A circuit maintenance fault performance detection method and system

Abstract

The application provides a circuit maintenance fault performance detection method and system, relates to the technical field of circuit maintenance, and captures subtle changes in operation by acquiring power grid multi-position operation signals and time alignment, combining signal propagation characteristics of physical structures and dynamic correction, and jointly optimizing defect sources and medium distortion parameters. From the complex superimposed signal, separate the characteristics and positions of multiple independent signal sources, solve the problem that weak signals are easily covered and multiple defects coexist, and analyze the correlation of the separated signal sources. Based on the pure signal after environmental stress compensation, the severity and development trend are evaluated, environmental interference is overcome, and the bottleneck of traditional implicit defect identification is broken. The method can realize early and accurate identification and positioning of implicit defects and trend prediction, provide reliable support for intelligent operation and maintenance of power grids, and significantly improve power supply quality and system reliability.

Description

Technical Field

[0001] This application relates to the technical field of circuit maintenance, and more particularly, to a method and system for detecting circuit maintenance fault performance. Background Art

[0002] In the power distribution link of large urban power grids, traditional circuit maintenance methods are difficult to effectively identify "hidden defects" - these defects do not directly cause power outages, but will continuously degrade the power supply quality and may develop into serious faults. Their concealment is strong, and conventional outage detections (such as withstand voltage tests and partial discharge detections) often fail due to the "memory effect" of the defects (showing state changes with the accumulation of operation history and stress). For example, a small air gap in a cable joint causes partial discharge when heated under high load, but the air gap returns to normal after power outage and cooling, and conventional detections cannot capture the abnormality.

[0003] The weak signals of hidden defects are easily interfered by the complex environment of the power grid (such as transient disturbances generated by switch operations and load fluctuations), and their waveform, spectrum, and energy characteristics may be highly similar to non-fault events, resulting in misjudgment or missed judgment by the monitoring system.

[0004] To address this, the operation and maintenance team introduced online monitoring technology, and real-time collected electrical parameters through sensors at key nodes, but faced new problems: one is that it is difficult to efficiently screen out the precursors of faults from a large amount of data; the other is that when multiple defects coexist, complex interactions occur (such as local overheating accelerating insulation aging, and signal superposition and interference of multiple discharge points), and single-signal feature analysis cannot accurately identify the independent states, positions, and associated effects of each defect, and it is even more difficult to evaluate its threat to the overall performance of the system. Existing technologies have not solved the core problems such as the accurate identification of hidden defects, the interactive analysis of multiple defects, and the value mining of a large amount of data, which restricts the improvement of power supply reliability. Summary of the Invention

[0005] This application provides a method and system for detecting circuit maintenance fault performance, aiming to solve the technical problems in the power distribution link of large urban power grids, such as the difficulty of traditional maintenance methods in identifying hidden defects, the easy masking or imitation of weak signals, and the difficulty of signal analysis when multiple defects coexist.

[0006] In a first aspect, this application discloses a method for detecting circuit maintenance fault performance, including:

[0007] Obtaining the operation signals at multiple positions in the power grid and performing time alignment on the operation signals;

[0008] Obtaining the physical structure information of the power grid and dynamically correcting based on real-time operation parameters to determine the signal propagation characteristics of the signals in the physical structure of the power grid;

[0009] Based on the time-aligned running signal, the physical structure information, and the signal propagation characteristics, the features and locations of multiple independent signal sources are separated from the superimposed signal by jointly optimizing the defect source parameters and the medium distortion parameters.

[0010] Correlation analysis is performed on the isolated independent signal sources, and the severity and development trend of the independent signal sources are evaluated, wherein the correlation analysis is based on the pure defect signals after environmental stress compensation.

[0011] Secondly, this application discloses a circuit troubleshooting performance testing system, which includes:

[0012] The signal acquisition module is used to acquire operating signals from multiple locations in the power grid and to time-align the operating signals.

[0013] The structural information acquisition module is used to acquire the physical structural information of the power grid and determine the signal propagation characteristics of the signal in the physical structure of the power grid based on real-time operating parameters and dynamic correction.

[0014] The signal separation module is used to separate the features and locations of multiple independent signal sources from the superimposed signals based on the time-aligned running signals, the physical structure information, and the signal propagation characteristics, by jointly optimizing the defect source parameters and the medium distortion parameters.

[0015] The correlation analysis module is used to perform correlation analysis on the isolated independent signal sources and to evaluate the severity and development trend of the independent signal sources. The correlation analysis is based on the pure defect signal after environmental stress compensation.

[0016] Beneficial effects

[0017] This application discloses a circuit maintenance fault performance detection method. By acquiring and time-aligning operating signals from multiple locations within the power grid, and combining this with the power grid's physical structure information and dynamically corrected signal propagation characteristics, it can accurately capture subtle changes in power grid operation. Through joint optimization of defect source parameters and dielectric distortion parameters, this method can effectively separate the characteristics and locations of multiple independent signal sources from complex superimposed signals. This solves the problems in existing technologies where weak signals are easily masked or imitated by other normal operating events, and where signals are difficult to analyze when multiple defects coexist. Furthermore, this method performs correlation analysis on the separated independent signal sources and assesses their severity and development trend based on the pure defect signal after environmental stress compensation. This effectively overcomes the interference of environmental stress on fault judgment and avoids the challenges of traditional methods in identifying latent defects. Therefore, this application can achieve early and accurate identification and location of latent defects in the power grid, and scientifically predict their development trends, providing reliable technical support for intelligent operation and maintenance of the power grid, and significantly improving power supply quality and system reliability. Attached Figure Description

[0018] To illustrate this application more clearly, the accompanying drawings used in the embodiments will be briefly described below. Obviously, those skilled in the art can obtain other drawings based on these drawings without any creative effort.

[0019] Figure 1 The diagram above illustrates a flowchart of a circuit troubleshooting and fault performance testing method.

[0020] Figure 2 The diagram above illustrates a schematic of a circuit troubleshooting and fault detection system.

[0021] Figure reference numerals: 100, Circuit troubleshooting and fault performance testing system; 10, Signal acquisition module; 20, Structural information acquisition module; 30, Signal separation module; 40, Correlation analysis module. Detailed Implementation

[0022] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0023] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] Traditional circuit maintenance methods face numerous challenges in identifying "latent defects" that do not immediately cause power outages but continuously affect power quality and may gradually deteriorate into serious faults. These latent defects are often concealed, their symptoms are difficult to detect during routine power outages, and their performance subtly changes with operating history and stress accumulation, exhibiting a "memory effect" or "cumulative effect." Furthermore, the weak signals generated by these latent defects are easily masked or imitated by other normal operating events in the complex power grid operating environment, leading to misjudgments or omissions. When multiple latent defects coexist in the same line segment, they may generate complex "interactions," making the signals received by monitoring devices more complex and difficult to interpret. In this case, single signal feature analysis methods often cannot accurately identify the independent existence and specific location of each defect, let alone assess their correlation and impact on the overall system performance.

[0025] In this regard, such as Figure 1 The diagram illustrates an exemplary flowchart of a circuit troubleshooting fault performance testing method. This application proposes a circuit troubleshooting fault performance testing method, comprising:

[0026] S10, acquire operating signals from multiple locations in the power grid, and perform time alignment on the operating signals;

[0027] S20, Obtain the physical structure information of the power grid, and determine the signal propagation characteristics of the signal in the physical structure of the power grid based on real-time operating parameters and dynamic correction.

[0028] S30, based on the time-aligned running signal, the physical structure information, and the signal propagation characteristics, the features and locations of multiple independent signal sources are separated from the superimposed signal by jointly optimizing the defect source parameters and the medium distortion parameters;

[0029] S40, Perform correlation analysis on the isolated independent signal sources and assess the severity and development trend of the independent signal sources, wherein the correlation analysis is based on the pure defect signals after environmental stress compensation.

[0030] This application, by introducing signal propagation characteristics, jointly optimizing defect source parameters and medium distortion parameters, and performing correlation analysis based on pure defect signals after environmental stress compensation, can effectively overcome the limitations of traditional methods in identifying latent defects, distinguishing between true and false faults, and handling the interaction of multiple defects, thereby significantly improving the accuracy and reliability of power grid fault detection.

[0031] To better understand the circuit troubleshooting and fault detection method proposed in this application, the following will provide a detailed explanation of some key terms and implementation environments involved.

[0032] Operational signals refer to electrical parameters collected by various sensors and monitoring equipment during power grid operation, such as voltage, current, partial discharge signals, temperature, and vibration. These signals contain a wealth of information about the power grid's operating status and potential faults.

[0033] The physical structure information of a power grid refers to detailed data at the physical level, such as the grid's topology, line lengths, cable types, transformer parameters, and switchgear locations. This information is fundamental to understanding the propagation paths and characteristics of signals within the power grid.

[0034] Real-time operating parameters refer to the dynamic operating data of the power grid at the current moment, such as load, voltage, current, and ambient temperature. These parameters affect the propagation characteristics of signals in the power grid, and therefore require dynamic correction.

[0035] Signal propagation characteristics refer to the physical properties of signals, such as attenuation, delay, reflection, and refraction, exhibited when signals propagate through the physical structure of a power grid. These characteristics are affected by the physical structure of the power grid and real-time operating parameters.

[0036] Defect source parameters refer to characteristic parameters that describe potential defect sources in the power grid, such as the location, type, intensity, and duration of the defect.

[0037] Dielectric distortion parameters refer to parameters that describe changes in the electrical characteristics of the dielectric material (such as insulating material) in a power grid due to aging, damage, etc., such as dielectric constant and loss tangent. Dielectric distortion can affect signal propagation.

[0038] An independent signal source refers to a signal generated by a single defect or fault point that can be distinguished from other signal sources.

[0039] A pure defect signal refers to a signal that, after environmental stress compensation, has had its environmental stress effects removed and can truly reflect the characteristics of the defect itself.

[0040] The implementation environment of this application is typically the power distribution segment of a large urban power grid, which is equipped with numerous sensors and monitoring devices capable of continuously collecting power grid operation data. The entire detection process can be carried out without power interruption, enabling online monitoring and diagnosis of latent defects.

[0041] The core of the circuit fault performance detection method of this application lies in achieving accurate identification, location and evaluation of hidden defects in the power grid through a series of refined signal processing and analysis steps.

[0042] First, it is necessary to acquire operational signals from multiple locations within the power grid and time-align these signals. Acquiring these signals can be achieved by deploying various sensors at key nodes in the power grid. For example, high-frequency current sensors can be installed to monitor transient current changes in lines, or partial discharge sensors can be installed to capture weak discharge signals generated by insulation defects. These sensors convert the acquired analog signals into digital signals and transmit them to the data processing center via a communication network. To ensure the accuracy of subsequent analysis, time alignment of these operational signals from different locations is crucial. Time alignment can be achieved in several ways. For instance, a high-precision GPS timing module can be used to provide a unified time reference for each sensor, ensuring that all acquired data has accurate timestamps. At the data processing center, the data can be resampled or interpolated based on these timestamps, ensuring that all signals are fully synchronized on the timeline.

[0043] Secondly, it is necessary to obtain information about the physical structure of the power grid and dynamically adjust it based on real-time operating parameters to determine the signal propagation characteristics within the power grid's physical structure. This physical structure information can be obtained from the power grid's Geographic Information System (GIS) or design drawings, including line lengths, cable types, transformer locations, and switchgear connections. This information constitutes the physical path for signal propagation. However, the signal propagation characteristics within the power grid are not static; they are affected by real-time operating parameters. For example, changes in line load can cause slight changes in impedance, and rising ambient temperature can affect the dielectric constant of cables, all of which impact signal attenuation and propagation speed. Therefore, it is necessary to obtain real-time power grid operating parameters, such as real-time voltage, current, and power data obtained through a SCADA system, and use these parameters to dynamically adjust the signal propagation characteristics. For instance, a propagation model based on electromagnetic field theory can be established, using real-time operating parameters as input, to dynamically calculate the signal attenuation coefficient and phase velocity in different line segments.

[0044] Furthermore, based on time-aligned operating signals, physical structure information, and signal propagation characteristics, the characteristics and locations of multiple independent signal sources are separated from the superimposed signals by jointly optimizing defect source parameters and dielectric distortion parameters. In complex power grid environments, multiple latent defects may coexist, and their generated signals will superimpose, making the originally acquired operating signals very complex. To accurately identify each defect, these superimposed signals need to be separated. This application employs a joint optimization method that considers both the parameters of the defect source itself (such as location and intensity) and dielectric distortion parameters (such as insulation aging degree). For example, a power grid signal propagation model containing multiple potential defect sources and dielectric distortion effects can be constructed. This model uses the parameters of the potential defect sources and dielectric distortion parameters as optimization variables. By iteratively adjusting these variables, the difference between the signal predicted by the model and the actual time-aligned operating signal acquired is minimized. This optimization process can employ various numerical optimization algorithms, such as genetic algorithms, particle swarm optimization, or gradient descent. When the model converges, the optimized defect source parameters and dielectric distortion parameters can reflect the characteristics and locations of independent signal sources.

[0045] Finally, correlation analysis is performed on the isolated independent signal sources to assess their severity and development trends. This correlation analysis is based on the purified defect signals after environmental stress compensation. After isolating the independent signal sources, further analysis of their interrelationships and individual hazard levels is necessary. However, direct analysis of the original isolated signals may be affected by environmental stress. For example, drastic changes in ambient temperature may cause fluctuations in the intensity of the defect signal, but this does not necessarily mean that the defect itself has worsened. Therefore, environmental stress compensation is required before conducting correlation analysis and assessment to obtain "pure defect signals." Environmental stress compensation can be achieved by establishing a mapping relationship between environmental stress and signal distortion. For example, experiments can be conducted beforehand to measure the signal variation patterns generated by known defect sources under different environmental conditions such as temperature and humidity, thereby constructing a compensation model. After obtaining the pure defect signals, various statistical or machine learning methods can be used for correlation analysis. For example, it is possible to analyze whether there is temporal synchronicity or causal relationship between different defect signals. Simultaneously, based on the intensity, duration, and other characteristics of the pure defect signals, combined with historical operating data of the power grid equipment, the severity of each independent signal source can be assessed, and its future development trend can be predicted.

[0046] The circuit fault performance detection method of this application lays the foundation for subsequent signal analysis by acquiring and time-aligning operating signals from multiple locations in the power grid. Simultaneously, it acquires the physical structure information of the power grid and dynamically corrects signal propagation characteristics based on real-time operating parameters, enabling the signal propagation model to more accurately reflect the actual operating state of the power grid. Furthermore, by jointly optimizing defect source parameters and dielectric distortion parameters, the characteristics and locations of multiple independent signal sources are separated from the superimposed signals, effectively solving the problem of mutual masking and interference between multiple defect signals. Finally, correlation analysis is performed on the separated independent signal sources, and their severity and development trend are assessed. The correlation analysis is based on pure defect signals after environmental stress compensation, which further improves the accuracy and reliability of fault diagnosis and avoids interference from environmental factors in defect assessment.

[0047] Compared with traditional circuit repair methods, the circuit repair fault performance detection method of this application has significant advantages. While traditional online monitoring technologies can collect massive amounts of data, they have limitations in handling the superposition of multiple defect signals and environmental stress interference, easily leading to misjudgments or missed diagnoses. This application, by introducing signal propagation characteristics and jointly optimizing defect source parameters and dielectric distortion parameters, can more accurately separate independent defect signals from complex superimposed signals, thus overcoming the challenges of traditional methods in identifying latent defects. Furthermore, this application specifically emphasizes correlation analysis and evaluation based on the pure defect signal after environmental stress compensation. This makes the judgment of defect severity and development trend more objective and accurate, avoiding interference from environmental factors in the diagnostic results. For example, in traditional methods, a partial discharge signal generated by a cable joint in a high-temperature and high-humidity environment may be misjudged as defect deterioration. However, this application, through environmental stress compensation, can distinguish whether the signal enhancement is caused by environmental factors or whether the defect itself is indeed deteriorating. This sophisticated processing capability enables this application to more effectively identify "hidden defects" that do not immediately cause power outages but continuously affect power quality, and to provide early warnings and maintenance for them, thereby significantly improving the reliability and security of power grid operation.

[0048] In some embodiments described above, this application proposes a method based on time-aligned operating signals, power grid physical structure information, and signal propagation characteristics. By jointly optimizing defect source parameters and dielectric distortion parameters, the characteristics and locations of multiple independent signal sources are separated from superimposed signals. However, in actual power grid operation, in addition to signals caused by defects, there may be a large number of normal transient events, such as switching operations and load fluctuations. The signals generated by these events are superimposed on defect signals, which may make it difficult to effectively distinguish and accurately extract pure defect signal features when performing signal separation directly in the time domain, thus affecting the accuracy of fault source location. To address this, this application further proposes a more refined signal separation method, aiming to improve the accuracy and robustness of independent signal source separation through frequency domain processing and interference filtering.

[0049] The steps described above, based on time-aligned operating signals, the physical structure information, and the signal propagation characteristics, to separate the features and locations of multiple independent signal sources from superimposed signals by jointly optimizing defect source parameters and medium distortion parameters, include:

[0050] Convert time-aligned operating signals into a frequency domain representation and filter out frequency components of normal transient events in the power grid;

[0051] Based on the physical structure of the power grid and the dynamically corrected signal propagation characteristics, a frequency domain propagation model containing potential signal sources is constructed.

[0052] By jointly optimizing the defect source parameters and the medium distortion parameters, the characteristics and location of independent signal sources can be separated from the filtered frequency domain signal.

[0053] Specifically, converting time-aligned operating signals into a frequency domain representation is typically achieved using mathematical tools such as Fourier transforms. The aim is to decompose complex time-domain signals into different frequency components for easier identification and processing. This involves identifying and removing signals within specific frequency ranges that are related to normal grid operation (such as switching operations and load changes) and are not caused by defects. For example, this can be achieved by using band-stop filters or spectral analysis techniques to identify and suppress the frequency components of known interference. The goal is to eliminate background noise and interference from non-fault-related signals, thereby highlighting the characteristics of defective signals.

[0054] Furthermore, based on information about the physical structure of the power grid, such as line length, impedance, and topology, and the signal propagation characteristics obtained through dynamic correction of real-time operating parameters, a frequency domain propagation model is constructed. This model describes how a potential signal source generates a signal in the power grid, propagates through the power grid structure, and reaches the frequency domain response at various monitoring devices. Specifically, this model can be a matrix equation or a set of differential equations, aiming to simulate the attenuation, reflection, and superposition processes of signals in the power grid, providing a theoretical basis for subsequent signal source localization.

[0055] This process involves simultaneously adjusting parameters representing the characteristics of the defect source (such as location, intensity, and spectral features) and parameters representing anomalies in the medium along the signal propagation path (such as local dielectric constant variations and loss factors) to ensure that the signal predicted by the model best matches the actual filtered frequency domain signal. This process typically employs iterative optimization algorithms, such as least squares, gradient descent, or heuristic algorithms, aiming to accurately deduce the characteristics and location of independent signal sources by minimizing the difference between the theoretical response and actual observations, while also considering the influence of the medium on signal propagation.

[0056] The proposed solution converts time-aligned operating signals into a frequency domain representation and filters out frequency components of normal power grid transient events during this stage, effectively focusing attention on signal characteristics related to defects. This frequency domain conversion and filtering process allows superimposed signals, which are difficult to distinguish in the time domain, to be analyzed and purified in the frequency dimension, thus significantly reducing the interference of normal operating noise on defect signal identification. Subsequently, based on a precisely constructed frequency domain propagation model containing potential signal sources, combined with the filtered clean frequency domain signal, the true characteristics and location of the signal source can be more accurately inverted by jointly optimizing the defect source parameters and the dielectric distortion parameters. The introduction of dielectric distortion parameters enables the model to more realistically reflect the propagation path and attenuation characteristics of signals in complex power grid environments, further improving the accuracy and robustness of signal separation.

[0057] Through the above technical solution, this application effectively overcomes the limitations of traditional time-domain signal separation methods, which are susceptible to interference from normal transient events when processing complex power grid operation signals. By performing signal processing and interference filtering in the frequency domain, the accuracy of separating the characteristics and location of independent signal sources from superimposed signals is significantly improved. Furthermore, constructing a frequency-domain propagation model and jointly optimizing defect source parameters and dielectric distortion parameters makes the location of fault sources more precise and the description of defect characteristics more detailed. This provides a more reliable basis for power grid inspection and maintenance, effectively avoiding resource waste caused by misjudging normal transient events as faults.

[0058] For example, suppose a partial discharge (PD) defect occurs in a high-voltage transmission line, which generates a high-frequency transient signal. First, operational signals are collected by sensors installed at different locations along the transmission line and time-aligned. Then, these time-aligned signals are converted to a frequency domain representation. During this process, a pre-established spectral feature library of transient events during normal power grid operation (such as switch closing and lightning strikes) is used to filter the collected frequency domain signals, removing frequency components related to these normal events, thus obtaining a cleaner defect signal spectrum. Next, combining detailed physical structure information of the transmission line (including line length, tower locations, and insulator types) with a signal propagation model dynamically corrected by real-time load and environmental parameters, a frequency domain propagation model containing potential PD sources is constructed. Finally, through an iterative optimization algorithm, defect source parameters such as the possible location and discharge intensity of the PD source, as well as local distortion parameters of the line medium, are jointly adjusted to minimize the difference between the model-predicted frequency domain response and the filtered actual measured frequency domain signal. When the optimization process converges, the characteristics of the PD defect source (such as discharge type and intensity) and its specific location on the transmission line can be accurately separated.

[0059] In some embodiments, this application further proposes the following steps for separating the features and locations of independent signal sources from the filtered frequency domain signal by jointly optimizing the defect source parameters and the dielectric distortion parameters:

[0060] The dielectric distortion parameters are added as optimization variables to the frequency domain propagation model to iteratively adjust the number, location, and frequency domain characteristics of potential signal sources.

[0061] Calculate the theoretical frequency domain response of the potential signal source at each monitoring device in the power grid;

[0062] The theoretical frequency domain response is matched with the filtered frequency domain signal, and the parameters of multiple potential signal sources are optimized by minimizing the difference between the two until convergence.

[0063] Based on the optimized parameters of the potential signal sources, the characteristics and locations of the independent signal sources are separated.

[0064] Specifically, when constructing the signal propagation model, in addition to considering the characteristics of the signal source itself (such as quantity, location, and frequency domain characteristics), the effects of attenuation, phase shift, and dispersion experienced by the signal during propagation in the power grid medium are quantified as medium distortion parameters and treated as variables that can be adjusted during the optimization process. The aim is to more comprehensively characterize the actual physical effects along the signal propagation path, enabling the model to adapt to different medium environments. Iteratively adjusting the quantity, location, and frequency domain characteristics of potential signal sources allows for a cyclic optimization process, gradually correcting the estimation of potential signal sources to make it closer to reality. Here, the quantity of potential signal sources refers to the number of possible defect sources in the power grid; the location refers to the specific coordinates of the defect source within the physical structure of the power grid; and the frequency domain characteristics refer to the energy distribution and phase information of the defect signal at different frequencies.

[0065] Furthermore, based on the estimated parameters of potential signal sources and dielectric distortion parameters in the current iteration, the constructed frequency domain propagation model is used to predict the signals that should be received at each monitoring device in the power grid. The theoretical frequency domain response is matched with the filtered frequency domain signal, and the parameters of multiple potential signal sources are optimized by minimizing the difference between the two until convergence. This involves comparing the theoretical response predicted by the model with the actual acquired and filtered frequency domain signal. An error function (e.g., mean square error) is defined, and optimization algorithms (e.g., gradient descent, genetic algorithm) are used to continuously adjust the parameters of the potential signal sources (including dielectric distortion parameters, the number, location, and frequency domain characteristics of potential signal sources) to minimize the difference between the theoretical response and the actual signal. This process continues until the parameter change is less than a preset threshold or the error function value no longer decreases significantly, indicating convergence. After the optimization process converges, the optimal parameter set represents the true characteristics (e.g., frequency components, intensity) of the independent signal sources and their specific locations in the power grid.

[0066] By incorporating the dielectric distortion parameters into the iterative optimization process, the signal separation model can more accurately reflect the actual signal propagation path and environmental influences. This significantly improves the accuracy and robustness of separating independent signal sources from superimposed signals, especially under conditions of complex or uncertain power grid dielectric characteristics. Consequently, it enables more reliable identification of defect source characteristics and locations, providing a cleaner and more accurate data foundation for subsequent correlation analysis and fault diagnosis, thereby enhancing the overall efficiency of circuit maintenance and fault detection.

[0067] As a specific implementation method, a concrete example is given below. Suppose a partial discharge (PD) fault occurs in an underground cable section of a power grid. The resulting signal propagates through the cable and subsequent overhead lines and is collected by multiple monitoring devices. Due to the differences in the dielectric properties of the cable and overhead lines, which may vary with the environment, signal separation based solely on fixed dielectric parameters may lead to errors.

[0068] The specific implementation of the scheme in this application is as follows:

[0069] First, the acquired operating signals are time-aligned and converted into a frequency domain representation. The frequency components of normal transient events in the power grid are filtered out to obtain the filtered frequency domain signal.

[0070] Next, a frequency domain propagation model is constructed that includes potential PD signal sources and dielectric distortion parameters. Initially, the location, intensity, and frequency domain characteristics of the PD source can be estimated, and the dielectric distortion parameters of the cable and overhead line (e.g., attenuation coefficient and phase delay at different frequencies) can be initially set.

[0071] Then, the iterative optimization process begins. In each iteration:

[0072] 1. Based on the currently estimated PD source parameters and dielectric distortion parameters, calculate the theoretical frequency domain response of the PD signal at each monitoring device.

[0073] 2. Compare the calculated theoretical frequency domain response with the actual acquired filtered frequency domain signal, and calculate the difference between the two.

[0074] 3. Utilize optimization algorithms (e.g., gradient-based optimization methods) to adjust the location, intensity, frequency domain characteristics, and dielectric distortion parameters of the PD source to minimize the difference between the theoretical response and the actual signal.

[0075] 4. Repeat the above steps until the difference converges to below the preset threshold, or the parameter change is no longer significant.

[0076] Finally, based on the PD source parameters and dielectric distortion parameters obtained after optimization and convergence, the characteristics of the PD signal source (such as discharge type and intensity) and its precise location in the underground cable section can be accurately separated. In this way, even in complex power grid dielectric environments, high-precision fault source location and feature identification can be achieved.

[0077] In some embodiments of this application, the steps of performing correlation analysis on the isolated independent signal sources and assessing the severity and development trend of the independent signal sources can be implemented in the following ways:

[0078] Pre-set detection test signals are injected into key nodes of the power grid, and the characteristics of the detection test signals and the separated independent signal sources are collected simultaneously.

[0079] Obtain operational information of external environmental stress sources and construct virtual stress curves that reflect the influence of external stress;

[0080] Based on the virtual stress curve, the propagation characteristics of the detection test signal under the corresponding stress state are analyzed, and the mapping relationship between stress state and signal distortion is constructed.

[0081] The defect signal is reverse-compensated using the mapping relationship to obtain a clean defect signal, and time-frequency domain features are extracted from the clean defect signal to obtain the clean defect signal features.

[0082] Based on the characteristics of the pure defect signal, defect correlation analysis and trend assessment are performed to generate graded early warning and maintenance strategies.

[0083] Specifically, preset detection test signals are injected into key nodes of the power grid, and the characteristics of these detection test signals and isolated independent signal sources are simultaneously acquired. Key nodes refer to nodes in the power grid that have a significant impact on overall operational stability, power flow distribution, or power supply reliability, such as substations, important load centers, or main line connection points. The preset detection test signals have known characteristics, such as frequency, amplitude, and waveform, and are used to detect the power grid's response under specific conditions. Furthermore, it is ensured that the injection of the detection test signals and the acquisition of independent signal source characteristics are precisely aligned in time to facilitate subsequent correlation analysis.

[0084] Furthermore, operational information on external environmental stress sources is acquired to construct a virtual stress curve reflecting the impact of external stress. External environmental stress sources can be understood as factors such as temperature, humidity, solar radiation intensity, wind speed, and electromagnetic interference that may affect the performance of power grid equipment and signal propagation. Operational information refers to real-time or historical data of these stress sources. The virtual stress curve is generated by comprehensively analyzing and modeling the operational information of multiple external environmental stress sources, producing a curve that quantifies and reflects the comprehensive external stress level borne by the power grid.

[0085] Based on this, using virtual stress curves, the propagation characteristics of the probe test signal under corresponding stress states are analyzed, and a mapping relationship between stress state and signal distortion is constructed. The change in propagation characteristics refers to the variation of amplitude attenuation, phase shift, and waveform distortion of the probe test signal as it is transmitted through the power grid, depending on the environmental stress state. The mapping relationship between stress state and signal distortion is established through experiments, simulations, or historical data analysis, using a mathematical model or lookup table to establish the relationship between external stress level and changes in signal propagation characteristics.

[0086] Subsequently, the defect signal is inversely compensated using the aforementioned mapping relationship to obtain a pure defect signal. Time-frequency domain features are then extracted from the pure defect signal to obtain its characteristics. Here, the defect signal refers to the characteristics of an independent signal source separated from the superimposed signals, but these characteristics may be affected by environmental stress. Inverse compensation, based on the established mapping relationship, eliminates or reduces the influence of environmental stress on the defect signal, thereby obtaining a signal closer to the characteristics of the defect source itself. Time-frequency domain features, such as the signal's energy distribution, spectral characteristics, instantaneous frequency, and wavelet coefficients, are used to comprehensively describe the dynamic behavior of the defect signal.

[0087] Finally, based on the characteristics of the pure defect signals, defect correlation analysis and trend assessment are performed to generate tiered early warning and maintenance strategies. Defect correlation analysis examines whether there are causal relationships or mutual influences between different pure defect signal characteristics; for example, whether the occurrence of one defect leads to the aggravation of another. Trend assessment predicts the changing trend of pure defect signal characteristics over time, determining whether the defect tends to stabilize, worsen, or improve. Based on the severity of the defects, their development trends, and the results of the correlation analysis, different levels of early warning measures and corresponding maintenance plans are formulated.

[0088] Through the above technical solution, this application effectively overcomes the interference of environmental stress on defect signal analysis in traditional methods, significantly improving the accuracy and reliability of defect signal feature extraction. Specifically, by constructing a mapping relationship between stress state and signal distortion and performing reverse compensation, a pure defect signal can be obtained. This allows subsequent defect correlation analysis and trend assessment to be based on more realistic defect information, thereby avoiding misjudgments or assessment biases caused by environmental factors. Consequently, the generated hierarchical early warning and maintenance strategies will be more accurate and effective, helping power grid operation and maintenance personnel to identify potential faults earlier and more accurately, optimize maintenance plans, reduce fault risks, and improve the stability and reliability of power grid operation.

[0089] In some embodiments, this application further proposes the following steps for performing defect correlation analysis and trend assessment based on the pure defect signal to generate graded early warning and maintenance strategies:

[0090] The conditional causal inference method was used to evaluate the causal relationships among features of pure defect signals;

[0091] The severity of the pure defect signal characteristics is determined by quantitative evaluation based on the characteristics of the pure defect signal, and the changes of the quantitative indicators over time are continuously tracked to analyze the evolution trend of the pure defect signal characteristics.

[0092] Based on the severity and the evolution trend, a graded early warning and maintenance strategy is generated.

[0093] Specifically, by utilizing advanced statistical and machine learning techniques, and considering specific operating scenarios or external conditions, the causal relationships between different pure defect signal characteristics can be identified and quantified. This reveals the mechanisms of interaction between defects, such as whether the occurrence or development of one defect leads to the appearance or aggravation of another defect. This method can distinguish between correlation and causation, providing a more accurate basis for subsequent severity assessment and trend prediction.

[0094] This involves using multiple dimensions of pure defect signal characteristics, such as amplitude, frequency, duration, and energy, combined with preset fault thresholds and expert experience, to numerically evaluate the current state of each independent defect, thereby determining its severity level of impact on power grid operation. For example, multiple severity levels can be defined, such as "minor," "moderate," and "severe."

[0095] Furthermore, time-series analysis is performed on the above quantitative assessment results to monitor the dynamic changes in the defect severity indicators. By analyzing the rate of change, periodicity, and abrupt change points, the future development direction and speed of the defect can be predicted, such as whether it will stabilize, worsen, or be alleviated. The purpose is to provide forward-looking information to support preventative maintenance decisions.

[0096] Therefore, by comprehensively considering the current severity of defects and their future evolution trends, differentiated early warning levels and corresponding maintenance measures are formulated. For example, for defects with high severity and a worsening trend, the highest level of early warning may be triggered, and immediate shutdown for maintenance may be recommended; for defects with moderate severity but a stable trend, regular monitoring and planned maintenance may be recommended. The aim is to achieve precise and intelligent fault management.

[0097] Through the above technical solution, this application overcomes the causal confusion problem that may exist in traditional methods in defect correlation analysis, ensuring a more accurate and objective assessment of defect severity. Continuous tracking of quantitative indicators allows for real-time monitoring of defect dynamic evolution, enabling accurate prediction of fault development trends and providing a valuable time window for power grid operation and maintenance. Furthermore, based on precise severity and evolution trend information, the generated hierarchical early warning and maintenance strategies are more refined and intelligent, effectively avoiding over-maintenance or under-maintenance, significantly improving the efficiency of power grid fault diagnosis and the scientific nature of maintenance decisions, thereby reducing operating costs and improving power supply reliability.

[0098] For example, suppose two pure defect signal features, feature A and feature B, are detected in a power grid. Traditional methods may only find a correlation between them, but cannot determine whether A causes B, B causes A, or both are caused by a common third-party factor.

[0099] First, a conditional causal inference method is employed. For example, using Granger causality tests or structural equation modeling, the time-series data of features A and B are analyzed, taking into account operating scenario parameters such as grid load and ambient temperature. If the analysis results show that, under a specific load scenario, changes in feature A significantly precede and influence changes in feature B, then it can be inferred that A is one of the causal sources of B.

[0100] Next, features A and B are quantitatively evaluated. For example, feature A's amplitude exceeds a preset threshold and is quantified as "moderate severity"; feature B has a smaller frequency offset and is quantified as "mild severity". Simultaneously, the system continuously tracks the changes in these two quantitative indicators over the past few weeks. It was found that the amplitude of feature A is slowly and steadily increasing, indicating a worsening trend; while the frequency offset of feature B remains stable.

[0101] Ultimately, based on this information, a tiered early warning and maintenance strategy are generated. Since feature A is of moderate severity and shows a worsening trend, the system may issue a Level 2 warning, recommending a detailed inspection of equipment involving feature A within the next month. While feature B exists, its severity is minor and the trend is stable; the system may only issue a Level 3 warning, recommending its inclusion in routine inspections and continuous monitoring. This strategy, based on causality and dynamic trends, can more effectively allocate maintenance resources, avoid unnecessary downtime, and intervene promptly in potential serious failures.

[0102] In some embodiments, this application further proposes the following steps for evaluating the causal relationship between pure defect signal features using the conditional causal inference method:

[0103] Obtain the operating status parameters of the power grid and classify the operating scenarios of the power grid;

[0104] Multi-timescale evolution patterns are extracted from the pure defect signal;

[0105] In each operating scenario, a nonlinear causal path graph is constructed using the multi-timescale evolution model;

[0106] Identify feedback loops in the causal path graph and quantify the time delays in the feedback loops;

[0107] By combining the aforementioned causal path diagram, feedback loop, and time delay, the cross-effect of defects is assessed.

[0108] When the rate of change of the operating status parameters exceeds a preset threshold, it is marked as a situation transition period, and the causal path graphs of adjacent situations are dynamically merged.

[0109] Specifically, real-time data on the power grid, including voltage, current, power, frequency, temperature, and humidity, are collected at different points in time. These parameters reflect the current operating status of the power grid and external environmental conditions. Based on these operating parameters, the operation of the power grid can be divided into multiple discrete operating scenarios, such as light load, heavy load, normal operation, and abnormal fluctuations. Each scenario represents a set of relatively stable operating conditions. In particular, the purified defect signal after environmental stress compensation is analyzed in depth to reveal its variation patterns and characteristics at different time granularities (e.g., short-term fluctuations, medium-term trends, and long-term evolution). This can be achieved through signal processing techniques such as time-frequency analysis, wavelet transform, and empirical mode decomposition, thereby capturing the transient, periodic, and long-term evolution characteristics of the defect signal. In practical applications, for the defect evolution patterns extracted under specific operating scenarios, nonlinear modeling methods (e.g., based on Granger causality, mutual information, or deep learning models) are used to identify the nonlinear causal relationships between different defect features and visualize them as a causal path graph. This path graph clearly shows the direction and intensity of the mutual influence between defects, as well as the possible nonlinear dependencies. Furthermore, the constructed causal path graph identifies closed-loop paths, i.e., feedback loops, where defects mutually promote or inhibit each other. For example, the deterioration of defect A may lead to the emergence of defect B, and the emergence of defect B, in turn, accelerates the further development of defect A. Simultaneously, quantifying the time delay required for the transmission of causal effects in these feedback loops is crucial for understanding the dynamic process of defect evolution. Thus, by integrating the identified nonlinear causal relationships, feedback mechanisms, and their time delays, a comprehensive assessment of the complex interactions among multiple defects in the power grid is conducted. This helps to reveal the chain reactions, synergistic effects, and potential deterioration paths among defects. As a preferred implementation, when the power grid operating state changes rapidly, such as load surges or equipment switching, causing a rapid transition from one stable situation to another, the system identifies this transition period. During this period, to maintain the continuity and accuracy of the causal assessment, the causal path graphs of multiple operating situations adjacent to the current transition situation are dynamically weighted and fused to reflect the dynamic adjustment of defect causal relationships during the transition period.

[0110] Through the aforementioned technical solutions, this application significantly improves the accuracy and robustness of assessing the causal relationships between power grid defects. By meticulously dividing operating scenarios and extracting multi-timescale evolution patterns, causal analysis can fully consider the dynamics and complexity of power grid operation, avoiding the limitations that may exist in traditional methods under a single model. Causal path diagrams, feedback loop identification, and time delay quantification allow for a deeper understanding of the complex interactions between defects, enabling more accurate prediction of defect evolution trends and potential chain reactions. Furthermore, the dynamic fusion mechanism for scenario transition periods ensures the continuity and adaptability of causal assessment when the power grid operating state changes rapidly, providing solid technical support for developing more accurate and timely graded early warning and maintenance strategies, and effectively reducing the risk of power grid failures.

[0111] For example, suppose a power grid area experiences a change in operating conditions throughout the day, from low load (nighttime) to high load (daytime peak) and then to medium load (evening), while ambient temperature and humidity also fluctuate over time. First, the system continuously acquires the power grid's operating status parameters, such as current, voltage, transformer temperature, and ambient temperature and humidity for each line. Based on these parameters, the day is divided into multiple operating conditions, including "low load stable condition," "load rising transition condition," "high load stable condition," "load falling transition condition," and "medium load stable condition." Next, the system performs multi-timescale analysis on pure defect signals separated from the power grid (e.g., partial discharge signals, insulation degradation signals), extracting their evolution patterns at different timescales such as minutes, hours, and days. Examples include the pulse repetition rate of partial discharge, discharge intensity, and the slow decreasing trend of insulation resistance. Under the high load stable condition, a nonlinear causal path graph is constructed using the extracted evolution patterns. For example, a nonlinear causal relationship was found between localized overheating of transformer windings (evolution mode of defect A) and bushing insulation degradation (evolution mode of defect B). Specifically, when the winding temperature exceeds a certain threshold, it accelerates the degradation rate of the bushing insulation. Simultaneously, potential feedback loops were identified; for example, bushing insulation degradation leads to enhanced partial discharge, which in turn further exacerbates localized overheating of the windings. The time lag of this feedback effect was quantified; for example, the impact of bushing degradation on winding overheating becomes apparent several hours later. When the power grid rapidly switches from a high-load stable scenario to a load-decreasing transition scenario (e.g., a rapid load decrease in the evening), the system detects that the current change rate exceeds a preset threshold and immediately marks it as a transition period. At this time, the system dynamically merges the causal path diagrams of the "high-load stable scenario" and the "medium-load stable scenario." During the merging process, contribution weights are dynamically calculated and allocated based on the proximity of real-time operating parameters to the centers of these two adjacent scenarios, as well as the environmental stress state similarity factor (e.g., comparing the slope variance of temperature and humidity changes during the transition period with each scenario). For example, in the early stages of load reduction, the high-load scenario has a higher weight. As the load continues to decrease, the weight of the medium-load scenario gradually increases, thereby generating a fused causal path diagram that reflects the current dynamic transition state. This ensures that the assessment of the cross-impact of defects remains accurate and continuous during scenario switching. In this way, this application can gain a more comprehensive and accurate understanding of the complex dynamic behavior of power grid defects, providing a more reliable decision-making basis for intelligent operation and maintenance of the power grid.

[0112] In some embodiments, this application further proposes the following steps for generating the tiered early warning and maintenance strategy:

[0113] Obtain real-time topology information of the power grid and dynamically construct a simulation environment for the operation of the power grid;

[0114] Identify key nodes in the power grid, including dynamically adjusting load sensitivity thresholds and fault propagation path weights based on real-time operating mode information to assess node importance;

[0115] The impact of defects on critical nodes is simulated in the operating simulation environment, including voltage dips, current overloads, and power losses.

[0116] Based on the simulation results, assess the extent of the impact of the defects on voltage stability, power flow distribution, and power supply reliability.

[0117] Based on the aforementioned scope of impact, the operational risk level of the defect is dynamically quantified.

[0118] Specifically, acquiring real-time power grid topology information refers to collecting data on the connection relationships, operating status, and physical locations of various devices in the power grid (such as substations, lines, and switches) in real time through SCADA systems, Geographic Information Systems (GIS), or other data sources. Based on this, a digital model reflecting the current connection status and physical layout of the power grid is dynamically constructed. On this basis, a dynamic simulation environment for power grid operation is built. This can be understood as using the aforementioned real-time topology information, combined with power grid load data, power output data, etc., to establish a simulation platform capable of simulating power flow calculations, fault analysis, and other operational behaviors. Its purpose is to provide a virtual test field that closely approximates the actual operating state for subsequent defect impact assessments.

[0119] In the simulated operating environment, by analyzing the structural characteristics and operating modes of the power grid, nodes that significantly impact the overall stability and reliability of the power grid are identified. Specifically, this includes dynamically adjusting load sensitivity thresholds and fault propagation path weights based on real-time operating mode information to assess node importance. Load sensitivity thresholds can be dynamically adjusted based on factors such as the current power grid load level and reserve capacity to reflect the sensitivity of nodes to load changes under different operating modes. Fault propagation path weights can be dynamically assessed based on factors such as line impedance and protection configuration to determine the likelihood and scope of a fault propagating from one node to other nodes. By integrating these dynamically adjusted parameters, the importance of each node under the current operating scenario can be more accurately evaluated.

[0120] In practical applications, identified defects (e.g., insulation aging, partial discharge, etc.) are injected into the operational simulation environment in the form of specific fault models, and their specific impacts on key nodes of the power grid are simulated. These impacts include, but are not limited to, voltage dips (node ​​voltages falling below the normal operating range); current overloads (line or equipment currents exceeding their rated capacity); and power losses (increased energy loss during transmission). The aim is to quantify the direct impact of defects on the physical and electrical parameters of the power grid under different operating conditions.

[0121] Furthermore, based on the simulation results, the impact of the defect on voltage stability, power flow distribution, and power supply reliability is assessed. Voltage stability refers to the ability of the power grid to maintain the voltage at each node within the allowable range after being disturbed; power flow distribution refers to the transmission path and magnitude of electrical energy in the power grid; and power supply reliability refers to the ability of the power grid to continuously supply power to users. Through in-depth analysis of the simulation results, regions where the defect may cause voltage instability, regions with abnormal power flow, and potential power outages can be identified, thereby comprehensively assessing the potential hazards of the defect.

[0122] Therefore, based on the aforementioned scope of impact, the operational risk level of the defect is dynamically quantified. The operational risk level can be comprehensively assessed and graded based on factors such as the scope and degree of impact of the defect on power grid operation, as well as the vulnerability of the current power grid operating mode. For example, the risk level can be divided into three levels: low, medium, and high, corresponding to different early warning responses and maintenance priorities. The purpose is to provide decision-makers with intuitive and quantifiable risk assessment results so that appropriate maintenance measures can be taken in a timely manner.

[0123] Through the above technical solutions, this application overcomes the limitations that traditional methods may have in generating early warning and maintenance strategies. Specifically, by acquiring real-time topology information and dynamically constructing an operational simulation environment, this application can accurately simulate and quantify the actual impact of defects under the current power grid operating conditions, avoiding biases that may arise from static models or empirical judgments. Identifying key nodes and dynamically adjusting their importance assessment parameters allows resources to be prioritized in areas crucial to power grid operation, improving maintenance efficiency and the rationality of resource allocation. Furthermore, by simulating the impact of defects on specific physical quantities such as voltage dips, current overloads, and power losses, and further assessing their impact on voltage stability, power flow distribution, and power supply reliability, this application provides a comprehensive, multi-dimensional risk assessment perspective, making the quantified operational risk level more accurate and instructive. Therefore, the generated graded early warning and maintenance strategies can more accurately reflect the actual threat of defects to power grid operation, thereby effectively improving the power grid's fault early warning capabilities and the scientific nature of maintenance decisions, ensuring the safe and stable operation of the power grid.

[0124] As a specific implementation, suppose that in a regional power grid, a signal separation module detects a partial discharge defect located on a transmission line. According to the basic scheme, this defect might be assessed as of moderate severity and have a slow development trend. However, in order to generate a more operational maintenance strategy, the scheme of this application will further perform the following steps:

[0125] First, the system acquires real-time topology information of the power grid in the area, including currently operating lines, substation connection status, load distribution, etc., and dynamically constructs a digital simulation model that reflects the current operating status of the power grid as an operating simulation environment.

[0126] Next, in the simulated operating environment, the system identifies the two substations connected to the transmission line and their downstream critical load nodes as critical nodes. During the identification process, the system dynamically adjusts the load sensitivity threshold and fault propagation path weights of these nodes based on the real-time operating mode of the current power grid (e.g., whether there is a large load transfer, whether important generator units are out of service, etc.) to reflect their actual importance in the current situation.

[0127] Subsequently, in the simulated operating environment, the system injects the partial discharge defect into the corresponding transmission line as a specific fault model, simulating the potential voltage drop, current overload, and power loss under different operating conditions. For example, the simulation results show that during peak load periods, the defect may cause the voltage drop at a downstream important industrial park power supply node to exceed the allowable range, and may also cause a slight current overload on adjacent lines.

[0128] Based on these simulation results, the system further assesses the impact of the defect on the entire power grid. For example, the assessment results indicate that the defect may affect the voltage stability of local areas, cause slight shifts in the power flow distribution of some lines, and, in extreme cases, may reduce the power supply reliability of the industrial park.

[0129] Ultimately, based on the stated impact range, the system dynamically quantifies the operational risk level of the defect. Since this defect may affect the power supply to a critical industrial park during peak load periods, its operational risk level is dynamically assessed as "medium-high risk," generating corresponding tiered early warnings (e.g., immediately initiating enhanced online monitoring and planning a power outage for maintenance during the next low load period) and maintenance strategies. This strategy, based on real-time simulation and dynamic risk quantification, can more accurately guide maintenance decisions compared to judging solely based on the severity of the defect itself, avoiding the risks of blind or delayed maintenance.

[0130] In some embodiments, this application further proposes that the steps of constructing a nonlinear causal path graph using the multi-timescale evolution pattern in each operating scenario include:

[0131] The rate of change and duration of power grid operating status parameters are continuously monitored. When the rate of change of the operating status parameters exceeds a preset threshold or the duration is less than a preset stabilization period, the current operating situation is marked as a situation transition period.

[0132] During the scenario transition period, multiple running scenario partitions adjacent to the current transition scenario are activated, and the contribution weight of each scenario partition is dynamically calculated based on the proximity of the real-time running status parameters to the center of each adjacent scenario partition.

[0133] In each activated adjacent context partition, a corresponding causal path graph is constructed, which reflects the nonlinear causal relationship between defect evolution modes;

[0134] The causal path graphs constructed in different scenario partitions are weighted and fused according to their corresponding contribution weights to generate a fused causal path graph. The contribution weights include an environmental stress state similarity factor, which is calculated by comparing the slope variance of the virtual stress curves of each partition.

[0135] Specifically, the system acquires and analyzes key operating parameters of the power grid in real time, such as voltage, current, power, and frequency, and calculates the rate of change of these parameters over time. When the rate of change of any parameter exceeds a preset threshold—for example, if the voltage drops by more than 5% in a short period, or if the duration of the current operating state fails to reach a preset stabilization period (for example, a stable operating situation typically requires at least 10 minutes, but the current state only lasts 2 minutes)—the current operating situation is marked as a transition period. This aims to accurately identify the dynamic process of the power grid transitioning from one stable operating state to another, avoiding the simplistic categorization of the complexity of the transition period into a single stable situation.

[0136] During the scenario transition period, the system identifies multiple predefined scenario partitions that are geographically adjacent to the current transition scenario based on the current operating state parameters. For example, if the current scenario is transitioning from "light load stable" to "heavy load stable," then the "light load stable partition," "medium load transition partition," and "heavy load stable partition" may be activated. The system calculates the distance (e.g., Euclidean or Mahalanobis distance) between the current real-time operating parameters and the center of each activated scenario partition (e.g., the average or typical value of the operating parameters within that partition) and converts this distance into weights; the closer the distance, the greater the weight. This ensures that the scenario partition most representative of the current actual operating state has a larger weight in subsequent fusion.

[0137] Within each activated adjacent context partition, multi-timescale evolution patterns extracted from historical data within that partition are used to construct a graph reflecting the nonlinear causal relationships between defect evolution patterns. These causal path graphs depict in detail how different defect characteristics influence and interact with each other under specific operational contexts.

[0138] The causal path graphs constructed in each of the aforementioned scenario partitions are linearly or nonlinearly combined based on their dynamically calculated contribution weights to form a fusion graph that comprehensively reflects the causal relationships of defects during the current scenario's transition period. The contribution weights include an environmental stress state similarity factor, which is calculated by comparing the slope variance of the virtual stress curves of each partition. Specifically, the environmental stress state similarity factor measures the similarity between the current environmental stress state and the typical environmental stress states represented by each scenario partition. For example, this similarity is quantified by comparing the slope variance of the virtual stress curve of the current environmental stress (such as temperature, humidity, vibration, etc.) with the slope variance of the corresponding virtual stress curves in the historical data of each scenario partition. The slope variance reflects the drasticness and stability of environmental stress changes; similar slope variances indicate similarity in environmental stress change patterns. This factor is incorporated into the fusion weights, ensuring that when environmental stress changes drastically or is highly similar to the environmental stress pattern of a specific scenario partition, the causal path graph of that scenario partition receives a higher weight in the fusion, thus more accurately reflecting the impact of environmental stress on the causal relationships of defects.

[0139] Through the above technical solutions, this application can significantly improve the accuracy and adaptability of defect causal relationship assessment during the transition period of power grid operation scenarios. Specifically, by accurately identifying the transition period and dynamically weighting and fusing multiple scenario partitions, the limitations of a single scenario model in dynamically changing environments are avoided, enabling the constructed causal path diagram to more realistically reflect the evolution of defects under complex operating conditions. In particular, the introduction of an environmental stress state similarity factor into the fusion weights allows the system to more comprehensively consider the impact of external environmental stress on defect causal relationships, thereby improving the robustness and reliability of causal inference. This helps to more accurately identify potential correlations between defects, more precisely predict defect development trends, and thus provide more scientific and refined decision support for the graded early warning and maintenance strategies of the power grid, effectively reducing false alarm and false negative rates, and improving the safety and reliability of power grid operation.

[0140] For example, suppose a power grid experiences a rapid increase in load from normal to heavy load during the summer peak season due to a sudden rise in temperature and a surge in air conditioning load, accompanied by short-term voltage fluctuations. During this time, the system continuously monitors the grid's operating parameters, such as voltage, current, and power. When the rate of voltage change exceeds a preset threshold (e.g., a decrease of more than 1% per minute) and the duration of this change is insufficient to constitute a stable heavy load situation, the current operating situation is marked as a transition period.

[0141] During this transition period, the system activates multiple operating scenario partitions adjacent to the current transition scenario, such as the "Normal Load Stable Partition," "Medium Load Rising Partition," and "Heavy Load Stable Partition." The system dynamically calculates the contribution weight of each partition based on the proximity of real-time voltage, current, and other parameters to the center values ​​of these three partitions. For example, if the current operating parameters are closer to the center of the "Medium Load Rising Partition," that partition will have a higher weight. Simultaneously, the system acquires operating information from external environmental stress sources, such as a sharp rise in ambient temperature, and constructs corresponding virtual stress curves. By comparing the slope variance of the current environmental stress curve with the slope variance of the corresponding virtual stress curves in the historical data of each scenario partition, an environmental stress state similarity factor is calculated.

[0142] Subsequently, within each activated scenario partition, a corresponding causal path graph is constructed. These graphs reflect the causal relationships between defect evolution modes such as partial discharge and insulation aging under that specific scenario. Finally, the causal path graphs constructed in these three scenario partitions are weighted and fused according to their dynamically calculated contribution weights (which include environmental stress state similarity factors) to generate a fused causal path graph that accurately reflects the causal relationships of defects in the current power grid under rapid load increases and high-temperature stress. In this way, even under complex dynamic transition scenarios, an accurate defect causal relationship model can be obtained, thus providing a more reliable basis for power grid fault diagnosis and preventive maintenance.

[0143] In some embodiments, this application further proposes, after the step of weighting and fusing the causal path graphs constructed in the different context partitions according to their corresponding contribution weights to generate a fused causal path graph, the method further includes:

[0144] Continuously track the consistency of changes in multi-timescale evolution patterns before and after context switching;

[0145] When a significant inconsistency is detected in the evolution pattern of a defect after a context switch, a causal relationship adaptive adjustment mechanism is activated. This mechanism prioritizes adjusting the strength and direction of the causal connections between the defect and other defects to ensure that the fused causal path diagram accurately reflects the causal relationships between the defects that are dynamically adjusted with the context.

[0146] Specifically, when power grid operating conditions change, a comparative analysis is conducted on the evolution patterns of specific defects across multiple time scales before and after the change. This can be achieved by calculating similarity indices of defect evolution patterns at different time scales (e.g., using the Dynamic Time Warping (DTW) algorithm or correlation coefficients). The aim is to promptly identify potential changes in defect behavior patterns caused by changes in operating conditions.

[0147] Specifically, when a significant inconsistency is detected in the evolution pattern of a defect after a context switch, an adaptive causal relationship adjustment mechanism is activated. That is, if the similarity index falls below a preset threshold, it indicates that the evolution pattern of the defect has undergone a non-negligible change after the context switch, requiring local or global adjustments to the fused causal path graph. The purpose of this mechanism is to ensure the dynamic adaptability of the causal path graph, enabling it to accurately reflect the true causal relationships between defects in the current operating context.

[0148] In practical applications, after activating the causal adaptive adjustment mechanism, the system prioritizes reassessing the causal connections between defects with inconsistent evolution patterns and other defects in the power grid. This may involve recalculating Granger causality, mutual information, or transit entropy metrics between the defect and other defects, and updating the corresponding connection strength and direction in the fused causal path graph based on the new calculation results. The aim is to quickly and accurately correct the causal path graph, avoiding misleading the overall correlation analysis due to inconsistent defect evolution patterns.

[0149] The proposed solution addresses the problem that traditional methods may fail to accurately reflect the true correlations between defects when the power grid operating environment dynamically changes. This is achieved by continuously tracking the consistency of defect evolution patterns after scenario switching and selectively adjusting the causal connections in the fused causal path graph. Through this dynamic adaptive adjustment, the accuracy and reliability of defect correlation analysis can be ensured.

[0150] By employing the aforementioned technical solutions, potential inconsistencies in defect evolution patterns can be promptly identified when power grid operating conditions change. This allows for dynamic adjustments to the causal relationships between defects, significantly improving the accuracy and adaptability of the fused causal path diagram. This enables more precise correlation analysis and trend assessment of independent signal sources, providing a more reliable basis for power grid fault early warning and maintenance strategies. It effectively avoids misjudgments or omissions caused by changing circumstances, thereby enhancing the safety and stability of power grid operation.

[0151] Suppose the power grid switches from a "normal operation scenario" to a "heavy load operation scenario." Before the switch, the evolution pattern of a specific defect (e.g., transformer partial discharge) shows a slow growth. After the scenario switch, the system continuously monitors the multi-timescale evolution pattern of this partial discharge defect. If the system detects a sudden acceleration in the evolution pattern of this partial discharge defect under the heavy load scenario, and a significant inconsistency with the historical evolution pattern under the normal scenario (e.g., the Euclidean distance or correlation coefficient of its time-frequency characteristics is below a preset threshold), a causal relationship adaptive adjustment mechanism will be immediately activated. At this time, the system will prioritize reassessing the causal connection strength and direction between this partial discharge defect and other related defects in the power grid (such as insulation aging, overheating, etc.). For example, it may find that the causal connection strength between partial discharge and insulation aging is significantly enhanced, or that partial discharge may become the main cause of the accelerated development of other defects. In this way, the fused causal path diagram can be dynamically updated to accurately reflect the new and closer causal relationships between partial discharge defects and other defects under the heavy load scenario, thereby providing maintenance personnel with more accurate early warning information and maintenance recommendations.

[0152] On the other hand, such as Figure 2 As shown, an exemplary structural schematic diagram of a circuit troubleshooting fault performance testing system is illustrated. This application proposes a circuit troubleshooting fault performance testing system 100, comprising:

[0153] The signal acquisition module 10 is used to acquire operating signals from multiple locations in the power grid and to time-align the operating signals.

[0154] The structural information acquisition module 20 is used to acquire the physical structure information of the power grid and determine the signal propagation characteristics of the signal in the physical structure of the power grid based on real-time operating parameters and dynamic correction.

[0155] The signal separation module 30 is used to separate the features and locations of multiple independent signal sources from the superimposed signals based on the time-aligned running signals, the physical structure information, and the signal propagation characteristics, by jointly optimizing the defect source parameters and the medium distortion parameters.

[0156] The correlation analysis module 40 is used to perform correlation analysis on the isolated independent signal sources and to evaluate the severity and development trend of the independent signal sources, wherein the correlation analysis is based on the pure defect signal after environmental stress compensation.

[0157] This system, through the coordinated operation of its various functional modules, effectively addresses the challenges of traditional circuit maintenance methods in identifying latent defects, distinguishing between true and false faults, and handling the interaction of multiple defects. The signal acquisition module comprehensively collects and synchronizes power grid operation data, providing an accurate foundation for subsequent analysis. The structural information acquisition module accurately models signal propagation characteristics by combining the power grid's physical structure with real-time operating parameters. Based on this, the signal separation module uses a joint optimization algorithm to accurately identify and separate independent defect signal sources from complex superimposed signals. Finally, the correlation analysis module conducts in-depth analysis of these separated, pure defect signals, assessing their severity and development trends. This enables early and accurate warning and diagnosis of latent faults in the power grid, significantly improving the reliability and safety of power grid operation.

[0158] The circuit fault detection system proposed in this application forms a complete and efficient fault detection solution through the close collaboration of its signal acquisition module, structural information acquisition module, signal separation module, and correlation analysis module. Compared with traditional circuit repair methods, this system can more effectively separate independent defect sources from superimposed complex signals and perform pure analysis to remove environmental interference, thereby significantly improving the accuracy of latent defect identification, location precision, and trend prediction capability. For example, in traditional methods, a partial discharge signal generated by a cable joint in a high-temperature and high-humidity environment may be misjudged as defect deterioration. However, this system, through environmental stress compensation, can distinguish whether the signal enhancement is caused by environmental factors or whether the defect itself is actually deteriorating. This refined processing capability enables the system to more effectively identify "latent defects" that do not immediately cause power outages but continuously affect power supply quality, and to provide early warning and maintenance for them. This provides more reliable technical support for early warning and preventive maintenance of the power grid, effectively improving the reliability and safety of power grid operation.

[0159] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for testing the performance of circuit troubleshooting, characterized in that, include: Acquire operating signals from multiple locations in the power grid and perform time alignment on the operating signals; The physical structure information of the power grid is obtained, and the signal propagation characteristics of the signal in the physical structure of the power grid are determined by dynamically correcting the real-time operating parameters. Based on the time-aligned running signal, the physical structure information, and the signal propagation characteristics, the features and locations of multiple independent signal sources are separated from the superimposed signal by jointly optimizing the defect source parameters and the medium distortion parameters. Correlation analysis is performed on the isolated independent signal sources, and the severity and development trend of the independent signal sources are evaluated, wherein the correlation analysis is based on the pure defect signals after environmental stress compensation.

2. The circuit troubleshooting and fault performance testing method according to claim 1, characterized in that, The steps of separating the features and locations of multiple independent signal sources from the superimposed signal by jointly optimizing defect source parameters and medium distortion parameters, based on the time-aligned running signal, the physical structure information, and the signal propagation characteristics, include: Convert time-aligned operating signals into a frequency domain representation and filter out frequency components of normal transient events in the power grid; Based on the physical structure of the power grid and the dynamically corrected signal propagation characteristics, a frequency domain propagation model containing potential signal sources is constructed. By jointly optimizing the defect source parameters and the medium distortion parameters, an independent signal is separated from the filtered frequency domain signal.

3. The circuit troubleshooting and fault performance testing method according to claim 2, characterized in that, The step of separating the features and locations of independent signal sources from the filtered frequency domain signal by jointly optimizing the defect source parameters and the medium distortion parameters includes: The dielectric distortion parameters are added as optimization variables to the frequency domain propagation model to iteratively adjust the number, location, and frequency domain characteristics of potential signal sources. Calculate the theoretical frequency domain response of the potential signal source at each monitoring device in the power grid; The theoretical frequency domain response is matched with the filtered frequency domain signal, and the parameters of multiple potential signal sources are optimized by minimizing the difference between the two until convergence. Based on the optimized parameters of the potential signal sources, the characteristics and locations of the independent signal sources are separated.

4. The circuit troubleshooting and fault performance testing method according to claim 1, characterized in that, The steps of performing correlation analysis on the isolated independent signal sources and assessing the severity and development trend of the independent signal sources include: Pre-set detection test signals are injected into key nodes of the power grid, and the characteristics of the detection test signals and the separated independent signal sources are collected simultaneously. Obtain operational information of external environmental stress sources and construct virtual stress curves that reflect the influence of external stress; Based on the virtual stress curve, the propagation characteristics of the detection test signal under the corresponding stress state are analyzed, and the mapping relationship between stress state and signal distortion is constructed. The defect signal is reverse-compensated using the mapping relationship to obtain a clean defect signal, and time-frequency domain features are extracted from the clean defect signal to obtain the clean defect signal features. Based on the characteristics of the pure defect signal, defect correlation analysis and trend assessment are performed to generate graded early warning and maintenance strategies.

5. The circuit troubleshooting and fault performance testing method according to claim 4, characterized in that, The steps of performing defect correlation analysis and trend assessment based on the characteristics of the pure defect signal to generate graded early warning and maintenance strategies include: The conditional causal inference method was used to evaluate the causal relationships among features of pure defect signals; The severity of the pure defect signal characteristics is determined by quantitative evaluation based on the characteristics of the pure defect signal, and the changes of the quantitative indicators over time are continuously tracked to analyze the evolution trend of the pure defect signal characteristics. Based on the severity and the evolution trend, a graded early warning and maintenance strategy is generated.

6. The circuit troubleshooting and fault performance testing method according to claim 5, characterized in that, The steps for evaluating the causal relationships between features of pure defect signals using the conditional causal inference method include: Obtain the operating status parameters of the power grid and classify the operating scenarios of the power grid; Multi-timescale evolution patterns are extracted from the pure defect signal; In each operating scenario, a nonlinear causal path graph is constructed using the multi-timescale evolution model; Identify feedback loops in the causal path graph and quantify the time delays in the feedback loops; By combining the aforementioned causal path diagram, feedback loop, and time delay, the cross-effect of defects is assessed. When the rate of change of the operating status parameters exceeds a preset threshold, it is marked as a situation transition period, and the causal path graphs of adjacent situations are dynamically merged.

7. The circuit fault detection method according to claim 5, characterized in that, The steps for generating the tiered early warning and maintenance strategy include: Obtain real-time topology information of the power grid and dynamically construct a simulation environment for the operation of the power grid; Identify key nodes in the power grid, including dynamically adjusting load sensitivity thresholds and fault propagation path weights based on real-time operating mode information to assess node importance; The impact of defects on critical nodes is simulated in the operating simulation environment, including voltage dips, current overloads, and power losses. Based on the simulation results, assess the extent of the impact of the defects on voltage stability, power flow distribution, and power supply reliability. Based on the aforementioned scope of impact, the operational risk level of the defect is dynamically quantified.

8. The circuit troubleshooting performance testing method according to claim 6, characterized in that, The step of constructing a nonlinear causal path graph using the multi-timescale evolution model in each operating scenario includes: The rate of change and duration of power grid operating status parameters are continuously monitored. When the rate of change of the operating status parameters exceeds a preset threshold or the duration is less than a preset stabilization period, the current operating situation is marked as a situation transition period. During the scenario transition period, multiple running scenario partitions adjacent to the current transition scenario are activated, and the contribution weight of each scenario partition is dynamically calculated based on the proximity of the real-time running status parameters to the center of each adjacent scenario partition. In each activated adjacent context partition, a corresponding causal path graph is constructed, which reflects the nonlinear causal relationship between defect evolution modes; The causal path graphs constructed in different scenario partitions are weighted and fused according to their corresponding contribution weights to generate a fused causal path graph. The contribution weights include an environmental stress state similarity factor, which is calculated by comparing the slope variance of the virtual stress curves of each partition.

9. The circuit troubleshooting and fault performance testing method according to claim 8, characterized in that, After the step of weighting and fusing the causal path graphs constructed in different context partitions according to their corresponding contribution weights to generate a fused causal path graph, the method further includes: Continuously track the consistency of changes in multi-timescale evolution patterns before and after context switching; When a significant inconsistency is detected in the evolution pattern of a defect after a context switch, a causal relationship adaptive adjustment mechanism is activated. This mechanism prioritizes adjusting the strength and direction of the causal connections between the defect and other defects to ensure that the fused causal path diagram accurately reflects the causal relationships between the defects that are dynamically adjusted with the context.

10. A circuit fault detection system, employing the circuit fault detection method as described in any one of claims 1-9, characterized in that, The system includes: The signal acquisition module is used to acquire operating signals from multiple locations in the power grid and to time-align the operating signals. The structural information acquisition module is used to acquire the physical structural information of the power grid and determine the signal propagation characteristics of the signal in the physical structure of the power grid based on real-time operating parameters and dynamic correction. The signal separation module is used to separate the features and locations of multiple independent signal sources from the superimposed signals based on the time-aligned running signals, the physical structure information, and the signal propagation characteristics, by jointly optimizing the defect source parameters and the medium distortion parameters. The correlation analysis module is used to perform correlation analysis on the isolated independent signal sources and to evaluate the severity and development trend of the independent signal sources. The correlation analysis is based on the pure defect signal after environmental stress compensation.

Images