A power network topology identification method and system based on an intelligent fusion terminal

Abstract

This invention provides a power network topology identification method and system based on an intelligent fusion terminal. The technical solution involves constructing an orthogonal spreading code sequence and distributing it to each end node. Using the power line zero-crossing point as the physical synchronization reference for the entire network, all end nodes are controlled to simultaneously perform low-power parallel physical modulation. At the fusion terminal, the acquired superimposed waveform data undergoes differential preprocessing, and an iterative despreading algorithm based on in-situ waveform template learning is used to analyze the signal strength of each node. Finally, multi-port contention decision logic is combined to determine the physical topology attribution. This invention solves the technical problems of low topology identification efficiency and low identification accuracy in environments with strong background noise and signal crosstalk.

Description

Technical Field

[0001] This invention relates to the field of low-voltage distribution network monitoring technology, specifically to a power network topology identification method and system based on an intelligent fusion terminal. Background Technology

[0002] With the construction of new power systems, low-voltage distribution substations have been connected to a massive number of distributed photovoltaic systems, electric vehicle charging piles, and smart home loads. In order to achieve precise control and lean management of the power generation, grid, load, and storage in the substations, it has become crucial to clarify the physical topology relationship between each end node and the distribution transformer side (substation fusion terminal).

[0003] Existing power network topology identification technologies mainly include methods based on power line carrier communication feature analysis and methods based on characteristic current or voltage pulse injection. Among them, existing technologies based on characteristic signal injection typically employ time-division multiplexing or polling mechanisms, whereby a central node controls each end node to sequentially send specific current pulses or create voltage dips, and the central node determines its affiliation by detecting the presence or absence of the signal.

[0004] However, this existing serial polling mechanism has significant limitations. First, when the number of nodes in the distribution area is large, such as hundreds of nodes, traversing all nodes takes an extremely long time, resulting in low identification efficiency and failing to meet the needs of real-time dynamic topology sensing. Second, the low-voltage distribution network environment is harsh, with strong power frequency fundamental waves, abundant harmonic noise, and load fluctuations (such as photovoltaic cloud shading and large load start-ups and shutdowns). Existing simple threshold detection methods are easily overwhelmed by background noise in low signal-to-noise ratio environments, leading to missed detections, or are subject to impulse noise interference, resulting in false detections, and have weak anti-interference capabilities. Furthermore, in scenarios with multi-phase or multi-circuit parallel laying, crosstalk generated by electromagnetic coupling between lines can cause induced signals to be detected even on non-connected phases. Existing technologies lack an effective competitive decision mechanism to distinguish between real connection signals and induced crosstalk signals, making phase or branch identification difficult.

[0005] To improve the signal-to-noise ratio, existing technologies often require significant current surges or voltage drops at the end nodes, which can cause flicker pollution to the power grid and even trigger the protection actions of sensitive equipment. Meanwhile, traditional identification algorithms fail to fully utilize the powerful edge computing capabilities of next-generation intelligent fusion terminals, making it difficult to handle complex signal demodulation and waveform analysis tasks. Therefore, how to leverage edge computing power to achieve efficient, parallel, and highly reliable topology identification of massive numbers of end nodes without affecting power quality is a pressing technical problem to be solved in this field. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention proposes a power network topology identification method and system based on an intelligent fusion terminal. This invention aims to solve the technical problems of low topology identification efficiency and low identification accuracy in environments with strong background noise and signal crosstalk.

[0007] The technical solution adopted in this invention is as follows: by constructing an orthogonal spreading code sequence and distributing it to each end node, using the power line zero-crossing point as the physical synchronization reference for the entire network, all end nodes are controlled to simultaneously perform low-power parallel physical modulation; on the fusion terminal side, the collected superimposed waveform data is differentially preprocessed, and a serial interference cancellation algorithm based on in-situ waveform template learning is used for iterative despreading to analyze the signal strength of each node; finally, the physical topology assignment is determined by combining multi-port contention decision logic.

[0008] In a first possible implementation, the method operates in a power network system comprising a central edge computing node and multiple end nodes, and includes the following steps:

[0009] The central edge computing node constructs a set of mutually orthogonal spreading code sequences and distributes the spreading code sequences to each end node in the power network system.

[0010] In response to the end node receiving the spreading code sequence, the modulation start time is locked according to a unified time base;

[0011] Each of the terminal nodes performs modulation on the electrical physical quantities of the connection point according to its spreading code sequence logic, generating superimposed composite waveform data on the power line composed of multiple terminal node modulation signals and power network background signals.

[0012] The central edge computing node collects superimposed and synthesized waveform data on the power line at a preset sampling rate;

[0013] The superimposed waveform data is preprocessed to extract feature sequences containing modulation features;

[0014] In the edge computing environment, the central edge computing node iteratively processes the feature sequence and analyzes the signal component strength corresponding to each spreading code sequence.

[0015] The physical topology affiliation between each end node and the monitoring port of the central edge computing node is determined based on the signal component strength.

[0016] Furthermore, determining the unified time base includes the following steps:

[0017] Each terminal node uses the zero-crossing point of the power line voltage waveform as a reference point for a common time base;

[0018] The end node initiates modulation at the Nth reference point after receiving the spreading code sequence.

[0019] Furthermore, the spreading code sequence logic specifically includes:

[0020] The duration of each symbol in the spreading code sequence is configured to be an integer multiple of the power line frequency period or an integer multiple of half a period.

[0021] The spreading code sequence is configured such that, at any modulation time, the number of end nodes performing modulation actions and the number of end nodes remaining silent are maintained within a preset balance ratio range.

[0022] Furthermore, modulation operations are performed on the electrical physical quantities of the connection point, specifically including:

[0023] The end node controls the impedance elements, load elements and / or power supply elements connected inside it to switch between different states according to the logic level in the spreading code sequence.

[0024] The modulation action is configured as a low-power modulation mode, so that the signal fluctuation amplitude generated by a single end node is lower than the base amplitude of the power line background noise.

[0025] Furthermore, the superimposed waveform data is preprocessed to extract feature sequences containing modulation features, including:

[0026] The power frequency component in the superimposed and synthesized waveform data is removed based on the power line power frequency reference.

[0027] Normalization mapping is performed on the superimposed composite waveform data after removing the power frequency component to generate a feature sequence that logically matches the spreading code sequence.

[0028] Furthermore, the feature sequence is iteratively processed to analyze the signal component strength corresponding to each spreading code sequence. Specific steps include:

[0029] Perform parallel correlation operations between the current feature sequence and all unidentified spreading code sequences to lock the target spreading code with the largest correlation peak, and use the absolute value of the correlation peak as the signal component strength of the target spreading code;

[0030] Based on the symbol timing of the target spreading code, the current feature sequence is divided into multiple symbol segments;

[0031] All code segments corresponding to logic 1 are superimposed and averaged to generate a positive feature template;

[0032] All code segments corresponding to logic -1 or 0 are superimposed and averaged to generate a negative feature template;

[0033] According to the timing logic of the target spreading code, the positive feature template and the negative feature template are concatenated to generate the reconstructed waveform of the target end node;

[0034] Subtract the reconstructed waveform from the current feature sequence to obtain residual waveform data;

[0035] In response to the residual waveform data not meeting the preset termination condition, the residual waveform data is updated to the current feature sequence, and the parallel correlation operation and subsequent steps are repeated.

[0036] In response to the residual waveform data satisfying the preset termination condition, the iteration is terminated and all target spreading codes and their corresponding signal component strengths are output.

[0037] The preset termination conditions include:

[0038] The root mean square value of the residual waveform data is calculated as the noise basis for the current iteration.

[0039] Calculate the ratio of the signal component intensity recorded in the current iteration to the noise floor;

[0040] In response to the ratio being less than a preset minimum signal-to-noise ratio threshold or the current iteration number reaching a preset maximum iteration number, it is determined that the termination condition is met;

[0041] If the ratio is greater than or equal to a preset minimum signal-to-noise ratio threshold or the current iteration count reaches a preset maximum iteration count, it is determined that the termination condition is not met.

[0042] Furthermore, an average is calculated by stacking the values. The specific steps include:

[0043] Extract K code segments belonging to the same logical state from the feature sequence;

[0044] The initial reference vector is obtained by performing an arithmetic average on all the aforementioned symbol segments;

[0045] Calculate the correlation coefficient between each symbol segment and the initial reference vector as the weight of the current symbol segment:

[0046]

[0047] in, This indicates the weight of the current code segment. Indicates the current code segment, Indicates the current code segment The average of all elements in the set. Represents the initial reference vector. This represents the average of all initial reference vectors. The L2 norm of a vector;

[0048] A feature template is generated by weighting the current symbol segment using the weights of the current symbol segment:

[0049]

[0050] in, Represents a feature template. This indicates the weight of the current code segment. This indicates the current code segment.

[0051] Furthermore, the physical topology affiliation between each end node and the monitoring port of the central edge computing node is determined based on the signal component strength, specifically including:

[0052] Read the intensity of all signal components in the current identification period;

[0053] For each spreading code sequence, the ratio of its corresponding signal component intensity to the original total energy of the feature sequence is calculated to obtain the energy proportion coefficient;

[0054] In response to the absolute value of the signal component strength of the spreading code sequence being greater than a preset hard decision threshold and its energy proportion coefficient being greater than a preset correlation coefficient threshold, it is determined that the end node of the spreading code sequence is physically connected to the monitoring port of the current central edge computing node.

[0055] If the signal component strength or the energy ratio coefficient does not meet the threshold condition, it is determined that the end node is not connected to the current monitoring port.

[0056] Furthermore, when the central edge computing node includes multiple monitoring ports, determining the physical topology affiliation between each end node and the monitoring port of the central edge computing node also includes multi-port contention determination, specifically including the following steps:

[0057] The central edge computing node simultaneously collects the feature sequences of the multiple monitoring ports and independently executes the iterative processing to construct an intensity feature vector for the same end node under different monitoring ports;

[0058] Identify the maximum and second largest values ​​in the intensity feature vector, and calculate the ratio of the difference between the maximum and the second largest values;

[0059] In response to the maximum value exceeding a preset effective signal threshold and the difference ratio exceeding a preset crosstalk suppression tolerance, it is determined that the end node uniquely belongs to the monitoring port corresponding to the maximum value.

[0060] If the difference ratio is lower than the crosstalk suppression tolerance, the terminal node is determined to be in the signal crosstalk region, no attribution determination is performed, and an abnormal alarm is generated.

[0061] In conjunction with the first feasible method, a second feasible method provides a power network topology identification system based on an intelligent fusion terminal, characterized by comprising:

[0062] The intelligent converged terminal, as a central edge computing node, includes a data acquisition module, a communication unit, and a processor, wherein the processor provides an edge computing environment.

[0063] Multiple end nodes are distributed on the branches or user side of the power network, and each end node is equipped with a synchronization triggering unit, a zero-crossing detection circuit and a power modulation circuit.

[0064] The processor is used to construct a set of mutually orthogonal spreading code sequences and distribute the spreading code sequences to each end node in the power network system through the communication unit.

[0065] The end node is used to respond to the synchronous triggering unit receiving the spreading code sequence, use the zero-crossing detection circuit to capture the voltage zero-crossing point to lock a unified time reference, and control the power modulation circuit to synchronously perform modulation actions on the electrical physical quantities of the connection point according to its spreading code sequence logic, thereby generating a hybrid characteristic signal on the power line.

[0066] The acquisition module is used to acquire superimposed composite waveform data on power lines at a preset sampling rate; the superimposed composite waveform data includes power network background signals and the mixed feature signals;

[0067] The processor is also configured to execute a computer program in the edge computing environment to preprocess the superimposed composite waveform data, extract feature sequences containing modulation features, iteratively process the feature sequences, parse the signal component strengths corresponding to each spreading code sequence, and determine the physical topology affiliation between each end node and the monitoring port of the central edge computing node based on the signal component strengths.

[0068] As can be seen from the above technical solution, the beneficial technical effects of the present invention are as follows:

[0069] 1. By introducing orthogonal spread spectrum coding technology, it supports a large number of end nodes to perform parallel physical modulation at the same time, breaking the bottleneck of the traditional time-division multiplexing polling mechanism and realizing real-time perception of the network topology of the entire distribution area.

[0070] 2. By employing serial interference cancellation and differential preprocessing algorithms, it can not only effectively filter out strong power frequency fundamental wave interference, but also accurately extract weak feature signals in environments with low signal-to-noise ratio and waveform distortion.

[0071] 3. The multi-port competition decision logic effectively eliminates phase or branch misjudgment caused by inter-line electromagnetic coupling; at the same time, the low-power spread spectrum modulation method avoids voltage flicker pollution to the power grid, ensuring the stability of power grid operation. Attached Figure Description

[0072] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0073] Figure 1 This is a flowchart of the method in Embodiment 1 of the present invention;

[0074] Figure 2 This is a waveform example diagram of Embodiment 1 of the present invention. Figure 2 (a) is a power frequency voltage waveform diagram of Embodiment 1 of the present invention. Figure 2 (b) is a waveform diagram of the modulation action of node A in Embodiment 1 of the present invention. Figure 2 (c) is a waveform diagram of the modulation action of node B in Embodiment 1 of the present invention. Figure 2 (d) is a superimposed waveform diagram of Embodiment 1 of the present invention. Figure 2 (e) is a feature sequence diagram of Embodiment 1 of the present invention;

[0075] Figure 3 The actual synthesized waveform and the fundamental reference waveform after superposition features in Embodiment 1 of the present invention are shown in the figure.

[0076] Figure 4 This is a system structure diagram of Embodiment 2 of the present invention. Detailed Implementation

[0077] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0078] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0079] Example 1

[0080] The working principle of Embodiment 1 is described in detail below. This embodiment provides a power network topology identification method based on an intelligent fusion terminal. The method relies on an intelligent fusion terminal deployed on the distribution transformer side as a central edge computing node, and several end nodes installed on each distributed power source or load side.

[0081] The method flowchart of this embodiment is as follows: Figure 1 As shown, it includes the following steps:

[0082] The central edge computing node constructs a set of mutually orthogonal spreading code sequences and distributes the spreading code sequences to each end node in the power network system.

[0083] In response to the end node receiving the spreading code sequence, the modulation start time is locked according to a unified time base;

[0084] Each of the terminal nodes performs modulation operations on the electrical physical quantities of the connection point according to its spreading code sequence logic, generating a mixed characteristic signal on the power line composed of modulation signals from multiple terminal nodes.

[0085] The central edge computing node collects superimposed and synthesized waveform data on the power line at a preset sampling rate;

[0086] The superimposed composite waveform data includes power network background signals and the mixed characteristic signals;

[0087] The superimposed waveform data is preprocessed to extract feature sequences containing modulation features;

[0088] In the edge computing environment, the central edge computing node iteratively processes the feature sequence and analyzes the signal component strength corresponding to each spreading code sequence.

[0089] The physical topology affiliation between each end node and the monitoring port of the central edge computing node is determined based on the signal component strength.

[0090] To enable those skilled in the art to better understand the present invention, the following will be combined with... Figure 1 The technical solutions in the embodiments of the present invention will be clearly and completely described.

[0091] In this embodiment, further, during the system initialization or topology change phase, the intelligent fusion terminal first assesses the number M of online end nodes under the current distribution area.

[0092] The terminal constructs a Hadamard matrix of order N (N≥M, and N is a power of 2). Each row of the matrix constitutes an orthogonal spreading code sequence, and the elements in the matrix consist of +1 and -1. The first row of the Hadamard matrix is ​​unbalanced, which would cause a continuous voltage drop. Therefore, in this embodiment, the first row is removed and not assigned to any terminal node.

[0093] The spreading code sequence is a code group with positive and negative balance characteristics. At any modulation time, the number of end nodes performing modulation actions and the number of end nodes remaining silent are maintained within a preset balance ratio range. When a large number of nodes operate simultaneously, although the physical signals are superimposed, the total voltage fluctuation after superposition tends to remain at a low level due to the statistical positive and negative balance of the selected orthogonal spreading code sequence, thus ensuring that the power quality of the power grid is less affected.

[0094] In this embodiment, the terminal further distributes the remaining N-1 rows of orthogonal spreading code sequences to each end node sequentially through the existing HPLC communication network of the substation. Upon receiving the instruction, the end node does not act immediately but instead starts a local timer.

[0095] The waveform example diagram of this embodiment is shown below. Figure 2 As shown, node A and node B are arbitrarily selected from the terminal nodes, and four power line frequency cycles (20ms per cycle) are selected as an example. Figure 2 (a) is a waveform diagram of the power frequency voltage. The zero-crossing detection circuit of the terminal node captures the physical zero-crossing point of the power line voltage in real time. To prevent zero-crossing jitter caused by power grid harmonics, the terminal node uses a moving average algorithm to calculate the average period of the most recent 20 power frequency cycles. Based on the physical zero-crossing point and average period of the most recently captured power frequency voltage, the theoretical start time of the lower k cycles is calculated:

[0096]

[0097] in, Indicates the start time. Indicates the physical zero-crossing point. Indicates the average period.

[0098] All end nodes initiate modulation using this theoretical start time as a reference point. Node A is assigned the spreading code logic sequence [1,-1,1,-1], while node B is assigned the sequence [1,1,-1,-1]. The modulation actions of node A and node B are as follows: Figure 2 (b) and Figure 2 As shown in (c).

[0099] In this embodiment, further, the modulation operation on the electrical physical quantities of the connection point specifically includes:

[0100] The end node controls the internal disturbance element (in this embodiment, a bypass resistor) to switch between different states according to the logic level in the spreading code sequence, in order to generate voltage amplitude fluctuations on the power line.

[0101] If the current bit of the spreading code is logic 1, it controls the internal disturbance element to conduct, causing a slight voltage drop in the power line.

[0102] If the current bit is logic -1, it remains in the disconnected state without disturbance.

[0103] In other embodiments, the disturbance element includes a bypass resistor, a variable impedance element, a switchable load element, or an auxiliary power supply element.

[0104] In this embodiment, the intelligent fusion terminal further samples the voltage of the power line continuously using a high-frequency ADC to obtain superimposed and synthesized waveform data.

[0105] Because all nodes operate simultaneously, the actual waveform on the power line is a linear superposition of the modulated signals from all nodes. Adding the original background signal of the power grid (fundamental frequency + noise), the resulting composite waveform is shown below. Figure 2 As shown in (d).

[0106] During the first period (0-20ms), the symbol logic of both nodes A and B is 1. Both nodes conduct their bypass resistors, resulting in a superimposed enhancement. This causes the actual synthesized waveform to drop relative to the fundamental reference waveform due to the disturbance caused by the two nodes. For example... Figure 2 In the superimposed waveform diagram shown in (d), the actual waveform reaches its peak value at approximately 5ms in the first period, as follows: Figure 3 The figure shows the actual composite waveform after superposition features and the fundamental reference waveform. It can be seen that the actual waveform after superposition features (i.e., the composite waveform) is lower than the fundamental reference waveform (i.e., the power frequency voltage waveform).

[0107] During the second and third cycles (20ms-60ms), the symbol logic of node A is opposite to that of node B, resulting in a voltage drop caused by the disturbance of only one node.

[0108] During the fourth cycle (60ms-80ms), the symbol logic of both node A and node B is -1. Both nodes remain silent and do not conduct bypass resistors. The power line voltage returns to the fundamental state, and no additional voltage drop occurs.

[0109] In this embodiment, further, after the central edge computing node acquires the aforementioned superimposed composite waveform data at a preset sampling rate, it further performs a preprocessing step using digital signal processing methods to extract, such as Figure 2(e) shows a feature sequence containing modulation features, specifically including:

[0110] The central edge computing node first uses the power line frequency reference monitored in real time to remove the 50Hz power frequency fundamental component from the original superimposed composite waveform through a periodic difference algorithm. Furthermore, the algorithm performs normalization mapping processing on the superimposed composite waveform data after removing the power frequency component, converting it from physical voltage fluctuations into logically matched feature values.

[0111] In this normalization mapping process, the system establishes a zero-mean mathematical mapping relationship, such as... Figure 2 The feature sequence shown in (e) can accurately reflect the spreading code logic of each node.

[0112] During the first cycle (0-20ms), due to voltage drops at both nodes A and B, the preprocessed feature sequence exhibits the strongest positive pulse with a normalized amplitude of approximately 2 units, representing the superposition enhancement feature of logic 1 at both nodes. During the second to third cycles (20ms-60ms), due to the out-of-direction logic of the two nodes, only the drop feature of a single node is retained on the power line. At this time, the amplitude of the feature sequence is halved relative to the first cycle, reflecting the initial cancellation effect of orthogonal codes at the physical level. During the fourth cycle (60ms-80ms), since neither node conducts bypass resistors, there is no actual waveform drop. After normalization mapping, this state is mapped to the negative maximum amplitude of the feature sequence, thus ensuring the mathematical zero-mean characteristic of the entire feature sequence.

[0113] Through the above preprocessing steps, the system transforms the weak physical fluctuations submerged in strong power frequency noise into feature sequences with significantly improved signal-to-noise ratios. These sequences serve as input data for subsequent iterative processing, enabling central and edge computing nodes to identify the signal component strengths corresponding to each spreading code sequence at the edge with relatively low computational cost.

[0114] The dual-node modulation consisting of node A and node B described above aims to clearly demonstrate the basic principles of orthogonal superposition and physical cancellation in this invention. In practical low-voltage distribution substation applications, the number of end nodes typically reaches hundreds or even thousands. At this time, the mixed characteristic signals on the power line physically appear as a linear superposition of numerous tiny voltage drops.

[0115] In terms of the numerical distribution of the feature sequence, as the number of parallel nodes increases, when hundreds of end nodes simultaneously perform modulation actions according to their respective spreading codes, the feature sequence will exhibit continuous fluctuation characteristics in the time domain, and its amplitude range will dynamically change with the number of simultaneously activated nodes. In such a highly complex background signal, traditional threshold detection can no longer distinguish the characteristics of any single node.

[0116] Therefore, in this embodiment, the feature sequence is further iteratively processed to analyze the signal component strength corresponding to each spreading code sequence. Specific steps include:

[0117] The central edge computing node performs parallel cross-correlation operations between the preprocessed feature sequence and the locally pre-stored spreading code library to lock the target spreading code with the largest correlation peak, and uses the absolute value of the correlation peak as the signal component strength of the target spreading code.

[0118] Because the spreading codes of different end nodes have good orthogonality, a significant correlation peak will only be generated when the code sequence of the target node is perfectly aligned with the components in the feature sequence. The algorithm locks the spreading code with the largest absolute value of the correlation peak and defines the absolute value of its corresponding correlation peak as the signal component strength of that node. This strength directly reflects the physical coupling gain between the end node and the current monitoring port.

[0119] In this embodiment, further, in order to extract accurate physical features from complex power grid load fluctuations, the current superimposed composite waveform data is segmented into multiple symbol segments according to the symbol timing of the target spreading code;

[0120] All code segments corresponding to logic 1 are superimposed and averaged to generate a positive feature template;

[0121] All code segments corresponding to logic -1 are superimposed and averaged to generate a negative feature template;

[0122] The steps for performing an overlay averaging include:

[0123] Extract K code segments belonging to the same logical state from the superimposed and synthesized waveform data;

[0124] The initial reference vector is obtained by performing an arithmetic average of all the aforementioned symbol segments:

[0125]

[0126] in, Represents the initial reference vector. Indicates the current code element segment;

[0127] Considering the presence of random impulse noise in the power line environment, if a segment is severely contaminated by noise, its waveform will deviate from the reference vector. Therefore, the Pearson correlation coefficient between each symbol segment and the initial reference vector is calculated as the weight of the current symbol segment. This weight represents the similarity between the segment and common features; the more similar the segment is to common features, the higher the weight; the more distorted the waveform, the lower the weight. The specific formula is as follows:

[0128]

[0129] in, This indicates the weight of the current code segment. Indicates the current code segment, express The average of all elements in the set. Represents the initial reference vector. This represents the average of all initial reference vectors. This represents the L2 norm of a vector.

[0130] The weight in response to the current symbol segment If the value is less than the preset validity threshold, then... Set to zero to remove abnormal segments contaminated by sudden impulse noise.

[0131] Using the weights of the current symbol segment, a weighted average is calculated on the symbol segments to generate a feature template:

[0132]

[0133] in, Represents a feature template. This indicates the weight of the current code segment. This indicates the current code segment.

[0134] After obtaining the positive and negative feature templates of the target node, the system splices the templates according to the timing logic of the target spreading code to reconstruct the complete modulation signal waveform of the node in the current period. Subsequently, the reconstructed waveform is subtracted from the current feature sequence to obtain the residual waveform data.

[0135] After each round of the above calculations, the system calculates the root mean square value of the residual waveform data as the noise basis for the current iteration round;

[0136] Calculate the ratio of the signal component intensity recorded in the current iteration to the noise floor:

[0137]

[0138] in, Indicates the signal-to-noise ratio. Indicates the signal component strength. This represents the noise floor.

[0139] If the ratio is greater than or equal to a preset minimum signal-to-noise ratio threshold or the current iteration count reaches a preset maximum iteration count, it is determined that the termination condition is not met.

[0140] If the termination condition is not met, it means that there are still valid node signals that can be parsed hidden in the data. The residual waveform data is then updated to the current feature sequence, the next iteration begins, and the process returns to the step of performing parallel correlation operations.

[0141] In response to the ratio being less than a preset minimum signal-to-noise ratio threshold (indicating that the rest are noise) or the current iteration count reaching a preset maximum iteration count, it is determined that the termination condition is met.

[0142] The iteration is terminated and all spreading codes and their corresponding signal component strengths parsed within the cycle are output.

[0143] In this embodiment, further, determining the physical topology affiliation between each end node and the monitoring port of the central edge computing node based on the signal component strength specifically includes:

[0144] For the identification results of a single monitoring port, the system reads the signal component intensity corresponding to each spreading code sequence within the current identification period, and calculates the ratio of this intensity value to the original total energy of the preprocessed feature sequence, thereby obtaining the energy proportion coefficient.

[0145] In response to the absolute value of the signal component strength of the spreading code sequence being greater than a preset hard decision threshold and its energy proportion coefficient being greater than a preset correlation coefficient threshold, it is determined that the end node of the spreading code sequence is physically connected to the monitoring port of the current central edge computing node.

[0146] If the signal component strength or the energy ratio coefficient does not meet the threshold condition, it is determined that the end node is not connected to the current monitoring port.

[0147] In complex transformer substation scenarios with multiple phases or multiple circuits laid in parallel, the central edge computing node is usually configured with multiple monitoring ports. Due to the electromagnetic coupling between power lines, the characteristic signal modulated by a certain end node may crosstalk to unconnected adjacent lines through mutual inductance, causing multiple monitoring ports to detect the signal component of the spreading code.

[0148] Therefore, in this embodiment, the multi-port contention determination logic is further executed. Specific steps include: the central edge computing node simultaneously collects data from all monitoring ports and independently performs the aforementioned iterative processing to construct an intensity feature vector for the same end node under different monitoring ports. The algorithm identifies the maximum and second-largest values ​​in this intensity feature vector in real time and calculates the ratio of their differences.

[0149] If the maximum strength exceeds the preset effective signal threshold, and the difference between the maximum and the second largest value exceeds the preset crosstalk suppression tolerance, it indicates that the signal strength at a specific port is dominant and sufficient to eliminate inter-line coupling interference. In this case, the terminal node is determined to belong uniquely to the monitoring port corresponding to the maximum value, thereby accurately completing the identification of phase or specific branch affiliation.

[0150] If the ratio of the difference between the maximum and second-largest values ​​is lower than the crosstalk suppression tolerance, it indicates that the signal strength detected by multiple ports is similar, and it is impossible to clearly distinguish between the actual connection signal and the induced crosstalk signal at the physical level. In this case, the end node is determined to be in a signal crosstalk region or a cross-phase coupling region. In this situation, the system will not perform a final attribution determination and will generate an abnormal alarm message to prompt maintenance personnel that there is serious electromagnetic coupling or wiring abnormality in this area.

[0151] Through this mechanism, this embodiment solves the identification drift and phase misjudgment problems that occur in traditional topology identification methods when multiple cables are laid side by side, and improves the determinism of transformer area topology mapping.

[0152] To further eliminate single misjudgments caused by random non-stationary noise such as the start-up and shutdown of high-power inductive loads and lightning-induced pulses within the transformer area, this embodiment also includes a time-domain statistical stability verification step.

[0153] After determining the initial affiliation of the end nodes, the central edge computing node does not immediately write it into the global topology database. Instead, it continuously executes M identification cycles (set to 10 power frequency cycles in this embodiment) and establishes a topology status cache queue for each end node.

[0154] The system updates the queue in real time based on the preliminary judgment results output for each recognition cycle.

[0155] Subsequently, the algorithm uses a majority voting mechanism for final confirmation, the specific steps of which include:

[0156] For each end node, count the number of times (K) it is stably identified as belonging to a specific monitoring port within M identification cycles of the cache queue.

[0157] The central edge computing node calculates the ratio of this number of times to the total number of cycles.

[0158] In response to the ratio being greater than the preset stability probability threshold, the system determines that the physical connection relationship of the end node has a high degree of consistency on the time axis, thereby formally confirming and locking the physical topology affiliation relationship, and updating it to the topology mapping view of the substation fusion terminal.

[0159] If the ratio is lower than the threshold, it indicates that the signal of the node fluctuates or drifts instantaneously on the time axis. The system determines that the determination result is affected by occasional noise interference. At this time, the system chooses to maintain the topology state of the previous moment unchanged until the stability criterion is met in subsequent cycles.

[0160] Through the above embodiments, the present invention utilizes the edge computing capabilities of intelligent fusion terminals to achieve high-precision and high-efficiency power network topology identification.

[0161] Example 2

[0162] In conjunction with the method described in Embodiment 1, Embodiment 2 provides a power network topology identification system based on an intelligent fusion terminal.

[0163] The system structure diagram of this embodiment is as follows: Figure 4 As shown, the specific structure includes:

[0164] The intelligent converged terminal, as a central edge computing node, includes a data acquisition module, a communication unit, and a processor, wherein the processor provides an edge computing environment.

[0165] Multiple end nodes are distributed on the branches or user side of the power network, and each end node is equipped with a synchronization triggering unit, a zero-crossing detection circuit and a power modulation circuit.

[0166] The processor is used to construct a set of mutually orthogonal spreading code sequences and distribute the spreading code sequences to each end node in the power network system through the communication unit.

[0167] The end node is used to respond to the synchronous triggering unit receiving the spreading code sequence, use the zero-crossing detection circuit to capture the voltage zero-crossing point to lock a unified time reference, and control the power modulation circuit to synchronously perform modulation actions on the electrical physical quantities of the connection point according to its spreading code sequence logic, thereby generating a hybrid characteristic signal on the power line.

[0168] The acquisition module is used to acquire superimposed composite waveform data on power lines at a preset sampling rate; the superimposed composite waveform data includes power network background signals and the mixed feature signals;

[0169] The processor is also configured to execute a computer program in the edge computing environment to preprocess the superimposed composite waveform data, extract feature sequences containing modulation features, iteratively process the feature sequences, parse the signal component strengths corresponding to each spreading code sequence, and determine the physical topology affiliation between each end node and the monitoring port of the central edge computing node based on the signal component strengths.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A power network topology identification method based on an intelligent fusion terminal, characterized in that, The method operates in a power network system containing a central edge computing node and multiple end nodes, and includes the following steps: The central edge computing node constructs a set of mutually orthogonal spreading code sequences and assigns the spreading code sequences to each of the end nodes in the power network system. In response to the end node receiving the spreading code sequence, the modulation start time is locked according to a unified time base; Each of the terminal nodes performs modulation on the electrical physical quantities of the connection point according to its spreading code sequence logic, generating superimposed composite waveform data on the power line composed of multiple terminal node modulation signals and power network background signals. The central edge computing node collects the superimposed composite waveform data on the power line at a preset sampling rate; The superimposed waveform data is preprocessed to extract feature sequences containing modulation features; In the edge computing environment, the central edge computing node iteratively processes the feature sequence and analyzes the signal component strength corresponding to each spreading code sequence. The physical topology affiliation between each of the terminal nodes and the monitoring port of the central edge computing node is determined based on the signal component strength.

2. The method according to claim 1, characterized in that, The unified time base is determined by the following steps: Each terminal node uses the zero-crossing point of the power line voltage waveform as a reference point for a common time base; The end node initiates modulation at the Nth reference point after receiving the spreading code sequence.

3. The method according to claim 1, characterized in that, The spreading code sequence logic specifically includes: The duration of each symbol in the spreading code sequence is configured to be an integer multiple of the power line frequency period or an integer multiple of half a period. The spreading code sequence is configured such that, at any modulation time, the number of end nodes performing modulation actions and the number of end nodes remaining silent are maintained within a preset balance ratio range.

4. The method according to claim 1, characterized in that, Modulation of the electrical physical quantities at the connection point, specifically including: The end node controls the impedance elements, load elements and / or power supply elements connected inside it to switch between different states according to the logic level in the spreading code sequence. The modulation action is configured as a low-power modulation mode, so that the signal fluctuation amplitude generated by a single end node is lower than the base amplitude of the power line background noise.

5. The method according to claim 1, characterized in that, The superimposed waveform data is preprocessed to extract feature sequences containing modulation features, including: The power frequency component in the superimposed and synthesized waveform data is removed based on the power line power frequency reference. Normalization mapping is performed on the superimposed composite waveform data after removing the power frequency component to generate a feature sequence that logically matches the spreading code sequence.

6. The method according to claim 1, characterized in that, The feature sequence is iteratively processed to analyze the signal component strength corresponding to each spreading code sequence. Specific steps include: Perform parallel correlation operations between the current feature sequence and all unidentified spreading code sequences to lock the target spreading code with the largest correlation peak, and use the absolute value of the correlation peak as the signal component strength of the target spreading code; Based on the symbol timing of the target spreading code, the current feature sequence is divided into multiple symbol segments; All code segments corresponding to logic 1 are superimposed and averaged to generate a positive feature template; All code segments corresponding to logic -1 or 0 are superimposed and averaged to generate a negative feature template; According to the timing logic of the target spreading code, the positive feature template and the negative feature template are concatenated to generate the reconstructed waveform of the target end node; Subtract the reconstructed waveform from the current feature sequence to obtain residual waveform data; In response to the residual waveform data not meeting the preset termination condition, the residual waveform data is updated to the current feature sequence, and the parallel correlation operation and subsequent steps are repeated. In response to the residual waveform data satisfying the preset termination condition, the iteration is terminated and all parsed target spreading codes and their corresponding signal component strengths are output. The preset termination conditions include: The root mean square value of the residual waveform data is calculated as the noise basis for the current iteration. Calculate the ratio of the signal component intensity recorded in the current iteration to the noise floor; In response to the ratio being less than a preset minimum signal-to-noise ratio threshold or the current iteration number reaching a preset maximum iteration number, it is determined that the termination condition is met; If the ratio is greater than or equal to a preset minimum signal-to-noise ratio threshold or the current iteration count reaches a preset maximum iteration count, it is determined that the termination condition is not met.

7. The method according to claim 6, characterized in that, The steps for performing an overlay averaging include: Extract K code segments belonging to the same logical state from the feature sequence; The initial reference vector is obtained by performing an arithmetic average on all the aforementioned symbol segments; Calculate the correlation coefficient between each symbol segment and the initial reference vector as the weight of the current symbol segment: in, This indicates the weight of the current code segment. Indicates the current code segment, Indicates the current code segment The average of all elements in the set. Represents the initial reference vector. This represents the average of all initial reference vectors. The L2 norm of a vector; A feature template is generated by weighting the current symbol segment using the weights of the current symbol segment: in, Represents a feature template. This indicates the weight of the current code segment. This indicates the current code segment.

8. The method according to claim 1, characterized in that, The physical topology affiliation between each end node and the monitoring port of the central edge computing node is determined based on the signal component strength, specifically including: Read the intensity of all signal components in the current identification period; For each spreading code sequence, the ratio of its corresponding signal component intensity to the original total energy of the feature sequence is calculated to obtain the energy proportion coefficient; In response to the absolute value of the signal component strength of the spreading code sequence being greater than a preset hard decision threshold and its energy proportion coefficient being greater than a preset correlation coefficient threshold, it is determined that the end node of the spreading code sequence is physically connected to the current monitoring port of the central edge computing node. If the signal component strength or the energy ratio coefficient does not meet the threshold condition, it is determined that the end node is not connected to the current monitoring port.

9. The method according to claim 1, characterized in that, When the central edge computing node includes multiple monitoring ports, the physical topology affiliation between each end node and the monitoring port of the central edge computing node is determined, which also includes multi-port contention determination. The specific steps include: The central edge computing node simultaneously collects the feature sequences of the multiple monitoring ports and independently executes the iterative processing to construct an intensity feature vector for the same end node under different monitoring ports; Identify the maximum and second largest values ​​in the intensity feature vector, and calculate the ratio of the difference between the maximum and the second largest values; In response to the maximum value exceeding a preset effective signal threshold and the difference ratio exceeding a preset crosstalk suppression tolerance, it is determined that the end node uniquely belongs to the monitoring port corresponding to the maximum value; If the difference ratio is lower than the crosstalk suppression tolerance, the terminal node is determined to be in the signal crosstalk region, no attribution determination is performed, and an abnormal alarm is generated.

10. A power network topology identification system based on an intelligent fusion terminal, characterized in that, include: The intelligent converged terminal, as a central edge computing node, includes a data acquisition module, a communication unit, and a processor, wherein the processor provides an edge computing environment. Multiple end nodes are distributed on the branches or user side of the power network, and each end node is equipped with a synchronization triggering unit, a zero-crossing detection circuit and a power modulation circuit. The processor is used to construct a set of mutually orthogonal spreading code sequences and distribute the spreading code sequences to each end node in the power network topology identification system through the communication unit. The end node is used to respond to the synchronous triggering unit receiving the spreading code sequence, use the zero-crossing detection circuit to capture the voltage zero-crossing point to lock a unified time reference, and control the power modulation circuit to synchronously perform modulation actions on the electrical physical quantities of the connection point according to its spreading code sequence logic, thereby generating a hybrid characteristic signal on the power line. The acquisition module is used to acquire superimposed composite waveform data on power lines at a preset sampling rate; the superimposed composite waveform data includes power network background signals and the mixed feature signals; The processor is also configured to execute a computer program in the edge computing environment to preprocess the superimposed composite waveform data, extract feature sequences containing modulation features, iteratively process the feature sequences, parse the signal component strengths corresponding to each spreading code sequence, and determine the physical topology affiliation between each end node and the monitoring port of the central edge computing node based on the signal component strengths.

Images