A DSP line fault information acquisition system integrating traveling wave detection algorithm
By constructing a Laplace matrix and using graph domain processing, combined with a fault parameter optimization module, accurate identification and location of fault traveling waves in complex topology power grids were achieved, solving the problem of inaccurate fault location in existing technologies and improving the efficiency of power grid fault repair.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHENGZHOU LONGHUA ELECTRICAL & MECHANICAL ENG CO LTD
- Filing Date
- 2026-03-17
- Publication Date
- 2026-06-02
AI Technical Summary
Existing traveling wave fault location technology has difficulty in accurately distinguishing between the initial fault traveling wave and the branch reflected wavefront in complex power grid topologies, resulting in inaccurate fault location.
The DSP line fault information acquisition system, which adopts an integrated traveling wave detection algorithm, locates the specific line branch with the fault by constructing a Laplace matrix and performing spectral domain processing, and achieves high-precision encoding of fault distance and adaptive correction wave velocity through a fault parameter optimization module.
It enables accurate identification and location of fault traveling waves in complex topology power grids, improves the effectiveness and accuracy of fault location, simplifies the fault investigation process, and enhances the response speed and operation and maintenance efficiency of power grid fault repair.
Smart Images
Figure CN122131071A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power system fault monitoring and relay protection technology, and more specifically, to a DSP line fault information acquisition system integrating a traveling wave detection algorithm. Background Technology
[0002] With the rapid development of digital signal processing technology and embedded systems, traveling wave fault location technology has become the mainstream technology for fault monitoring of 110kV and above high voltage transmission lines, and a core supporting monitoring technology for intelligent power distribution systems, thanks to its advantages of fast response speed and high positioning accuracy.
[0003] Existing conventional traveling wave fault location devices mostly rely on single-end or double-end traveling wave detection architectures. The technical approach involves capturing the time-domain signal of the transient traveling wave generated by the fault, identifying the arrival time of the traveling wave front using edge detection methods, and calculating the fault distance by combining the fixed traveling wave propagation speed with the line length. This technology has been widely applied in single-branch or simple-branch transmission lines, providing fundamental technical support for the safe and stable operation of the power grid. However, with the increasing complexity of modern power grid structures and the growing prevalence of multi-T-connection and multi-branch topologies in transmission lines, conventional single-end or double-end traveling wave detection architectures exhibit certain application limitations. For example, when a transmission line has a T-shaped branch structure, the traveling wave, after reflection at the T-connection node and branch bifurcation point, generates multiple superimposed wavefronts. The wavefront sequence received at the measurement end contains multiple reflected waves from the fault point, branch node, and the opposite bus, making it difficult to accurately distinguish the initial fault traveling wave from the branch reflected wavefronts from the complex wavefront sequence. This results in the inability to accurately pinpoint the specific line branch where the fault occurred, thus affecting the accuracy and effectiveness of fault location in complex power grid topologies.
[0004] In view of this, the present invention proposes a DSP line fault information acquisition system integrating traveling wave detection algorithm to solve the above problems. Summary of the Invention
[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a DSP line fault information acquisition system integrating a traveling wave detection algorithm, comprising:
[0006] The DSP main control module is used to construct the Laplace matrix, monitor the energy entropy change of the multi-channel traveling wave signal, and determine the time-domain traveling wave data matrix when the energy entropy change triggers the interrupt request signal.
[0007] The spectral domain processing module is used to perform spatial transformation on the time-domain traveling wave data matrix using the Laplace matrix to obtain the spectral domain signal matrix. The principal eigenmode components in the spectral domain signal matrix are used to locate the specific faulty line branch, and the inverse transformation is used to obtain the precise arrival time of the traveling wave and the set of line parameters.
[0008] The fault parameter optimization module is used to construct a multi-dimensional solution space based on the accurate arrival time of the traveling wave and the set of line parameters. It initializes the multi-dimensional solution space by establishing an evaluation function to generate an initial optimization state matrix, and iterates the initial optimization state matrix to obtain high-precision fault distance and adaptive correction wave velocity.
[0009] The communication reporting module is used to encode high-precision fault distance and adaptive correction wave velocity to form a holographic fault report.
[0010] Furthermore, the Laplace matrix is constructed, including:
[0011] Based on the collected transmission line topology, a power grid association graph containing nodes and edges is constructed. Substations and towers of each transmission line in the transmission line topology are marked as nodes, conductors between adjacent nodes are marked as edges, and the actual length of the conductors is marked as the edge weight, thus completing the construction of the power grid association graph.
[0012] Construct a Laplace matrix with the total number of nodes in the power grid interconnection diagram as the dimension. Superimpose the weights of all edges connected to the corresponding nodes on the diagonal of the Laplace matrix, and fill in the negative or zero values of the corresponding edge weights on the off-diagonal positions to complete the construction of the Laplace matrix.
[0013] Further, the time-domain traveling wave data matrix is determined, including:
[0014] The real-time acquired multi-channel traveling wave signals are written into the internal ring buffer of the DSP, and fixed-length multi-channel traveling wave signal data segments are read from the ring buffer in sequence as the current analysis window.
[0015] Extract the instantaneous energy of all sampling points within the current analysis window and summarize it to form the total instantaneous energy value of the current analysis window; analyze the change in the total instantaneous energy value of the current analysis window relative to the previous analysis window to determine the energy entropy change of the multi-channel traveling wave signal within the current analysis window;
[0016] The energy entropy change is compared with a preset morphological threshold. If the energy entropy change is greater than the morphological threshold, an interruption request signal is generated. Based on the interruption request signal, multi-channel traveling wave signals with specified periods forward and backward from the interruption request trigger time are extracted from the circular buffer. The multi-channel traveling wave signals are arranged in order of channel and sampling time to determine the time-domain traveling wave data matrix.
[0017] Furthermore, spatial transformations are performed on the time-domain traveling wave data matrix, including:
[0018] Perform graph decomposition on the Laplacian matrix to determine the graph frequency matrix of the Laplacian matrix and the graph basis vector matrix composed of graph basis vectors;
[0019] A graphical Fourier transform is performed on the time-domain traveling wave data matrix based on the transpose of the graphical basis vector matrix to obtain the spectral domain signal matrix. The rows of the spectral domain signal matrix correspond to the graphical frequency components, and the columns correspond to the graphical frequency coefficients at each sampling time.
[0020] Furthermore, pinpoint the specific branch of the faulty circuit, including:
[0021] The energy of the spectral domain signal matrix is collected row by row to form an energy sequence corresponding to each graph frequency component; the maximum energy value in the energy sequence is searched, and the graph frequency component corresponding to the maximum energy value is taken as the principal eigenmode component.
[0022] Based on the row index of the principal intrinsic mode components in the spectral domain signal matrix, the spectral basis vectors corresponding to the row index are extracted from the spectral basis vector matrix; the magnitude of each element in the spectral basis vector is analyzed, and the nodes corresponding to the two elements with the largest magnitude are identified as the nodes at both ends of the faulty line; the specific faulty line branch is locked based on the corresponding edge of the node in the power grid correlation diagram.
[0023] Furthermore, the precise arrival times of the traveling waves and the set of line parameters are obtained, including:
[0024] Set all elements in the spectral domain signal matrix except for the row containing the principal eigenmode components to zero to obtain the filtered spectral domain signal matrix; perform a graph inverse Fourier transform on the filtered spectral domain signal matrix based on the graph spectral basis vector matrix to obtain the time-domain reconstructed signal corresponding to the principal eigenmode components.
[0025] The time corresponding to the first peak position in the time-domain reconstructed signal is used as the precise arrival time of the traveling wave; the transmission line topology is traversed to match the line length, inductance per unit length, and capacitance per unit length corresponding to the specific faulty line branch, and these are combined to form a set of line parameters.
[0026] Furthermore, a multidimensional solution space is constructed, including:
[0027] The line length in the line parameter set is extracted as the search boundary of the fault distance optimization vector; the theoretical wave speed is determined based on the unit length inductance and unit length capacitance in the line parameter set, and the preset floating range of the theoretical wave speed is used as the search boundary of the equivalent wave speed optimization vector.
[0028] The search boundary enclosed by the fault distance optimization vector and the equivalent wave velocity optimization vector is used to construct a multidimensional solution space; where each coordinate point in the multidimensional solution space corresponds to a set of fault distance values and equivalent wave velocity values.
[0029] Further, an initial optimization state matrix is generated, including:
[0030] An evaluation function is established, which takes a set of fault distance values and equivalent wave velocity values as inputs and the root mean square error between the theoretical traveling wave propagation time and the precise traveling wave arrival time as the output value.
[0031] Select several random coordinate points in the multidimensional solution space; import the fault distance value and equivalent wave velocity value corresponding to each random coordinate point into the evaluation function to obtain the output value of each random coordinate point; arrange the fault distance value, equivalent wave velocity value and corresponding output value of several random coordinate points in rows to form an initial optimization state matrix.
[0032] Furthermore, high-precision fault distance and adaptively corrected wave velocity are obtained, including:
[0033] The initial optimization state matrix is iterated over a loop. The average of the historical best positions of all random coordinate points in each iteration is taken as the average best position of the current iteration. The position coordinates of each random coordinate point in the multidimensional solution space and the output value in the evaluation function are updated based on the average best position.
[0034] The updated output value of each random coordinate point is compared with the output value of that random coordinate point at the individual's historical best position. If the updated output value is better, the global best position of the group is determined based on the updated position coordinates of the random coordinate point.
[0035] Determine whether the current iteration meets the preset iteration termination condition. If the iteration termination condition is met, determine the fault distance value corresponding to the global optimal position of the group at the time of iteration termination as the high-precision fault distance, and determine the corresponding equivalent wave velocity value as the adaptive correction wave velocity.
[0036] Furthermore, the encoding forms a holographic fault report, including:
[0037] Obtain the high-precision timestamp of the current moment; encode and splice the high-precision fault distance, adaptive correction wave velocity, specific fault line branch, and high-precision timestamp to form a holographic fault report.
[0038] The technical effects and advantages of the DSP line fault information acquisition system integrating traveling wave detection algorithm of the present invention are as follows:
[0039] 1. This invention transforms complex, multi-branched power grid topologies into algorithmically processable power grid correlation diagrams by constructing a Laplace matrix, overcoming the limitation of conventional traveling wave technology, which is only applicable to simple linear lines. By performing spectral decomposition of the Laplace matrix and executing a graphical Fourier transform, the principal eigenmode components are extracted and the specific faulty line branch is identified. This achieves spectral domain separation and purification of time-domain mixed traveling wave front signals, as well as effective differentiation between fault characteristic signals and branch reflection clutter signals. It accurately identifies the source of the fault traveling wave and locates the faulty line branch, significantly improving the effectiveness of fault location in complex topology power grids. By performing an inverse transform on the filtered spectral domain signal to obtain the time-domain reconstructed signal and extracting the precise arrival time of the traveling wave, it achieves accurate extraction and time calibration of fault characteristic waveforms, providing accurate benchmark data for subsequent fault parameter optimization.
[0040] 2. This invention constructs a multi-dimensional solution space containing optimization vectors for fault distance and equivalent wave velocity, transforming the line parameter set into fault distance search boundaries and equivalent wave velocity search boundaries, thus achieving precise adaptation of the optimization range to the actual operating conditions of the line. By establishing an evaluation function and performing iterative optimization on the initial optimization state matrix, it achieves joint dynamic optimization of fault distance and equivalent wave velocity, eliminating the influence of factors such as ambient temperature and conductor aging, and significantly improving the accuracy of fault distance detection. By iteratively optimizing and outputting high-precision fault distance and adaptively corrected wave velocity encoding to form a holographic fault report, it realizes the integration of information throughout the entire process from original signal acquisition, fault branch identification, precise time calibration to joint parameter optimization, providing complete and intuitive fault data support for power grid operation and maintenance personnel, simplifying the fault troubleshooting process, and significantly improving the response speed and operation and maintenance efficiency of power grid fault repair. Attached Figure Description
[0041] Figure 1 This is a schematic diagram of the structure of a DSP line fault information acquisition system integrating a traveling wave detection algorithm according to the present invention;
[0042] Figure 2 This is a flowchart illustrating the process of determining the time-domain traveling wave data matrix in this invention;
[0043] Figure 3 This is a flowchart illustrating how the present invention obtains high-precision fault distance and adaptively corrected wave velocity. Detailed Implementation
[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0045] Example 1, please refer to Figure 1 , Figure 2 and Figure 3 As shown in this embodiment, a DSP line fault information acquisition system integrating a traveling wave detection algorithm includes:
[0046] The DSP main control module is used to construct the Laplace matrix, monitor the energy entropy change of the multi-channel traveling wave signal, and determine the time-domain traveling wave data matrix when the energy entropy change triggers the interrupt request signal.
[0047] Constructing the Laplace matrix includes:
[0048] Based on the collected transmission line topology, a power grid association graph containing nodes and edges is constructed. Substations and towers of each transmission line in the transmission line topology are marked as nodes, conductors between adjacent nodes are marked as edges, and the actual length of the conductors is marked as the edge weight, thus completing the construction of the power grid association graph.
[0049] Retrieve the transmission line topology stored internally in the DSP. The transmission line topology includes node data (including substation number, tower number, node geographic coordinates and node type identifier), edge data (including starting node identifier, ending node identifier, actual conductor length, conductor type, inductance per unit length and capacitance per unit length), and topology relationship data (including node adjacency relationship table and line branch relationship).
[0050] Construct a Laplace matrix with the total number of nodes in the power grid interconnection diagram as the dimension. Superimpose the weights of all edges connected to the corresponding nodes on the diagonal of the Laplace matrix, and fill in the negative or zero values of the corresponding edge weights on the off-diagonal positions to complete the construction of the Laplace matrix.
[0051] For example, suppose the power grid graph contains 3 nodes: node 1, node 2 and node 3, and the edge weight between node 1 and node 2 is 5; the edge weight between node 2 and node 3 is 3.
[0052] The Laplace matrix is then: .
[0053] Determine the time-domain traveling wave data matrix, including:
[0054] The real-time acquired multi-channel traveling wave signals are written into the DSP's internal circular buffer, and fixed-length multi-channel traveling wave signal data segments are sequentially read from the circular buffer as the current analysis window. The circular buffer consists of several equal-sized storage units, each storing a fixed-length multi-channel traveling wave signal data segment. A read pointer is set in the circular buffer, initially pointing to the starting address of the buffer, and the step size of each read pointer movement is set to one storage unit. Starting from the address currently pointed to by the read pointer, data for a certain number of sampling times is continuously read. Each sampling time contains the sampled values of all channels at that time. The data from all channels and all times are combined together as the first current analysis window.
[0055] After the first current analysis window is processed, the read pointer is moved one memory unit to the right, and the next fixed-length multi-channel traveling wave signal data segment is read from the new pointer position as the starting point, which becomes the next current analysis window.
[0056] It should be explained that the fixed length represents the number of consecutive sampling points read from the circular buffer at one time, that is, the number of sampling points contained in each current analysis window. The setting of the fixed length should be based on the complete capture of the traveling wave front, ensuring that the time span of the current analysis window can cover the complete transient process of the fault traveling wave from the beginning to the peak and then back to stability, while taking into account the sensitivity of energy entropy change to signal abrupt changes.
[0057] For example, if the sampling rate for each channel is 10MHz and the fixed length is set to 256 sampling points, then each current analysis window contains all channel data for 256 sampling times.
[0058] Extract the instantaneous energy of all sampling points within the current analysis window and sum them to form the total instantaneous energy value for the current analysis window. Calculate the square of the amplitude of each sampling point as the instantaneous energy of that sampling point. Accumulate all instantaneous energies within the current analysis window to obtain the total instantaneous energy value. The total instantaneous energy value is used to measure the overall intensity of the traveling wave signal within the current analysis window.
[0059] Analyzing the change in the instantaneous total energy value of the current analysis window relative to the previous analysis window determines the energy entropy change of the multi-channel traveling wave signal within the current analysis window. Energy entropy change can capture the abrupt energy change points of the traveling wave signal. When a fault occurs, the signal energy jumps from the normal level to its peak value in a very short time at the instant the traveling wave front reaches the measurement end (manifested as a sharp pulse in energy entropy change). By detecting this pulse, the arrival time range of the faulty traveling wave can be accurately pinpointed, providing a time reference for accurately extracting the waveforms before and after the fault from the circular buffer.
[0060] The energy entropy change is compared with a preset morphological threshold. If the energy entropy change is greater than the morphological threshold, an interrupt request signal is generated. If the energy entropy change is less than or equal to the morphological threshold, it is determined that no fault has occurred, and the process returns to continue monitoring the energy entropy change in the next analysis window.
[0061] It should be explained that the morphological threshold is used to distinguish between normal fluctuations and fault events. The morphological threshold is set comprehensively based on the background noise level under normal line operation, the amplitude of load fluctuations, and the statistical characteristics of historical fault data.
[0062] Based on the interrupt request signal, the multi-channel traveling wave signal of the specified period before and after the interrupt request trigger time is extracted from the circular buffer. In response to the interrupt request signal, the DSP pauses the writing of new multi-channel traveling wave signal data to the circular buffer and completely freezes all multi-channel traveling wave signal data currently stored in the circular buffer.
[0063] It should be explained that the forward and backward specified periods are used to determine the range of multi-channel traveling wave signals extracted from the ring buffer. The forward specified period is used to preserve the normal operating waveform of the line before the fault occurs as a comparison benchmark. The backward specified period is used to completely record the entire process of traveling wave propagation, reflection, and attenuation after the fault occurs. The length of the forward and backward specified periods is usually set according to the line length, traveling wave propagation time, and the operating time window of the protection device. For short-distance lines, 3 to 5 power frequency cycles can be used, while for long-distance lines or scenarios requiring analysis of multiple traveling wave reflections, 5 to 10 power frequency cycles can be used.
[0064] For example, if each power frequency cycle is 20 milliseconds, the forward specified cycle is set to 5 power frequency cycles, and the backward specified cycle is set to 5 power frequency cycles, then a total of 10 power frequency cycles with a total length of 200 milliseconds of multi-channel traveling wave signal can be extracted. This length is sufficient to cover the complete process of the fault traveling wave from its occurrence, propagation to the opposite bus, and reflection back to the measurement end.
[0065] The multi-channel traveling wave signals are arranged according to channel and sampling time order to determine the time-domain traveling wave data matrix. The multi-channel traveling wave signals extracted from the forward and backward directions are spliced to form a continuous multi-channel traveling wave signal data sequence that includes the complete process before and after the fault. A time-domain traveling wave data matrix is created within the DSP, with the number of rows equal to the total number of channels of the multi-channel traveling wave signal and the number of columns equal to the sum of the total number of sampling points extracted from the forward and backward directions. The multi-channel traveling wave signal data sequence is then filled into the time-domain traveling wave data matrix row by row and column by column according to channel order and sampling time order.
[0066] The spectral domain processing module is used to perform spatial transformation on the time-domain traveling wave data matrix using the Laplace matrix to obtain the spectral domain signal matrix. The principal eigenmode components in the spectral domain signal matrix are used to locate the specific line branch of the fault, and the inverse transformation is used to obtain the accurate arrival time of the traveling wave and the set of line parameters.
[0067] Spatial transformation of the time-domain traveling wave data matrix includes:
[0068] Graph decomposition is performed on the Laplacian matrix to determine its graph frequency matrix and the graph basis vector matrix composed of graph basis vectors. Eigenvalue decomposition is then performed on the Laplacian matrix to obtain all eigenvalues and their corresponding eigenvectors. These eigenvalues are arranged and stored in ascending order to form a diagonal graph frequency matrix. Each eigenvalue in the graph frequency matrix represents a graph frequency component; smaller eigenvalues indicate lower graph frequencies, and larger eigenvalues indicate higher graph frequencies.
[0069] Each eigenvector is a numerical sequence of length equal to the total number of nodes. All eigenvectors are arranged in order of their corresponding eigenvalues to form a graph basis vector matrix with rows and columns equal to the total number of nodes. Each column of the graph basis vector matrix is an eigenvector corresponding to a specific graph frequency component, and each row corresponds to the projection coefficient of a node under different graph frequency bases.
[0070] A graphical Fourier transform is performed on the time-domain traveling wave data matrix using the transpose of the graphical basis vector matrix to obtain the spectral domain signal matrix. The rows of this matrix correspond to the graphical frequency components, and the columns correspond to the graphical frequency coefficients at each sampling time. The graphical basis vector matrix is then transposed by interchanging its rows and columns to generate its transpose. This transpose is used as the left multiplication matrix, and the time-domain traveling wave data matrix is used as the right multiplication matrix. The product of the transpose and the time-domain traveling wave data matrix is calculated and defined as the spectral domain signal matrix.
[0071] Identify the specific branch of the line that is faulty, including:
[0072] The energy of the spectral domain signal matrix is collected row by row to form an energy sequence corresponding to each graph frequency component. A one-dimensional array with a length equal to the number of rows in the spectral domain signal matrix is allocated internally in the DSP. All elements in the one-dimensional array are initialized to 0 and defined as the energy sequence. A row index variable is set for the spectral domain signal matrix (the row index variable starts from 0 and increments to the maximum row index of the spectral domain signal matrix): For each row indicated by the row index variable in the spectral domain signal matrix, each graph frequency coefficient in that row is traversed; the square of the graph frequency coefficient is calculated to obtain the instantaneous energy of the corresponding graph frequency component; the instantaneous energy is accumulated to the element at the corresponding index position in the energy sequence; this operation is repeated until all rows of the spectral domain signal matrix have been traversed, completing the collection of energy for each graph frequency component.
[0073] Search for the maximum energy value in the energy sequence and take the graph frequency component corresponding to the maximum energy value as the principal eigenmode component.
[0074] It should be explained that, as a locally abrupt signal, the energy of a fault traveling wave is mainly concentrated in the high-frequency graph components related to the fault location. Since the fault energy is highly concentrated in specific high-frequency components, while noise energy is relatively uniformly distributed across components, the graph frequency component with the highest energy has the highest signal-to-noise ratio. In the graph basis vector corresponding to this graph frequency component, the amplitude of nodes near the fault point is the most significant, and it can most clearly reflect the fault location information. Therefore, by using the graph frequency component corresponding to the maximum energy value as the principal eigenmode component, the dominant mode that best represents the fault characteristics can be extracted from the complex multi-component signal.
[0075] Based on the row indices of the principal eigenmode components in the spectral domain signal matrix, the spectral basis vectors corresponding to the row indices are extracted from the spectral basis vector matrix. According to the maximum energy index of the principal eigenmode components, the maximum energy index column is located in the spectral basis vector matrix, and all elements of the maximum energy index column are extracted to form the spectral basis vector corresponding to the principal eigenmode components; the length of this spectral basis vector is equal to the total number of nodes in the power grid interconnection diagram.
[0076] Analyze the magnitude of each element in the graph basis vector, and identify the nodes corresponding to the two elements with the largest magnitudes as the nodes at both ends of the faulty line. Based on the corresponding edges of these nodes in the power grid interconnection diagram, pinpoint the specific branch of the faulty line. Take the absolute value of each element in the graph basis vector, and sort the elements in descending order. Select the two elements with the largest values after sorting, and read their original position indices in the index array. Define these two original position indices as the node numbers at both ends of the faulty line. Find the edge connecting these two faulty node numbers in the power grid interconnection diagram to complete the identification of the specific branch of the faulty line.
[0077] It should be explained that when the principal eigenmode with the highest energy is selected for inverse transformation, the resulting time-domain reconstructed signal, in its spatial domain (i.e., at each node), will have significantly higher amplitudes at the two ends of the branch where the fault point is located, as these are the starting and ending points (or boundaries) of the traveling wave propagation path. Therefore, by finding the two nodes with the largest amplitudes in the spectral basis vectors, the specific line branch where the fault is located can be identified.
[0078] Obtain the precise arrival time of the traveling wave and the set of line parameters, including:
[0079] By setting all elements in all rows of the spectral domain signal matrix except those containing the principal eigenmode components to zero, we obtain the filtered spectral domain signal matrix. This filtered spectral domain signal matrix is used to retain only the information of the principal eigenmode components, while completely filtering out information from other frequency components.
[0080] A graphical inverse Fourier transform is performed on the filtered spectral domain signal matrix based on the graphical basis vector matrix to obtain the time-domain reconstructed signal corresponding to the principal eigenmode components. Using the graphical basis vector matrix as the transform kernel, matrix multiplication is performed on the filtered spectral domain signal matrix, left-multiplying the graphical basis vector matrix by the filtered spectral domain signal matrix to obtain the time-domain reconstructed signal matrix. The row with the largest amplitude is selected from the time-domain reconstructed signal matrix as the time-domain reconstructed signal corresponding to the principal eigenmode components.
[0081] It's important to explain that the inverse graphical Fourier transform is used to remap the spectral domain signal back to the time domain. When only the principal eigenmode components are retained and the remaining components are set to zero, the inverse transform yields the time-domain waveform contributed solely by the principal eigenmode components. Since the principal eigenmode components represent the vibration modes where fault energy is most concentrated, their reconstructed waveforms contain only the traveling wave components related to the fault, filtering out irrelevant components such as branch node reflections and noise interference. Therefore, the first peak in the time-domain reconstructed signal can more accurately reflect the true moment when the fault traveling wave first arrives at the measurement end.
[0082] The moment corresponding to the first peak position in the reconstructed time-domain signal is taken as the precise arrival time of the traveling wave. The reconstructed time-domain signal is scanned point by point, and the amplitude of adjacent sampling points is compared sequentially starting from the initial moment. The sampling moment corresponding to the first turning point where the upward trend turns into a downward trend is determined as the moment corresponding to the peak position, i.e., the precise arrival time of the traveling wave.
[0083] Traverse the transmission line topology, match the line length, inductance per unit length, and capacitance per unit length corresponding to the specific faulty line branch, and combine them to form a set of line parameters.
[0084] The fault parameter optimization module is used to construct a multi-dimensional solution space based on the accurate arrival time of the traveling wave and the set of line parameters. It initializes the multi-dimensional solution space by establishing an evaluation function to generate an initial optimization state matrix, and iterates the initial optimization state matrix to obtain high-precision fault distance and adaptive correction wave velocity.
[0085] Constructing a multidimensional solution space, including:
[0086] The line length in the line parameter set is extracted as the search boundary of the fault distance optimization vector. The search boundary of the fault distance optimization vector lies between the minimum boundary value (zero) and the maximum boundary value (the actual physical line length of the specific line branch with the fault). For example, if the line length is 50km, the search boundary of the fault distance optimization vector is [0, 50]km.
[0087] The theoretical wave velocity is determined based on the unit length inductance and unit length capacitance in the circuit parameter set. A preset floating range of the theoretical wave velocity is used as the search boundary for the equivalent wave velocity optimization vector. The theoretical wave velocity is the reciprocal of the square root of the product of the unit length inductance and unit length capacitance.
[0088] It should be explained that the setting of the theoretical wave speed fluctuation range needs to comprehensively consider the influence of actual factors such as conductor operating temperature, ambient temperature and humidity, conductor aging degree, and sag changes on the wave speed. The fluctuation range should be sufficient to cover possible changes in wave speed under various operating conditions, while avoiding a decrease in optimization efficiency due to an excessively large range.
[0089] For example, in the line parameter set: the inductance per unit length is 0.9337 × 10⁻³ H / km, and the capacitance per unit length is 1.274 × 10⁻ 8 F / km. The calculated theoretical wave velocity is approximately 2.98 × 10⁻⁶. 5 km / s.
[0090] Considering the slight increase in inductance due to increased conductor temperature during the high-temperature period in summer, the wave velocity may decrease to 95% of the theoretical value; while during the low-temperature period in winter, the wave velocity may increase to approximately 102% of the theoretical value. Taking these factors into account, the fluctuation range of the theoretical wave velocity is set to 90% to 108% of the theoretical value; that is, the search boundary for the equivalent wave velocity optimization vector is 2.68 × 10⁻⁶. 5 km / s to 3.22×10 5 km / s.
[0091] A multidimensional solution space is constructed by enclosing the search boundaries of the fault distance optimization vector and the equivalent wave velocity optimization vector. Each coordinate point in the multidimensional solution space corresponds to a set of fault distance values and equivalent wave velocity values. The search boundary of the fault distance optimization vector is mapped to the range of values on the horizontal axis of a two-dimensional coordinate system, and the search boundary of the equivalent wave velocity optimization vector is mapped to the range of values on the vertical axis of a two-dimensional coordinate system. A rectangular region is formed by enclosing the horizontal and vertical axis ranges, and this rectangular region is defined as the multidimensional solution space. Any coordinate point within the multidimensional solution space contains a valid fault distance value and a valid equivalent wave velocity value.
[0092] Generate the initial optimization state matrix, including:
[0093] An evaluation function is established, which takes a set of fault distance values and equivalent wave velocity values as inputs and the root mean square error between the theoretical traveling wave propagation time and the precise traveling wave arrival time as the output value.
[0094] The expression for the evaluation function is defined as follows: ;
[0095] Where d represents the current fault distance value; This represents the current equivalent wave velocity value; The output value of the evaluation function represents the fault distance value. and equivalent wave velocity value The corresponding root mean square error; This indicates the number of independent observation data points involved in the fault location process, i.e., the number of independent traveling wave arrival times used for evaluation (the observation samples can be multiple traveling wave fronts or multiple independent fault events under the same fault). Indicates the wave head ordinal number; Indicates the first The precise arrival time of the traveling wave observed.
[0096] Within the multidimensional solution space, several random coordinate points are selected (using a uniformly distributed random sampling method). The fault distance value and equivalent wave velocity value corresponding to each random coordinate point are imported into the evaluation function to obtain the output value for each random coordinate point. It should be noted that the number of random coordinate points is set according to the range of the multidimensional solution space and the required optimization accuracy. For example, for transmission lines with a length of less than 100km, the number of random coordinate points is set to 40. This number ensures population diversity while meeting the real-time requirement of the DSP completing initialization before the next sampling window arrives.
[0097] It should be explained that the smaller the output value of the coordinate point, the closer the combination of fault distance and equivalent wave velocity is to the actual situation.
[0098] The fault distance values, equivalent wave velocity values, and corresponding output values of several random coordinate points are arranged in rows to form an initial optimization state matrix. That is, the first column stores the fault distance values, the second column stores the equivalent wave velocity values, and the third column stores the output values of the evaluation function.
[0099] High-precision fault distance and adaptively corrected wave velocity are obtained, including:
[0100] Perform a loop iteration on the initial optimization state matrix, and take the average of the individual historical best positions of all random coordinate points in each iteration as the average best position of this iteration.
[0101] Before the first iteration, the current coordinates of each random point in the initial optimization state matrix are recorded as the individual's historical best position; the output value of the evaluation function corresponding to each random point is recorded as the individual's historical best output value. Simultaneously, the position corresponding to the minimum value among all individual historical best output values is recorded as the population's global best position.
[0102] The position coordinates of each random point in the multidimensional solution space and the output value in the evaluation function are updated based on the average optimal position.
[0103] The updated output value of each random coordinate point is compared with the output value of that random coordinate point at its historical best position. If the updated output value is better, the group's global best position is determined based on the updated coordinates of the random coordinate point. After updating the historical best positions of all random coordinate points, all historical best output values are iterated to find the minimum value. If this minimum value is less than the current group's global best output value, the group's global best position is updated to the random coordinate point position corresponding to the minimum value, and the group's global best output value is updated simultaneously; otherwise, the group's global best position remains unchanged.
[0104] Determine whether the current iteration meets the preset iteration termination condition. If the iteration termination condition is met, determine the fault distance value corresponding to the global optimal position of the group at the time of iteration termination as the high-precision fault distance, and determine the corresponding equivalent wave velocity value as the adaptive correction wave velocity.
[0105] It should be explained that the preset iteration termination conditions include the maximum number of iterations and the minimum fitness threshold. Specifically, after each iteration cycle, the output value corresponding to the current global position of the population is calculated. This output value is compared with the minimum fitness threshold. If the output value is less than the minimum fitness threshold, the convergence termination condition is directly satisfied. If the output value is not less than the minimum fitness threshold, i.e., the convergence termination condition is not satisfied, the current iteration count is further read, and it is checked whether the iteration count is greater than or equal to the preset maximum number of iterations. If it is, the iteration termination condition is satisfied. If neither of the above two conditions is satisfied, the process returns to continue executing the next iteration cycle, repeating the iteration termination condition comparison process.
[0106] It should be explained that the minimum fitness threshold is set based on the line fault location accuracy, combined with the industry accuracy standards for traveling wave fault location in high-voltage transmission lines. The threshold is determined by selecting the critical value corresponding to the minimum error interval within the output range of the evaluation function. This ensures that the fault parameter combination corresponding to reaching the minimum fitness threshold meets the high-precision location requirements, while also adapting to the differences in operating conditions of transmission lines at different voltage levels.
[0107] The maximum number of iterations is set based on the DSP's computing power, real-time response time limit, multi-dimensional solution space scale, and number of random coordinate points. This ensures that invalid iterations do not occupy computing resources and delay fault information reporting. At the same time, it matches the timeliness requirements of emergency response to transmission line faults and ensures that parameter optimization is completed within the limited computing cycle.
[0108] The communication reporting module is used to encode high-precision fault distance and adaptive correction wave velocity to form a holographic fault report.
[0109] The encoding generates a holographic fault report, including:
[0110] Obtain the high-precision timestamp of the current moment; encode and splice the high-precision fault distance, adaptive correction wave velocity, specific fault line branch, and high-precision timestamp to form a holographic fault report.
[0111] In this embodiment, by constructing a Laplace matrix, the complex multi-branch power grid topology is transformed into a power grid correlation diagram that can be processed by algorithms, overcoming the limitation of conventional traveling wave technology being only applicable to simple linear lines. By performing spectral decomposition on the Laplace matrix and executing a graphical Fourier transform, the principal eigenmode components are extracted and the specific line branch of the fault is located. This achieves spectral domain separation and purification of the time-domain mixed traveling wave front signal, as well as effective differentiation between fault feature signals and branch reflection clutter signals, accurately identifying the source of the fault traveling wave and locating the line branch to which the fault belongs, significantly improving the effectiveness of fault location in complex topology power grids. By performing an inverse transform on the filtered spectral domain signal to obtain the time-domain reconstructed signal and extracting the precise arrival time of the traveling wave, the accurate extraction and time calibration of the fault feature waveform are achieved, providing accurate reference data for subsequent fault parameter optimization.
[0112] By constructing a multidimensional solution space containing optimization vectors for fault distance and equivalent wave velocity, the set of line parameters is transformed into search boundaries for fault distance and equivalent wave velocity, achieving precise adaptation of the optimization range to the actual operating conditions of the line. By establishing an evaluation function and performing iterative optimization on the initial optimization state matrix, joint dynamic optimization of fault distance and equivalent wave velocity is achieved, eliminating the influence of factors such as ambient temperature and conductor aging, and significantly improving the accuracy of fault distance detection. By iteratively optimizing and outputting high-precision fault distance and adaptively corrected wave velocity coding to form a holographic fault report, a complete information integration process from raw signal acquisition, fault branch identification, precise time calibration to joint parameter optimization is realized, providing complete and intuitive fault data support for power grid operation and maintenance personnel, simplifying the fault troubleshooting process, and significantly improving the response speed and operation and maintenance efficiency of power grid fault repair.
[0113] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0114] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0115] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.
[0116] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A DSP line fault information acquisition system integrating traveling wave detection algorithm, characterized in that, The DSP line fault information acquisition system integrating traveling wave detection algorithm includes: The DSP main control module is used to construct the Laplace matrix, monitor the energy entropy change of the multi-channel traveling wave signal, and determine the time-domain traveling wave data matrix when the energy entropy change triggers the interrupt request signal. The spectral domain processing module is used to perform spatial transformation on the time-domain traveling wave data matrix using the Laplace matrix to obtain the spectral domain signal matrix. The principal eigenmode components in the spectral domain signal matrix are used to locate the specific faulty line branch, and the inverse transformation is used to obtain the precise arrival time of the traveling wave and the set of line parameters. The fault parameter optimization module is used to construct a multi-dimensional solution space based on the accurate arrival time of the traveling wave and the set of line parameters. It initializes the multi-dimensional solution space by establishing an evaluation function to generate an initial optimization state matrix, and iterates the initial optimization state matrix to obtain high-precision fault distance and adaptive correction wave velocity. The communication reporting module is used to encode high-precision fault distance and adaptive correction wave velocity to form a holographic fault report.
2. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 1, characterized in that, The construction of the Laplacian matrix includes: Based on the collected transmission line topology, a power grid association graph containing nodes and edges is constructed. Substations and towers of each transmission line in the transmission line topology are marked as nodes, conductors between adjacent nodes are marked as edges, and the actual length of the conductors is marked as the edge weight, thus completing the construction of the power grid association graph. Construct a Laplace matrix with the total number of nodes in the power grid interconnection diagram as the dimension. Superimpose the weights of all edges connected to the corresponding nodes on the diagonal of the Laplace matrix, and fill in the negative or zero values of the corresponding edge weights on the off-diagonal positions to complete the construction of the Laplace matrix.
3. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 2, characterized in that, The determined time-domain traveling wave data matrix includes: The real-time acquired multi-channel traveling wave signals are written into the internal ring buffer of the DSP, and fixed-length multi-channel traveling wave signal data segments are read from the ring buffer in sequence as the current analysis window. Extract the instantaneous energy of all sampling points within the current analysis window and summarize it to form the total instantaneous energy value of the current analysis window; analyze the change in the total instantaneous energy value of the current analysis window relative to the previous analysis window to determine the energy entropy change of the multi-channel traveling wave signal within the current analysis window; The energy entropy change is compared with a preset morphological threshold. If the energy entropy change is greater than the morphological threshold, an interruption request signal is generated. Based on the interruption request signal, multi-channel traveling wave signals with specified periods forward and backward from the interruption request trigger time are extracted from the circular buffer. The multi-channel traveling wave signals are arranged in order of channel and sampling time to determine the time-domain traveling wave data matrix.
4. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 3, characterized in that, The spatial transformation of the time-domain traveling wave data matrix includes: Perform graph decomposition on the Laplacian matrix to determine the graph frequency matrix of the Laplacian matrix and the graph basis vector matrix composed of graph basis vectors; A graphical Fourier transform is performed on the time-domain traveling wave data matrix based on the transpose of the graphical basis vector matrix to obtain the spectral domain signal matrix. The rows of the spectral domain signal matrix correspond to the graphical frequency components, and the columns correspond to the graphical frequency coefficients at each sampling time.
5. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 4, characterized in that, The specific line branch where the fault is located includes: The energy of the spectral domain signal matrix is collected row by row to form an energy sequence corresponding to each graph frequency component; the maximum energy value in the energy sequence is searched, and the graph frequency component corresponding to the maximum energy value is taken as the principal eigenmode component. Based on the row index of the principal intrinsic mode components in the spectral domain signal matrix, the spectral basis vectors corresponding to the row index are extracted from the spectral basis vector matrix; the magnitude of each element in the spectral basis vector is analyzed, and the nodes corresponding to the two elements with the largest magnitude are identified as the nodes at both ends of the faulty line; the specific faulty line branch is locked based on the corresponding edge of the node in the power grid correlation diagram.
6. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 5, characterized in that, The obtained precise traveling wave arrival time and line parameter set include: Set all elements in the spectral domain signal matrix except for the row containing the principal eigenmode components to zero to obtain the filtered spectral domain signal matrix; perform a graph inverse Fourier transform on the filtered spectral domain signal matrix based on the graph spectral basis vector matrix to obtain the time-domain reconstructed signal corresponding to the principal eigenmode components. The time corresponding to the first peak position in the time-domain reconstructed signal is used as the precise arrival time of the traveling wave; the transmission line topology is traversed to match the line length, inductance per unit length, and capacitance per unit length corresponding to the specific faulty line branch, and these are combined to form a set of line parameters.
7. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 6, characterized in that, The construction of the multidimensional solution space includes: The line length in the line parameter set is extracted as the search boundary of the fault distance optimization vector; the theoretical wave speed is determined based on the unit length inductance and unit length capacitance in the line parameter set, and the preset floating range of the theoretical wave speed is used as the search boundary of the equivalent wave speed optimization vector. The search boundary enclosed by the fault distance optimization vector and the equivalent wave velocity optimization vector is used to construct a multidimensional solution space; where each coordinate point in the multidimensional solution space corresponds to a set of fault distance values and equivalent wave velocity values.
8. The DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 7, characterized in that, The generation of the initial optimization state matrix includes: An evaluation function is established, which takes a set of fault distance values and equivalent wave velocity values as inputs and the root mean square error between the theoretical traveling wave propagation time and the precise traveling wave arrival time as the output value. Select several random coordinate points in the multidimensional solution space; import the fault distance value and equivalent wave velocity value corresponding to each random coordinate point into the evaluation function to obtain the output value of each random coordinate point; arrange the fault distance value, equivalent wave velocity value and corresponding output value of several random coordinate points in rows to form an initial optimization state matrix.
9. A DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 8, characterized in that, The process of obtaining high-precision fault distance and adaptively corrected wave velocity includes: The initial optimization state matrix is iterated over a loop. The average of the historical best positions of all random coordinate points in each iteration is taken as the average best position of the current iteration. The position coordinates of each random coordinate point in the multidimensional solution space and the output value in the evaluation function are updated based on the average best position. The updated output value of each random coordinate point is compared with the output value of that random coordinate point at the individual's historical best position. If the updated output value is better, the global best position of the group is determined based on the updated position coordinates of the random coordinate point. Determine whether the current iteration meets the preset iteration termination condition. If the iteration termination condition is met, determine the fault distance value corresponding to the global optimal position of the group at the time of iteration termination as the high-precision fault distance, and determine the corresponding equivalent wave velocity value as the adaptive correction wave velocity.
10. A DSP line fault information acquisition system integrating traveling wave detection algorithm according to claim 9, characterized in that, The encoding forms a holographic fault report, including: Obtain the high-precision timestamp of the current moment; encode and splice the high-precision fault distance, adaptive correction wave velocity, specific fault line branch, and high-precision timestamp to form a holographic fault report.