Power distribution terminal fault detection method and system based on data processing
By employing noise suppression and signal enhancement techniques, combined with traveling wave and impedance-guided ranging methods, adaptive trigger state backtracking, and reliability assessment, the accuracy and reliability issues of fault location in distribution terminals in complex power grid environments have been resolved, achieving higher precision and more stable fault location.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI LANJIAN ELECTRIC EQUIP CO LTD
- Filing Date
- 2026-03-18
- Publication Date
- 2026-05-29
AI Technical Summary
Existing fault location methods for distribution terminals are susceptible to noise interference in complex power grid environments, resulting in insufficient accuracy and reliability in fault location.
Real-time sampling data is acquired through interactive power distribution terminals, noise suppression and signal enhancement are implemented, line fault probability is calculated, state change chain is constructed by adaptively triggering state backtracking, and fault location is guided by traveling wave and impedance. Reliable measurement and reconstruction of associated signals are also performed to obtain more accurate fault location results.
It improves the accuracy and stability of fault location, ensuring stable and reliable fault location results in complex power grid environments, and enhances response speed and flexibility.
Smart Images
Figure CN122109680A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and more specifically to a method and system for detecting faults in power distribution terminals based on data processing. Background Technology
[0002] As a critical piece of equipment in a power system, the accuracy of fault detection and location in distribution terminals directly impacts the stability and reliability of the power system. With the continuous development of power systems, the requirements for fault location technology in distribution terminals are becoming increasingly stringent. Traditional fault location methods typically rely on impedance measurement or traveling wave analysis. However, in practical applications, due to the complexity of the power system environment, these methods are easily affected by noise and interference, leading to reduced location accuracy and making it difficult to meet the needs of modern power systems. Summary of the Invention
[0003] This application provides a data processing-based method and system for detecting faults in distribution terminals, which addresses the technical problems of insufficient accuracy and reliability of existing fault location methods in complex power grid environments due to noise interference.
[0004] The first aspect of this application provides a data processing-based method for fault detection in a distribution terminal. The method includes: an interactive distribution terminal acquiring real-time sampled data of a target line, performing noise suppression and signal enhancement on the real-time sampled data to generate a real-time line state sequence; calculating a line fault probability based on the real-time line state sequence, and adaptively triggering a backtracking of the reference neighbor state of the real-time line state sequence based on the line fault probability to construct a line state change chain; performing traveling wave-guided fault location based on the line state change chain to obtain a first fault location result, and simultaneously performing impedance-guided fault location based on the line state change chain to obtain a second fault location result; performing correlation signal reliability quantification on the first and second fault location results to determine a first location reliability and a second location reliability; and reconstructing the location result reliability based on the first and second location reliability according to a predetermined reliability to obtain a reliable fault location result.
[0005] A second aspect of this application provides a data processing-based power distribution terminal fault detection system, the system comprising: a real-time data sampling module, used to interactively acquire real-time sampling data of a target line from a power distribution terminal, and to perform noise suppression and signal enhancement on the real-time sampling data to generate a real-time line state sequence; a neighboring point state backtracking module, used to calculate the line fault probability based on the line real-time state sequence, and to adaptively trigger the reference neighboring point state backtracking of the line real-time state sequence based on the line fault probability to construct a line state change chain; a fault location module, used to perform traveling wave guided fault location based on the line state change chain to obtain a first fault location result, and simultaneously perform impedance guided fault location based on the line state change chain to obtain a second fault location result; a signal reliability quantification module, used to perform associated signal reliability quantification on the first fault location result and the second fault location result to determine a first location reliability and a second location reliability; and a location result reconstruction module, used to reconstruct the location result reliability based on the first location reliability and the second location reliability, according to a predetermined reliability, to obtain a reliable fault location result.
[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: The data processing-based fault detection method and system for distribution terminals provided in this application pertain to the field of data processing technology. It acquires real-time data through an interactive distribution terminal, implements noise suppression and signal enhancement, calculates the probability of line faults, and adaptively triggers state backtracking to construct a state change chain. Combining traveling wave and impedance-guided ranging, it performs reliable quantification and reconstruction of the ranging results to obtain more accurate fault ranging results. This solves the technical problems of existing distribution terminal fault ranging methods being susceptible to noise interference in complex power grid environments, resulting in insufficient accuracy and reliability in fault location. It achieves the technical effect of improving fault ranging accuracy and stability through noise suppression and signal enhancement, along with reliability assessment and result reconstruction. Attached Figure Description
[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0008] Figure 1 This is a schematic flowchart of a power distribution terminal fault detection method based on data processing provided in an embodiment of this application. Figure 2 This is a schematic diagram of the structure of a power distribution terminal fault detection system based on data processing, provided in an embodiment of this application.
[0009] Figure labeling: Real-time data sampling module 11, neighboring point status backtracking module 12, fault ranging module 13, signal reliability quantification module 14, ranging result reconstruction module 15. Detailed Implementation
[0010] This application provides a data processing-based method and system for detecting faults in distribution terminals, which addresses the technical problems of insufficient accuracy and reliability of existing fault location methods in complex power grid environments due to noise interference.
[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.
[0012] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.
[0013] Example 1, as Figure 1 As shown, this application provides a data processing-based method for detecting faults in power distribution terminals, the method comprising: P10: The interactive power distribution terminal acquires real-time sampling data of the target line, performs noise suppression and signal enhancement on the real-time sampling data, and generates a real-time status sequence of the line.
[0014] Furthermore, step P10 in this embodiment of the application also includes: P11: Perform signal feature analysis on the real-time sampled data to obtain a sampled signal feature set; P12: Based on the sampled signal feature set, drive the automatic gain controller to perform noise suppression and signal enhancement on the real-time sampled data to obtain the real-time state sequence of the line.
[0015] It should be understood that the interactive distribution terminal first acquires real-time sampled data from the target line, including parameters such as current, voltage, and frequency. These parameters reflect the electrical state of the line under normal operation or fault conditions. However, directly using this raw data for subsequent processing may be affected by noise and interference. Therefore, signal feature analysis is required to extract key features that characterize the line's operating state from the complex sampled data. Through signal feature analysis, effective and noise components in the signal can be identified, and information such as voltage amplitude, phase angle, and frequency fluctuations can be extracted to form a sampled signal feature set.
[0016] A sampled signal feature set is an abstract representation of the original signal, reflecting important information within it. For example, by performing a Fourier transform on a time-domain signal, the frequency-domain features can be obtained, further identifying anomalous frequency components. These features are then used to guide subsequent signal processing, effectively improving signal quality.
[0017] Next, based on the obtained feature set of the sampled signal, an automatic gain controller (AGC) is used to suppress noise and enhance the signal of the real-time sampled data. According to the information in the feature set of the sampled signal, the AGC adjusts the gain value during signal processing. In this process, the AGC dynamically adjusts the gain based on different signal characteristics, such as amplitude and frequency, aiming to maximize the strength of the effective signal while suppressing noise. Specifically, when the noise in the signal is strong, the AGC reduces the gain to prevent further amplification of the noise; while when the signal itself is weak, the gain is appropriately increased to enhance the clarity of the effective signal.
[0018] During this process, the automatic gain controller continuously monitors the signal strength and dynamically adjusts it according to a preset algorithm. Gain control is not just a simple increase or decrease; it also includes real-time analysis of signal characteristics to ensure that the signal enhancement process does not introduce additional noise. In this way, noise components in the real-time sampled data are effectively suppressed, while the effective components of the signal are enhanced, ultimately resulting in a clear real-time line status sequence with a high signal-to-noise ratio, providing high-quality data support for subsequent fault detection and ranging analysis.
[0019] P20: Calculate the line fault probability based on the real-time state sequence of the line, and adaptively trigger the baseline neighbor state backtracking of the real-time state sequence of the line based on the line fault probability to construct the line state change chain.
[0020] Furthermore, based on the real-time status sequence of the line, the probability of line fault is calculated. In this embodiment, step P20 further includes: P21: Based on the power distribution terminal, retrieve the line fault log database of the target line and randomly perturb the line fault log database to obtain a line fault perturbation database; P22: Perform iterative supervised learning on a Bayesian network based on the line fault perturbation database to obtain a first fault probability prediction model; P23: Perform iterative supervised learning on a deep belief network based on the line fault perturbation database to obtain a second fault probability prediction model; P24: Distill and fuse the first fault probability prediction model and the second fault probability prediction model to obtain a line fault prediction node; P25: Input the real-time state sequence of the line into the line fault prediction node and output the line fault probability.
[0021] Optionally, the line fault probability is calculated based on the real-time state sequence of the line, and the baseline neighbor state backtracking of the real-time state sequence of the line is adaptively triggered based on the fault probability, thereby constructing the line state change chain.
[0022] To achieve this goal, firstly, fault log data is retrieved from the target line based on the distribution terminal. The fault log database contains historical fault information for the target line, including the time of occurrence, type, and location of the fault. To further enhance the accuracy and adaptability of the model, the system will perform certain perturbation processing on this historical fault data to generate a line fault perturbation database. Specifically, perturbation processing simulates different fault scenarios by randomly adjusting variables such as the time, type, and location of the fault data. In this application, randomly perturbing the data in the line fault log database can increase data diversity and improve the model's generalization ability to different fault conditions. For example, the fault occurrence time can be randomly shifted, the fault location can be randomly perturbed, or fault-related parameters can be randomly adjusted to generate the line fault perturbation database. Such perturbation can make the dataset more diverse, thereby improving the generalization ability of the fault prediction model and helping the model adapt to more different fault scenarios.
[0023] Then, the Bayesian network undergoes iterative supervised learning based on a line fault disturbance database. A Bayesian network is a machine learning method based on a probabilistic graphical model that represents causal relationships between variables by constructing probabilistic dependencies between nodes. The Bayesian network learns the correlation between different fault types and line states based on disturbed fault data. During training, the Bayesian network repeatedly adjusts its internal structure and parameters, gradually optimizing the prediction accuracy of fault probabilities. For example, data from the line fault disturbance database is used as training samples, and the parameters of the Bayesian network are iteratively optimized so that the network can accurately predict fault probabilities based on the input line state information. After multiple iterative learning iterations, a first fault probability prediction model is obtained, which can output the probability distribution of line faults based on line state information.
[0024] Simultaneously, a second fault probability prediction model is obtained by iteratively supervising the deep belief network (DRM) based on a line fault disturbance database. The DRM is a deep learning model composed of multiple stacked Restricted Boltzmann Machines (RBMs), possessing powerful feature learning and representation capabilities. Compared to Bayesian networks, DRMs have stronger expressive power when processing data, enabling them to capture more complex fault modes and implicit patterns. In this application, data from the line fault disturbance database is also used to perform iterative supervised learning on the DRM. By training the RBMs layer by layer, the DRM can learn the complex nonlinear relationship between line state and fault probability. After multiple iterations, a second fault probability prediction model is obtained, which can output predicted fault probabilities based on line state information.
[0025] Next, the prediction results of the Bayesian network and the deep belief network are fused. Distillation fusion technology combines the advantages of multiple models to form a more powerful model. Through weighted averaging or other fusion methods, the system can integrate the prediction outputs of the two models to obtain a more accurate fault probability prediction. Specifically, the outputs of the first and second fault probability prediction models are used as inputs and processed through a fusion network, such as a simple weighted average or neural network fusion, to obtain a comprehensive fault probability prediction result. This fused result can fully utilize the probabilistic reasoning ability of the Bayesian network and the feature learning ability of the deep belief network, improving the accuracy and reliability of fault probability prediction. The final line fault prediction node can output a more accurate fault probability based on the input line state information.
[0026] Finally, the distilled and fused line fault prediction node receives the input real-time line status sequence and outputs a specific line fault probability based on the previously established model. This fault probability is a value between 0 and 1, representing the likelihood of a line fault occurring. Based on this probability, the system can determine whether a line is at risk of fault and provide decision support for subsequent fault location and handling.
[0027] Furthermore, based on the line fault probability, the baseline neighbor state backtracking of the real-time state sequence of the line is adaptively triggered to construct the line state change chain. In this embodiment, step P20 further includes: P26: If the line fault probability is greater than or equal to the fault probability threshold, construct the adjacent sampling time sequence corresponding to the real-time sampling time point of the line's real-time state sequence; P27: Sort the adjacent sampling time sequences from near to far to obtain the first adjacent sampling time point; P28: Perform state sequence backtracking on the target line based on the first adjacent sampling time point to obtain the first neighboring point line state sequence; P29: Perform fault probability analysis on the target line based on the first neighboring point line state sequence to obtain the first neighboring point fault probability; P210: Determine whether the first neighboring point fault probability is less than the fault probability threshold; P211: If the first neighboring point fault probability is less than the fault probability threshold, use the first neighboring point line state sequence as the benchmark neighboring point state backtracking result, and generate the line state change chain based on the benchmark neighboring point state backtracking result and the line's real-time state sequence.
[0028] Specifically, the system determines whether the calculated line fault probability is greater than or equal to a preset fault probability threshold. If the fault probability meets the condition, it indicates a high risk of line fault, requiring further detailed analysis. In this case, the system constructs an adjacent sampling time series. Specifically, the adjacent sampling time series consists of data corresponding to the current real-time sampling time point and multiple historical sampling time points before and after it. This data helps the system understand the line's operational status and changes over a past period, providing necessary historical evidence for subsequent backtracking.
[0029] Subsequently, these adjacent sampled time series are sorted from most recent to furthest to determine the first adjacent sampled time point. This sorting method ensures that, starting from the current time point, the most recent historical states are analyzed first to more accurately capture state changes before the fault occurred. In this way, the historical state closest to the current fault time point can be quickly located.
[0030] Next, based on the first adjacent sampling time point, the target line is backtracked in terms of its state sequence to extract the historical line state information corresponding to that time point, forming the first adjacent point line state sequence. This process, by retrieving line state data related to the first adjacent sampling time point, provides a data foundation for subsequent fault probability analysis. State sequence backtracking involves analyzing historical data to gradually trace the changes in line state in order to determine the critical states before the fault occurred.
[0031] Next, the line state sequence of the first neighboring point is input into the line fault prediction node for fault probability analysis. Specifically, the system inputs this historical state sequence into the aforementioned line fault prediction node to calculate the fault probability. The purpose of this process is to assess the likelihood of a line fault occurring under this historical state sequence, thereby determining whether the sequence reflects the actual fault risk.
[0032] Next, it is determined whether the failure probability of the first neighboring node is less than a failure probability threshold. This determination is a crucial step in deciding whether to continue state backtracking. If the failure probability is lower than the threshold, it indicates that the failure risk of the neighboring node's state sequence is low and can be used as a reliable reference. Conversely, if the failure probability is high, it indicates that the historical state sequence cannot accurately reflect the current failure risk.
[0033] If the failure probability of the first neighboring point is less than the failure probability threshold, the line state sequence of the first neighboring point is used as the baseline neighboring point state backtracking result, and combined with the real-time line state sequence, a line state change chain is generated. This process combines the baseline neighboring point state backtracking result with the current real-time line state sequence to construct a complete line state change chain. This change chain reflects the evolution of the line from a normal state to a fault state, helping the system track the line's state evolution from before the fault to after the fault, revealing the specific path of the fault occurrence.
[0034] Furthermore, in determining whether the failure probability of the first neighboring point is less than the failure probability threshold, step P20 of this embodiment further includes: P212: If the failure probability of the first neighboring point is greater than or equal to the failure probability threshold, read the second neighboring sampling time point according to the adjacent sampling time sequence; P213: Perform state sequence backtracking on the target line according to the second neighboring sampling time point to obtain the state sequence of the second neighboring point line; P214: Calculate the failure probability of the second neighboring point according to the state sequence of the second neighboring point line; P215: If the failure probability of the second neighboring point is less than the failure probability threshold, use the state sequence of the second neighboring point line as the baseline neighboring point state backtracking result.
[0035] In one possible embodiment of this application, when the failure probability of the first neighboring point is greater than or equal to the failure probability threshold, it indicates that the current historical state sequence may not effectively reflect the failure risk. Therefore, the system will read the next historical sampling time point, i.e., the second adjacent sampling time point, based on the adjacent sampling time series. The purpose of this process is to further expand the scope of analysis and find potentially more suitable reference data by looking at data further away from the current time point.
[0036] Next, the target line is backtracked based on the second adjacent sampling time point to obtain the second neighboring line state sequence. This step is similar to the first neighboring point state sequence backtracking; by retrieving line state data related to the second adjacent sampling time point, the line state information at that time point is extracted to form the second neighboring line state sequence. This state sequence contains the operating status and changes of the target line at earlier time points, further enriching the historical data for backtracking.
[0037] Subsequently, the line state sequence of the second neighboring point is input into the line fault prediction node to calculate the fault probability of the second neighboring point. This step utilizes a pre-trained fault probability prediction model to analyze the line state sequence of the second neighboring point and calculate the fault probability at that time point. The purpose of this process is to verify whether the second neighboring point is in a fault state, thereby providing a basis for backtracking the state of the baseline neighboring point.
[0038] Then, it is determined whether the calculated failure probability of the second neighboring point is less than a failure probability threshold. If the failure probability is less than the preset threshold, it means that the line state sequence of the second neighboring point can accurately reflect the risk of the failure and can be used as a new reliable reference. At this time, the system will use the line state sequence of the second neighboring point as the baseline neighboring point state backtracking result for subsequent analysis and construction of the line state change chain.
[0039] In this way, more historical data can be gradually traced back when the probability of failure is high, in order to find the state sequence that best reflects the risk of line failure. This step-by-step backtracking method can improve the accuracy of fault prediction and ensure that the system can accurately capture changes in the state of the line.
[0040] P30: Based on the line state change chain, perform traveling wave guided fault location to obtain the first fault location result, and simultaneously perform impedance guided fault location based on the line state change chain to obtain the second fault location result.
[0041] Furthermore, based on the line state change chain, traveling wave guided fault location is implemented to obtain the first fault location result. In this embodiment, step P30 further includes: P31: Perform high-frequency energy mutation detection based on the line state change chain to determine the arrival time of the initial traveling wave; P32: Perform IMF high-frequency secondary peak identification based on the line state change chain to determine the arrival time of the reflected traveling wave; P33: Perform energy ratio verification based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave to obtain the energy ratio verification result; P34: If the energy ratio verification result meets the energy ratio verification condition, activate the traveling wave fault location model; P35: Based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave, and according to the traveling wave fault location model, output the first fault location result.
[0042] It should be understood that traveling wave guided fault location is implemented based on the line status change chain, and impedance guided fault location is carried out simultaneously.
[0043] First, high-frequency energy abrupt change detection is performed based on the line state change chain. Traveling wave fault location typically determines the fault location by detecting energy changes in the signal. Whenever the initial traveling wave arrives at the fault point, the energy in the signal changes abruptly; this change is called a high-frequency energy abrupt change. By detecting the signal, the system can accurately identify the arrival time of the initial traveling wave. This time point marks the moment when the initial traveling wave reaches the target location, providing an important reference for subsequent fault localization.
[0044] Next, the arrival time of the reflected traveling wave is further determined by identifying the high-frequency secondary peaks of the IMF (Inductively Coupled Function) through the line state change chain. The reflected traveling wave is the signal reflected back when the original traveling wave encounters a fault point or other impedance changes; this reflected wave will have obvious high-frequency secondary peaks. By performing IMF decomposition on the signal, the system can extract the high-frequency secondary peaks and determine the arrival time of the reflected traveling wave by the occurrence time of these peaks. This process helps the system identify signal propagation delays and reflections, thus providing accurate data for subsequent energy ratio verification.
[0045] Then, an energy ratio check is performed based on the arrival times of the initial traveling wave and the reflected traveling wave to obtain the energy ratio check result. The energy ratio check verifies whether signal propagation is normal by calculating the energy ratio of the initial traveling wave and the reflected traveling wave. Specifically, the energy ratio is calculated using the following formula: ;in, Represents the energy ratio. The energy representing the initial wave, This represents the energy of the reflected traveling wave. By calculating the energy ratio, the system can determine whether the signal propagation is as expected. If the energy ratio is less than a preset threshold, it indicates that the signal propagation is normal, and the fault location model can continue to be activated.
[0046] If the energy ratio verification result meets the set conditions, i.e., the energy ratio is less than a certain threshold, the system will activate the traveling wave fault location model. This model calculates the fault location based on the arrival times of the initial traveling wave and the reflected traveling wave. By activating this model, the system can begin the actual fault location calculation.
[0047] Finally, based on the arrival times of the initial traveling wave and the reflected traveling wave, the first result of fault location is output through the traveling wave fault location model. The traveling wave fault location model determines the location of the fault point using the aforementioned arrival times and a corresponding calculation formula. Assuming the fault location model utilizes a simple location formula, it can be expressed as: ;in, This is the fault location result. It is the impedance at the fault point. This is the impedance of the line. Using this model, the distance from the fault point to the monitoring point can be calculated, which is the first result of fault location.
[0048] In summary, by employing precise signal processing and energy ratio verification, combined with a traveling wave fault location model, the location of a line fault can be accurately determined. The entire process relies on the arrival times of the initial traveling wave and the reflected traveling wave, and the reliability of the fault location results is ensured through energy ratio verification.
[0049] Furthermore, impedance-guided fault location is performed synchronously based on the line state change chain to obtain a second fault location result. In this embodiment, step P30 further includes: P36: Based on the line state change chain, construct the pre-fault reference phasor and the transient phasor at the fault moment; P37: Based on the pre-fault reference phasor and the transient phasor at the fault moment, construct the fault impedance estimation model; P38: Based on the line unit length impedance of the target line, perform fault location analysis according to the fault impedance estimation model to generate the second fault location result.
[0050] Specifically, in addition to traveling wave-guided fault location, impedance-guided fault location also needs to be performed simultaneously based on the line state change chain.
[0051] First, two key phasors are constructed based on the line state change chain: the pre-fault reference phasor and the transient phasor at the fault moment. The pre-fault reference phasor represents the voltage and current state of the line during normal operation, while the transient phasor at the fault moment represents the voltage and current characteristics of the line when the fault occurs. Through these two phasors, the system can analyze the changes in line state before and after the fault, providing basic data for subsequent fault impedance estimation.
[0052] Next, a fault impedance estimation model is constructed based on the pre-fault reference phasor and the transient phasor at the fault moment. This model reflects the change in line impedance at the time of the fault by comparing the two phasors. In this way, the system can estimate the impedance at the fault point, thus providing a reliable basis for subsequent fault location. When constructing the model, the fault impedance can be expressed by the following formula: ;in, Indicates fault impedance. and These represent the voltage and current at the moment of the fault, respectively. and This represents the voltage and current before the fault. Using this formula, the system can estimate the impedance at the fault point and provide data support for subsequent distance measurement.
[0053] Finally, based on the impedance per unit length of the target line, fault location analysis is performed according to the fault impedance estimation model to generate a second fault location result. The impedance per unit length of the target line is the resistance and reactance value per unit length of the line, representing the electrical characteristics of the line. Combined with the fault impedance estimation model, the system can calculate the distance between the fault point and the distribution terminal. For example, the fault location analysis uses the following formula: ;in, Indicates the distance between the fault point and the power distribution terminal. It is the impedance at the fault point. This is the impedance per unit length of the line. The physical meaning of this formula is that the distance from the fault point to the monitoring point is directly proportional to the fault impedance and inversely proportional to the impedance per unit length of the line. By using this formula, the embodiments of this application can accurately perform impedance-guided fault location based on the line state change chain and output a second result of the fault location.
[0054] P40: Perform correlation signal reliability quantification on the first fault ranging result and the second fault ranging result to determine the first ranging reliability and the second ranging reliability.
[0055] Furthermore, step P40 in this embodiment of the application also includes: P41: Based on the first fault location result and the second fault location result, identify the associated signals of the line state change chain to obtain a first associated signal set and a second associated signal set; P42: Activate the first signal evaluation mechanism and the second signal evaluation mechanism. The first signal evaluation mechanism includes a first signal evaluation factor and a first evaluation weight configuration. The first signal evaluation factor includes high-frequency mutation clarity, traveling wave energy concentration, and multi-scale consistency. The second signal evaluation mechanism includes a second signal evaluation factor and a second evaluation weight configuration. The second signal evaluation factor includes phasor consistency, harmonic suppression effectiveness, and impedance curve fitting degree; P43: Perform multi-dimensional reliability evaluation fusion on the first associated signal set according to the first signal evaluation mechanism to output the first ranging reliability; P44: Perform multi-dimensional reliability evaluation fusion on the second associated signal set according to the second signal evaluation mechanism to output the second ranging reliability.
[0056] Optionally, to ensure the accuracy and reliability of fault location results, the reliability of the associated signals between the two fault location results can be quantified. Specifically, by identifying the associated signals of the first and second fault location results, the reliability of each result is evaluated, and the results are fused through a multi-dimensional signal evaluation mechanism to finally output the reliability of the fault location.
[0057] First, based on the first and second results of fault location, the associated signals of the line state change chain are identified. The purpose of this process is to extract key signals related to the fault location results from the line state change chain. By identifying these signals, the system obtains two sets of associated signals: a first set and a second set, corresponding to the results of traveling wave-guided fault location and impedance-guided fault location, respectively. Through associated signal identification, the system can more accurately understand the signal characteristics affecting the fault location results.
[0058] Next, the first and second signal evaluation mechanisms are activated to evaluate the two associated signal sets. The first signal evaluation mechanism focuses on assessing signal characteristics related to the traveling wave, including high-frequency abrupt change clarity, traveling wave energy concentration, and multi-scale consistency. High-frequency abrupt change clarity measures the salience of the wavefronts of the initial traveling wave and the reflected wave, helping to identify the starting and reflection points of the traveling wave. Traveling wave energy concentration assesses the concentration effect of traveling wave energy within a specific time window, reflecting the intensity and concentration effect of the traveling wave, and helping to determine the salience of the fault signal. Multi-scale consistency verifies the synchronicity of traveling wave abrupt changes at different wavelet scales, ensuring the consistency of traveling wave characteristics across different scales.
[0059] The second signal evaluation mechanism focuses on assessing impedance-related signal characteristics, including phasor consistency, harmonic suppression effectiveness, and impedance curve fit. Phasor consistency measures the stability of phasor changes near the fault point. This factor ensures that the electrical characteristics of the line remain consistent before and after the fault, thereby improving the accuracy of distance measurement. Harmonic suppression effectiveness assesses whether the impedance signal is affected by noise interference, especially harmonic noise in the power system. This factor helps ensure the reliability of distance measurement results in noisy environments. Impedance curve fit determines the stability and reliability of impedance distance measurement results. If the impedance curve accurately fits the impedance changes of the actual line, the distance measurement results will be more reliable.
[0060] After the evaluation mechanism is activated, a multi-dimensional reliability evaluation fusion is performed on the first associated signal set. This step involves combining three evaluation factors—high-frequency mutation clarity, traveling wave energy concentration, and multi-scale consistency—using a weighted average or other fusion algorithm to obtain a comprehensive reliability score. This score reflects the overall reliability of the first associated signal set and provides a quantitative indicator to assess the credibility of the first fault location result. For example, if the high-frequency mutation clarity, traveling wave energy concentration, and multi-scale consistency are assigned weights of 0.4, 0.3, and 0.3 respectively, and the standardized values of these three factors are 0.8, 0.7, and 0.9 respectively, then the calculation process for the first fault location reliability is as follows: First fault location reliability = (0.4 × 0.8) + (0.3 × 0.7) + (0.3 × 0.9) = 0.80. This result shows that, according to the first signal evaluation mechanism, the reliability of the first fault location result is 0.80, or 80%. This is a relatively high reliability score, indicating that the first fault location result is credible.
[0061] Similarly, a multi-dimensional reliability evaluation and fusion is performed on the second associated signal set according to the second signal evaluation mechanism. By integrating factors such as phasor consistency, harmonic suppression effectiveness, and impedance curve fitting degree, the evaluation results from multiple dimensions are fused to output the second ranging reliability. This reliability value represents the reliability of the impedance-guided fault ranging result, ensuring that the fault location result obtained by this ranging method remains accurate under different signal interferences.
[0062] P50: Based on the first ranging reliability and the second ranging reliability, the ranging result reliability is reconstructed according to the predetermined reliability of the first fault ranging result and the second fault ranging result to obtain the fault ranging reliability result.
[0063] Furthermore, step P50 in this embodiment of the application also includes: P51: If the first ranging reliability is less than the predetermined reliability, and the second ranging reliability is less than the predetermined reliability, output a ranging anomaly warning signal; P52: If the first ranging reliability is greater than or equal to the predetermined reliability, and the second ranging reliability is less than the predetermined reliability, output the first fault ranging result as the reliable fault ranging result; P53: If the first ranging reliability is greater than or equal to the predetermined reliability, and the second ranging reliability is greater than or equal to the predetermined reliability, reconstruct and fuse the first fault ranging result and the second fault ranging result according to the first ranging reliability and the second ranging reliability to generate the reliable fault ranging result; P54: If the first ranging reliability is less than the predetermined reliability, and the second ranging reliability is greater than or equal to the predetermined reliability, output the second fault ranging result as the reliable fault ranging result.
[0064] Specifically, the reliability of fault ranging results is reconstructed based on the first ranging reliability and the second ranging reliability to obtain the final reliable fault ranging result.
[0065] First, the system determines whether the reliability of both the first and second fault ranging results is below a predetermined reliability standard. If the reliability of both ranging results is below the predetermined threshold, it means that the results of both ranging methods are unreliable, and the system cannot guarantee the accuracy of fault location. In this case, the system will output a ranging anomaly warning signal, indicating that there may be a problem with the fault ranging and further inspection or adjustment is needed. In this way, the system can issue timely warnings when the ranging results are unstable, avoiding incorrect fault location.
[0066] If the reliability of the first ranging result is greater than or equal to the predetermined reliability standard, while the reliability of the second ranging result is lower, then the first ranging result is selected as the final reliable fault ranging result. In this case, the system considers the fault ranging result guided by the traveling wave to be more reliable, and therefore will output this result as the final ranging result.
[0067] If the reliability of both ranging results meets a predetermined criterion—that is, the reliability of both the first and second ranging results is greater than or equal to a predetermined reliability—the system will reconstruct and fuse the ranging results. This process involves weighting the two ranging results and combining their reliability to output a comprehensive fault ranging reliability result. Through weighted fusion, the system can combine the advantages of the two ranging methods, improving the accuracy of the final ranging result. For example, the fusion process can be weighted using the following formula: ;in, This is the final reliable result of fault location. and These are the fault location results for traveling wave guided fault location and impedance guided fault location, respectively. and These are the reliability weights of the corresponding ranging results. In this way, the system can weight the two ranging results according to their reliability, ensuring the accuracy of the final ranging result.
[0068] If the reliability of the first fault location result is lower than the predetermined standard, while the reliability of the second fault location result meets the predetermined standard, the system will output the second fault location result as the final reliable fault location result. This means that when the impedance-guided fault location result is more reliable, the system will prioritize that result. In this way, the system can flexibly output the best fault location result based on the reliability of different fault location methods, improving the location accuracy of the power distribution system under complex fault conditions and the overall stability of the system.
[0069] In summary, the embodiments of this application have at least the following technical effects: This application effectively eliminates noise interference and enhances signal quality through noise suppression and signal enhancement techniques, thereby improving the accuracy of fault location. By combining traveling wave guided and impedance guided location methods and integrating multiple location results, it ensures stable and reliable fault location results in complex power grid environments. Adaptive triggering of state backtracking based on line fault probability constructs a line state change chain, improving the response speed and flexibility of fault location. By conducting reliability assessment and reconstruction of the location results, the fault location results are optimized, further enhancing the robustness of the system.
[0070] The technology achieves the effect of improving the accuracy and stability of fault location by suppressing noise and enhancing signals, as well as conducting reliability assessments and reconstructing results.
[0071] Example 2 is based on the same inventive concept as the data processing-based power distribution terminal fault detection method in the previous examples, such as... Figure 2 As shown, this application provides a power distribution terminal fault detection system based on data processing. The system and method embodiments in this application are based on the same inventive concept. The system includes: The real-time data sampling module 11 is used to obtain real-time sampling data of the target line from the interactive power distribution terminal, and to perform noise suppression and signal enhancement on the real-time sampling data to generate a real-time status sequence of the line.
[0072] The neighboring point state backtracking module 12 is used to calculate the line fault probability based on the line real-time state sequence, and adaptively trigger the benchmark neighboring point state backtracking of the line real-time state sequence based on the line fault probability to construct the line state change chain.
[0073] The fault location module 13 is used to perform traveling wave guided fault location based on the line state change chain to obtain a first fault location result, and simultaneously perform impedance guided fault location based on the line state change chain to obtain a second fault location result.
[0074] The signal reliability quantification module 14 is used to perform correlation signal reliability quantification on the first fault ranging result and the second fault ranging result to determine the first ranging reliability and the second ranging reliability.
[0075] The ranging result reconstruction module 15 is used to reconstruct the ranging result reliability of the first fault ranging result and the second fault ranging result based on the first ranging reliability and the second ranging reliability, according to a predetermined reliability, so as to obtain the fault ranging reliable result.
[0076] Furthermore, the real-time data sampling module 11 is also used to perform the following steps: The real-time sampled data is analyzed for signal features to obtain a sampled signal feature set; based on the sampled signal feature set, the automatic gain controller is driven to perform noise suppression and signal enhancement on the real-time sampled data to obtain the real-time state sequence of the line.
[0077] Furthermore, the neighbor state backtracking module 12 is also used to perform the following steps: Based on the power distribution terminal, the line fault log database of the target line is retrieved, and the line fault log database is randomly perturbed to obtain a line fault perturbation database; based on the line fault perturbation database, a Bayesian network is subjected to iterative supervised learning to obtain a first fault probability prediction model; based on the line fault perturbation database, a deep belief network is subjected to iterative supervised learning to obtain a second fault probability prediction model; based on the first fault probability prediction model and the second fault probability prediction model, distillation and fusion are performed to obtain a line fault prediction node; the real-time state sequence of the line is input into the line fault prediction node, and the line fault probability is output.
[0078] Furthermore, the neighbor state backtracking module 12 is also used to perform the following steps: If the line fault probability is greater than or equal to a fault probability threshold, construct adjacent sampling time sequences corresponding to the real-time sampling time points of the line's real-time state sequence; sort the adjacent sampling time sequences from nearest to farthest to obtain the first adjacent sampling time point; perform state sequence backtracking on the target line based on the first adjacent sampling time point to obtain the first neighboring point line state sequence; perform fault probability analysis on the target line based on the first neighboring point line state sequence to obtain the first neighboring point fault probability; determine whether the first neighboring point fault probability is less than the fault probability threshold; if the first neighboring point fault probability is less than the fault probability threshold, use the first neighboring point line state sequence as the benchmark neighboring point state backtracking result, and generate the line state change chain based on the benchmark neighboring point state backtracking result and the line's real-time state sequence.
[0079] Furthermore, the neighbor state backtracking module 12 is also used to perform the following steps: If the failure probability of the first neighboring point is greater than or equal to the failure probability threshold, the second neighboring sampling time point is read according to the adjacent sampling time sequence; the target line is backtracked according to the second neighboring sampling time point to obtain the second neighboring point line state sequence; the failure probability of the second neighboring point is calculated according to the second neighboring point line state sequence; if the failure probability of the second neighboring point is less than the failure probability threshold, the second neighboring point line state sequence is used as the reference neighboring point state backtracking result.
[0080] Furthermore, the fault location module 13 is also used to perform the following steps: High-frequency energy mutation detection is performed based on the line state change chain to determine the arrival time of the initial traveling wave; high-frequency secondary peak identification of the IMF is performed based on the line state change chain to determine the arrival time of the reflected traveling wave; energy ratio verification is performed based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave to obtain the energy ratio verification result; if the energy ratio verification result meets the energy ratio verification condition, the traveling wave fault location model is activated; based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave, the first fault location result is output according to the traveling wave fault location model.
[0081] Furthermore, the fault location module 13 is also used to perform the following steps: Based on the line state change chain, a reference phasor before the fault and a transient phasor at the fault moment are constructed; based on the reference phasor before the fault and the transient phasor at the fault moment, a fault impedance estimation model is constructed; based on the line impedance per unit length of the target line, fault location analysis is performed according to the fault impedance estimation model to generate the second fault location result.
[0082] Furthermore, the signal reliability quantification module 14 is also used to perform the following steps: Based on the first fault location result and the second fault location result, the associated signal identification of the line state change chain is performed to obtain a first associated signal set and a second associated signal set; a first signal evaluation mechanism and a second signal evaluation mechanism are activated. The first signal evaluation mechanism includes a first signal evaluation factor and a first evaluation weight configuration. The first signal evaluation factor includes high-frequency mutation clarity, traveling wave energy concentration, and multi-scale consistency. The second signal evaluation mechanism includes a second signal evaluation factor and a second evaluation weight configuration. The second signal evaluation factor includes phasor consistency, harmonic suppression effectiveness, and impedance curve fitting degree; a multi-dimensional reliability evaluation fusion is performed on the first associated signal set according to the first signal evaluation mechanism to output the first ranging reliability; a multi-dimensional reliability evaluation fusion is performed on the second associated signal set according to the second signal evaluation mechanism to output the second ranging reliability.
[0083] Furthermore, the ranging result reconstruction module 15 is also used to perform the following steps: If the first ranging reliability is less than the predetermined reliability and the second ranging reliability is less than the predetermined reliability, a ranging anomaly warning signal is output; if the first ranging reliability is greater than or equal to the predetermined reliability and the second ranging reliability is less than the predetermined reliability, the first fault ranging result is output as the reliable fault ranging result; if the first ranging reliability is greater than or equal to the predetermined reliability and the second ranging reliability is greater than or equal to the predetermined reliability, the first fault ranging result and the second fault ranging result are reconstructed and fused according to the first ranging reliability and the second ranging reliability to generate the reliable fault ranging result; if the first ranging reliability is less than the predetermined reliability and the second ranging reliability is greater than or equal to the predetermined reliability, the second fault ranging result is output as the reliable fault ranging result.
[0084] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.
[0085] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.
[0086] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.
Claims
1. A method for detecting faults in power distribution terminals based on data processing, characterized in that, include: The interactive power distribution terminal acquires real-time sampling data of the target line, and performs noise suppression and signal enhancement on the real-time sampling data to generate a real-time status sequence of the line. The line fault probability is calculated based on the real-time state sequence of the line, and the baseline neighbor state backtracking of the real-time state sequence of the line is adaptively triggered based on the line fault probability to construct the line state change chain. Based on the line state change chain, traveling wave guided fault location is performed to obtain the first fault location result. Simultaneously, impedance guided fault location is performed based on the line state change chain to obtain the second fault location result. The first fault location result and the second fault location result are correlated with a reliable signal quantification to determine the first ranging reliability and the second ranging reliability. Based on the first ranging reliability and the second ranging reliability, the ranging result reliability is reconstructed by reconstructing the first fault ranging result and the second fault ranging result according to a predetermined reliability, so as to obtain a reliable fault ranging result.
2. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, The probability of line faults is calculated based on the real-time status sequence of the line, including: According to the power distribution terminal, the line fault log database of the target line is retrieved, and the line fault log database is randomly disturbed to obtain the line fault disturbance database. Based on the line fault disturbance library, the Bayesian network is iteratively supervised to learn and obtain the first model for fault probability prediction. Based on the line fault disturbance library, the deep belief network is subjected to iterative supervised learning to obtain a second model for fault probability prediction. Distillation and fusion are performed on the first fault probability prediction model and the second fault probability prediction model to obtain line fault prediction nodes; The real-time status sequence of the line is input into the line fault prediction node, and the line fault probability is output.
3. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, Based on the line fault probability, adaptive triggering of the baseline neighbor state backtracking of the real-time state sequence of the line is performed to construct the line state change chain, including: If the line fault probability is greater than or equal to the fault probability threshold, construct the adjacent sampling time sequence corresponding to the real-time sampling time point of the line real-time status sequence; The adjacent sampling time series are sorted from nearest to farthest to obtain the first adjacent sampling time point; Based on the first adjacent sampling time point, the target line is backtracked to obtain the state sequence of the first neighboring line. Based on the first neighboring point line state sequence, the target line is analyzed for fault probability to obtain the first neighboring point fault probability. Determine whether the failure probability of the first neighboring point is less than the failure probability threshold; If the failure probability of the first neighboring point is less than the failure probability threshold, the line state sequence of the first neighboring point is used as the benchmark neighboring point state backtracking result, and the line state change chain is generated based on the benchmark neighboring point state backtracking result and the real-time line state sequence.
4. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, Based on the aforementioned line state change chain, traveling wave-guided fault location is implemented to obtain the first fault location result, including: High-frequency energy mutation detection is performed based on the line state change chain to determine the arrival time of the initial traveling wave; IMF high-frequency peak identification is performed based on the line state change chain to determine the arrival time of the reflected traveling wave. The energy ratio is verified based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave to obtain the energy ratio verification result. If the energy ratio verification result meets the energy ratio verification condition, the traveling wave fault location model is activated. Based on the arrival time of the initial traveling wave and the arrival time of the reflected traveling wave, the first fault location result is output according to the traveling wave fault location model.
5. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, Synchronously perform impedance-guided fault location based on the line state change chain to obtain a second fault location result, including: Based on the line state change chain, construct the pre-fault reference phasor and the transient phasor at the fault moment; Based on the pre-fault reference phasor and the transient phasor at the fault moment, a fault impedance estimation model is constructed; Based on the line unit length impedance of the target line, fault location analysis is performed according to the fault impedance estimation model to generate the second fault location result.
6. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, The first fault location result and the second fault location result are correlated with a reliable signal quantification to determine the reliability of the first and second ranging methods, including: Based on the first fault location result and the second fault location result, the associated signal identification is performed on the line state change chain to obtain a first associated signal set and a second associated signal set; Activate the first signal evaluation mechanism and the second signal evaluation mechanism. The first signal evaluation mechanism includes a first signal evaluation factor and a first evaluation weight configuration. The first signal evaluation factor includes high-frequency mutation clarity, traveling wave energy concentration and multi-scale consistency. The second signal evaluation mechanism includes a second signal evaluation factor and a second evaluation weight configuration. The second signal evaluation factor includes phasor consistency, harmonic suppression effectiveness and impedance curve fitting degree. The first associated signal set is evaluated and fused using the first signal evaluation mechanism to output the first ranging reliability. The second associated signal set is evaluated and fused using the second signal evaluation mechanism to output the second ranging reliability.
7. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, Based on the first ranging reliability and the second ranging reliability, the ranging result reliability is reconstructed according to a predetermined reliability for the first fault ranging result and the second fault ranging result to obtain a reliable fault ranging result, including: If the first ranging reliability is less than the predetermined reliability, and the second ranging reliability is less than the predetermined reliability, a ranging anomaly warning signal is output; If the first ranging reliability is greater than or equal to the predetermined reliability, and the second ranging reliability is less than the predetermined reliability, the first fault ranging result is output as the reliable fault ranging result. If the first ranging reliability is greater than or equal to the predetermined reliability, and the second ranging reliability is greater than or equal to the predetermined reliability, the first fault ranging result and the second fault ranging result are reconstructed and fused according to the first ranging reliability and the second ranging reliability to generate the fault ranging reliable result; If the first ranging reliability is less than the predetermined reliability, and the second ranging reliability is greater than or equal to the predetermined reliability, the second fault ranging result is output as the fault ranging reliability result.
8. The power distribution terminal fault detection method based on data processing as described in claim 1, characterized in that, The real-time sampled data is subjected to noise suppression and signal enhancement to generate a real-time line status sequence, including: The real-time sampled data is analyzed for signal features to obtain a sampled signal feature set; Based on the sampled signal feature set, the automatic gain controller is driven to perform noise suppression and signal enhancement on the real-time sampled data to obtain the real-time state sequence of the line.
9. The power distribution terminal fault detection method based on data processing as described in claim 3, characterized in that, Determining whether the failure probability of the first neighboring point is less than the failure probability threshold includes: If the failure probability of the first neighboring point is greater than or equal to the failure probability threshold, the second adjacent sampling time point is read according to the adjacent sampling time sequence. Based on the second adjacent sampling time point, the target line is backtracked to obtain the second adjacent point line state sequence; Calculate the failure probability of the second neighboring point based on the line state sequence of the second neighboring point; If the failure probability of the second neighboring point is less than the failure probability threshold, the line state sequence of the second neighboring point is used as the baseline neighboring point state backtracking result.
10. A power distribution terminal fault detection system based on data processing, characterized in that, The system is used to implement the data processing-based power distribution terminal fault detection method according to any one of claims 1 to 9, and the system includes: The real-time data sampling module is used to acquire real-time sampling data of the target line through the interactive power distribution terminal, and to perform noise suppression and signal enhancement on the real-time sampling data to generate a real-time status sequence of the line. The neighboring point state backtracking module is used to calculate the line fault probability based on the line real-time state sequence, and adaptively trigger the benchmark neighboring point state backtracking of the line real-time state sequence based on the line fault probability to construct the line state change chain. The fault location module is used to perform traveling wave guided fault location based on the line state change chain to obtain a first fault location result, and simultaneously perform impedance guided fault location based on the line state change chain to obtain a second fault location result. The signal reliability quantification module is used to perform correlation signal reliability quantification on the first fault ranging result and the second fault ranging result to determine the first ranging reliability and the second ranging reliability. The ranging result reconstruction module is used to reconstruct the ranging result reliability of the first fault ranging result and the second fault ranging result based on the first ranging reliability and the second ranging reliability, according to a predetermined reliability, so as to obtain the fault ranging reliable result.