service positioning

    CN120446661A