Power failure pre-judgment method and system based on digital twinning

Abstract

The invention provides a power failure pre-judgment method and system based on digital twinning. The method comprises the following steps: firstly, acquiring first monitoring data and second monitoring data, and performing time synchronization processing to obtain a target acoustic signal and a target current signal; secondly, inputting the target acoustic signal and the target current signal into a digital twinborn model; then in a digital twinborn model, comparing the acoustic signal with a reference acoustic mode to identify a first abnormal feature, and comparing the current signal with a theoretical current curve to identify a second abnormal feature; then calculating an associated fault probability value according to the first abnormal feature and the second abnormal feature; and finally, when the associated fault probability value is greater than a preset probability threshold value, outputting a fault early warning instruction. According to the technical scheme provided by the invention, accurate early warning and root positioning of associated faults are realized, and the problems of fuzzy early warning and lack of fault path description in the prior art are solved.

Description

Technical Field

[0001] This application relates to the field of power fault prediction technology, and in particular to a power fault prediction method and system based on digital twins. Background Technology

[0002] In the field of power system operation and maintenance, the reliable operation of key equipment in high-voltage substations is of paramount importance. With the development of smart grids, the technical need to provide early warning of latent equipment faults and prevent them from evolving into systemic power outages is becoming increasingly urgent. Traditional periodic maintenance is blind, while post-incident maintenance is costly.

[0003] Currently, one existing solution to address the above needs is a fault prediction method based on multi-sensor data and artificial intelligence models. This method collects equipment operating data by deploying vibration sensors and current sensors on transformers and circuit breakers, respectively. It integrates multi-source data using data fusion technology and inputs it into a machine learning model, such as a deep neural network, trained on historical fault data for analysis. This machine learning model identifies abnormal patterns in the data and outputs an equipment-level health score or fault probability. When the probability exceeds a threshold, an early warning is triggered.

[0004] However, this existing solution has certain shortcomings. First, its analysis process often focuses on the independent state assessment of a single device, lacking explicit modeling of the fault propagation relationship between related devices. For example, it is difficult to distinguish whether the abnormal vibration of the transformer is due to its own core fault or a chain reaction caused by the impact of the operation of an adjacent circuit breaker. Second, the solution relies on a data-driven black box model, and its judgment logic is not transparent. The early warning results lack clear indication of the physical root cause of the fault, such as whether it is a loose core or contact erosion and its propagation path in the power grid topology. This makes it difficult for maintenance personnel to quickly locate the core fault point and formulate accurate maintenance strategies. Summary of the Invention

[0005] This application provides a power fault prediction method and system based on digital twins to solve the problems of ambiguous fault early warning and location and lack of characterization of fault transmission paths between related equipment in the prior art.

[0006] Firstly, this application provides a power fault prediction method based on digital twins, including: Acquire the first and second monitoring data; The first and second monitoring data are time-synchronized to obtain the target acoustic signal and the target current signal. The target acoustic signal and target current signal are input into a digital twin model, wherein the digital twin model includes a structural dynamics model of a transformer and an operating mechanism model of a circuit breaker; In the digital twin model, the acoustic signal is compared with the reference acoustic pattern pre-stored in the voiceprint database to identify the first abnormal feature corresponding to the transformer mechanical fault, and the current signal is compared with the theoretical current curve of the operating mechanism model to identify the second abnormal feature corresponding to the circuit breaker electrical fault. Calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature; When the associated fault probability value is greater than a preset probability threshold, a fault warning command is output.

[0007] Optionally, acquiring first monitoring data and second monitoring data includes: The original vibration waveform data of the transformer is collected by multiple vibration sensors arranged on the wall of the transformer tank at a first preset sampling frequency. The original current waveform data of the circuit breaker is collected at a second preset sampling frequency by a current sensor connected in series in the circuit breaker's opening and closing coil circuit. The original vibration waveform data is transformed in the frequency domain to obtain vibration spectrum data, and the spectral amplitude within a preset frequency band is extracted from the vibration spectrum data. The initial raw current waveform data is segmented to divide the current rise phase data and current stabilization phase data during circuit breaker operation. The spectral amplitude within the preset frequency band is marked as the first monitoring data; The duration of the current rising phase data and the average current value of the current stabilizing phase data are jointly labeled as the second monitoring data.

[0008] Optionally, the first monitoring data and the second monitoring data are time-synchronized to obtain the target acoustic signal and the target current signal, including: Receive the first monitoring data and the second monitoring data, wherein both the first monitoring data and the second monitoring data are accompanied by clock information; Based on the clock information, the first time point in the first monitoring data where the spectrum amplitude suddenly changes is identified, and the second time point in the second monitoring data where the current rise phase begins is identified; Calculate the initial time difference between the first time point and the second time point; The initial time difference is compared with a preset time difference threshold. If the initial time difference is less than or equal to the time difference threshold, the first monitoring data and the second monitoring data are determined to be a synchronized data group. The spectral amplitude of the first monitoring data identified as a synchronous data group is interpolated to obtain the target acoustic signal. At the same time, the current waveform data corresponding to the duration of the current rise phase is marked as the target current signal.

[0009] Optionally, in the digital twin model, the acoustic signal is compared with a reference acoustic pattern pre-stored in the voiceprint database to identify a first abnormal feature corresponding to a transformer mechanical fault, and the current signal is compared with the theoretical current curve of the operating mechanism model to identify a second abnormal feature corresponding to a circuit breaker electrical fault, including: The target acoustic signal is input into the structural dynamics model in the digital twin model to obtain the multi-point simulated vibration response corresponding to the transformer core and windings; The target current signal is input into the operating mechanism model in the digital twin model to obtain the simulated motion trajectory of the circuit breaker moving contact; The target reference acoustic pattern matching the current transformer model is retrieved from the acoustic pattern database. The reference acoustic pattern contains the reference spectral distribution of core vibration under normal conditions. The multi-point simulated vibration response is synthesized according to spatial location to obtain the overall vibration characteristic spectrum of the transformer tank, and the amplitude deviation between the overall vibration characteristic spectrum and the reference spectrum at the preset characteristic frequency point is calculated. The simulated motion trajectory is compared with the theoretical motion trajectory pre-stored in the operating mechanism model to obtain the time delay of the moving contact during the opening and closing process. The characteristic frequency point corresponding to the amplitude deviation exceeding the first preset threshold is marked as the first abnormal feature; The motion phase in which the time delay exceeds the second preset threshold is marked as the second abnormal feature.

[0010] Optionally, based on the first abnormal feature and the second abnormal feature, a related fault probability value is calculated, including: Obtain the set of feature frequency points corresponding to the first abnormal feature and the set of motion stages corresponding to the second abnormal feature; The transformer core condition score is calculated based on the amplitude deviation corresponding to each characteristic frequency point in the set of characteristic frequency points. The circuit breaker contact status score is calculated based on the time delay corresponding to each motion stage in the set of motion stages. Based on the transformer core condition score and the circuit breaker contact condition score, an initial correlation probability value is calculated using a weighted fusion method. Obtain the electrical distance information between the transformer and the circuit breaker recorded in the digital twin model; The initial association probability value is corrected based on the electrical distance information to obtain the associated fault probability value.

[0011] Optionally, based on the transformer core condition score and the circuit breaker contact condition score, an initial correlation probability value is calculated using a weighted fusion method, including: Based on the connection relationship between transformers and circuit breakers in the power grid topology, the first weighting coefficient corresponding to the transformer core condition score and the second weighting coefficient corresponding to the circuit breaker contact condition score are determined. The transformer core condition score is multiplied by the first weighting coefficient to obtain the first weighted score; The circuit breaker contact status score is multiplied by the second weighting coefficient to obtain the second weighted score; The first weighted score and the second weighted score are summed to obtain the comprehensive score. The comprehensive score value is input into a preset probability conversion function to output an initial correlation probability value.

[0012] Optionally, when the associated fault probability value is greater than a preset probability threshold, a fault warning instruction is output, including: When the associated fault probability value is greater than the preset probability threshold, the corresponding warning level is determined according to the different numerical ranges in which the associated fault probability value is located. Based on the set of characteristic frequency points corresponding to the first abnormal feature, the potential fault area of ​​the transformer core is determined. Based on the set of motion stages corresponding to the second abnormal feature, the potential fault type of the circuit breaker contacts is determined; Based on the electrical connection relationship between the transformer and the circuit breaker recorded in the digital twin model, information on the scope of the fault impact is generated; The warning level, potential fault area, potential fault type, and fault impact range information are combined to output a fault warning command.

[0013] Secondly, this application provides a power fault prediction system based on digital twins, comprising: The acquisition module is used to acquire the first monitoring data and the second monitoring data; The processing module is used to perform time synchronization processing on the first monitoring data and the second monitoring data to obtain the target acoustic signal and the target current signal; An input module is used to input the target acoustic signal and the target current signal into a digital twin model, wherein the digital twin model includes a structural dynamics model of a transformer and an operating mechanism model of a circuit breaker; The identification module is used to compare the acoustic signal with a reference acoustic pattern pre-stored in the voiceprint database in the digital twin model to identify a first abnormal feature corresponding to a transformer mechanical fault, and to compare the current signal with the theoretical current curve of the operating mechanism model to identify a second abnormal feature corresponding to a circuit breaker electrical fault. The calculation module is used to calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature; The output module is used to output a fault warning command when the associated fault probability value is greater than a preset probability threshold.

[0014] Thirdly, this application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a power fault prediction method based on digital twins as described in the first aspect above.

[0015] Fourthly, this application provides a computer storage medium storing a computer program, which, when executed by a computer, implements a power fault prediction method based on digital twins as described in the first aspect.

[0016] The beneficial effect of this application is that by constructing a digital twin model that integrates the power grid topology and equipment physical constraints, and by co-analyzing the transformer vibration acoustic signal and the circuit breaker operating current signal, it is possible to accurately identify the mechanical fault characteristics of the transformer core and the electrical fault characteristics of the circuit breaker contacts. On this basis, by calculating the correlation fault probability value between the two, a quantitative assessment of the fault risk of the transformer-circuit breaker, a key electrical connection node, is achieved. This method not only focuses on the independent state of individual devices, but also emphasizes revealing the transmission relationship of faults between related devices, thereby solving the problem of ambiguous fault early warning and location in the prior art.

[0017] Furthermore, when the associated fault probability value exceeds the threshold, the output fault warning command integrates multi-dimensional information such as warning level, potential fault area, potential fault type, and fault impact range. This enables maintenance personnel to clearly distinguish whether the root cause of the fault originates from the internal mechanical structure of the transformer or the circuit breaker operating mechanism, and to predict the possible propagation path of the fault in the power grid. This precise fault location and risk path characterization improves the interpretability and action guidance value of the warning information, effectively overcoming the deficiency of existing solutions in lacking the characterization of fault propagation paths between associated equipment. Therefore, this application can achieve precise early warning and root cause location of faults at the connection points of key equipment in high-voltage substations, providing a reliable basis for proactive maintenance decisions.

[0018] These or other aspects of this application will become more apparent from the description of the following embodiments. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart of a power fault prediction method based on digital twins provided in this application is shown; Figure 2 A schematic diagram of the structure of a power fault prediction system based on digital twin provided in this application is shown; Figure 3 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present application, the technical solution of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0022] In some of the processes described in the specification, claims, and accompanying drawings of this application, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as 101, 102, etc., are merely used to distinguish different operations and do not themselves represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a chronological order, nor do they limit "first" and "second" to different types.

[0023] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] Figure 1 This application provides a flowchart of a power fault prediction method based on digital twins, as shown below. Figure 1 As shown, the method includes: Step 101: Obtain the first monitoring data and the second monitoring data.

[0025] Optionally, step 101 may specifically include: Step 1011: Collect the original vibration waveform data of the transformer at a first preset sampling frequency using multiple vibration sensors arranged on the transformer tank wall.

[0026] Step 1012: The original current waveform data of the circuit breaker is acquired at a second preset sampling frequency by a current sensor connected in series in the circuit breaker's opening and closing coil circuit.

[0027] Step 1013: Perform frequency domain transformation on the original vibration waveform data to obtain vibration spectrum data, and extract the spectral amplitude within a preset frequency band from the vibration spectrum data.

[0028] Step 1014: The initial raw current waveform data is segmented to divide the current rising phase data and the current stabilizing phase data during circuit breaker operation.

[0029] Step 1015: Mark the spectral amplitude within the preset frequency band as the first monitoring data.

[0030] Step 1016: The duration of the current rising phase data and the average current value of the current stabilizing phase data are jointly marked as the second monitoring data.

[0031] In this step, the first monitoring data refers to the characteristic data extracted from the transformer vibration signal to characterize its mechanical state, which is used for subsequent fault analysis.

[0032] The second monitoring data refers to the characteristic data extracted from the circuit breaker operating current signal, which is used to characterize its electrical and mechanical state and is used for subsequent fault analysis.

[0033] A vibration sensor is a device that can measure the vibration signal of an object, and is used to collect the mechanical vibration of the surface of a transformer.

[0034] The first preset sampling frequency refers to the data acquisition speed preset when acquiring transformer vibration signals. It is used to ensure that effective vibration details can be captured and is determined through the settings of the data acquisition system.

[0035] Raw vibration waveform data refers to the unprocessed signal sequence of transformer surface vibration over time, directly acquired by vibration sensors, and is used for subsequent analysis.

[0036] The opening and closing coil circuit refers to the electrical path in a circuit breaker where the operating mechanism that controls the connection or disconnection of the circuit is located. It is used to supply current to drive the circuit breaker to operate and is the location where the current sensor is connected.

[0037] A current sensor is a device that can measure current signals in a circuit and is used to collect current changes during circuit breaker operation.

[0038] The second preset sampling frequency refers to the data acquisition speed preset when acquiring circuit breaker current signals. It is used to ensure that details of current changes can be captured and is determined through the settings of the data acquisition system.

[0039] Raw current waveform data refers to the unprocessed signal sequence of circuit breaker operating coil current changing over time, directly acquired by the current sensor, and used for subsequent analysis.

[0040] Vibration spectrum data refers to the data obtained after converting the original vibration waveform data from the time domain to the frequency domain. It is used to display the distribution of vibration energy at different frequency points and is obtained by processing the original vibration waveform data using the Fast Fourier Transform algorithm.

[0041] The preset frequency band range refers to one or more frequency intervals in the vibration spectrum data that are pre-set and related to specific mechanical faults of the transformer. It is used to focus on analyzing key fault characteristics and is determined in advance according to the transformer model and its typical fault characteristic frequencies.

[0042] Spectral amplitude refers to the intensity of a vibration signal at a specific frequency point or within a frequency band in vibration spectrum data. It is used to quantify the significance of fault characteristics and is obtained directly from vibration spectrum data or calculated from it.

[0043] Current rise phase data refers to the current data in the original current waveform data from the time the current begins to appear until it reaches its peak value, and is used to analyze the characteristics of the circuit breaker starting process.

[0044] Current stabilization phase data refers to the current data during the period when the current remains relatively stable after reaching its peak in the original current waveform data. It is used to analyze the characteristics of the circuit breaker's holding state.

[0045] Duration refers to the length of time corresponding to the current rise phase data, which is used to measure the speed of the circuit breaker's start-up process. It is obtained by calculating the time difference between the start and end points of the current rise phase data.

[0046] Average current value: refers to the arithmetic mean of all current points in the current stabilization phase data. It is used to characterize the current level when the circuit breaker is operating stably. It is obtained by summing all the values ​​in the current stabilization phase data and dividing by the number of data points.

[0047] In this step, firstly, multiple vibration sensors arranged on the transformer tank wall continuously collect raw vibration waveform data generated during transformer operation at a pre-set first sampling frequency. Simultaneously, current sensors connected in series in the circuit breaker's opening and closing coil circuit collect raw current waveform data flowing through the coil when the circuit breaker performs opening or closing operations at a pre-set second sampling frequency. Secondly, the collected raw vibration waveform data is converted from a time-domain signal to a frequency-domain signal using a fast Fourier transform algorithm, thereby obtaining vibration spectrum data that clearly shows the energy distribution of each frequency component. Subsequently, the spectral amplitude within the pre-set frequency band range is extracted from this vibration spectrum data based on a pre-determined frequency band range determined according to the transformer fault mechanism. Then, the collected raw current waveform data is analyzed to identify the current rise phase, in which the current increases sharply from zero to reach its peak, and the current stabilization phase, in which the current remains basically unchanged after reaching its peak. Thus, the complete waveform is divided into current rise phase data and current stabilization phase data. Finally, the spectral amplitude within the preset frequency band range extracted from the vibration spectrum data is marked as the first monitoring data for characterizing the transformer state. At the same time, the duration corresponding to the current rise phase data is calculated, and the average current value of all points in the current stabilization phase data is calculated. These two parameters are jointly marked as the second monitoring data for characterizing the circuit breaker state.

[0048] For example, in a substation, maintenance personnel first install several vibration sensors on the outer wall of the tank of main transformer A, setting the sampling frequency to 10kHz to collect the raw vibration waveform data of transformer A. Simultaneously, a current sensor is connected in series to the opening and closing coil circuit of circuit breaker B, which is connected to transformer A, and the sampling frequency is set to 50kHz to collect the raw current waveform data when circuit breaker B performs a closing operation. Next, a fast Fourier transform is performed on the raw vibration waveform data of transformer A to obtain its vibration spectrum, and the spectral amplitude within a preset frequency band of 100Hz to 500Hz, related to the core loosening fault, is extracted. Then, the raw current waveform data of circuit breaker B is analyzed to identify the current rise phase from 0A to 5A, and the current stabilization phase where the current stabilizes at around 5A. Finally, the extracted 100-500Hz spectral amplitude is marked as the first monitoring data, and the duration of the current rise phase (0.1 seconds) and the average current value of the current stabilization phase (5.1A) are jointly marked as the second monitoring data.

[0049] Step 102: Perform time synchronization processing on the first monitoring data and the second monitoring data to obtain the target acoustic signal and the target current signal.

[0050] Optionally, step 102 may specifically include: Step 1021: Receive the first monitoring data and the second monitoring data, wherein both the first monitoring data and the second monitoring data are accompanied by clock information.

[0051] Step 1022: Based on the clock information, identify the first time point in the first monitoring data where the spectral amplitude changes abruptly, and identify the second time point in the second monitoring data where the current rise phase begins.

[0052] Step 1023: Calculate the initial time difference between the first time point and the second time point.

[0053] Step 1024: Compare the initial time difference with a preset time difference threshold. If the initial time difference is less than or equal to the time difference threshold, then determine that the first monitoring data and the second monitoring data are a synchronized data group.

[0054] Step 1025: Interpolate the spectral amplitude of the first monitoring data that is determined to be a synchronous data group to obtain the target acoustic signal, and mark the current waveform data corresponding to the duration of the current rise phase as the target current signal.

[0055] In this step, the target acoustic signal refers to the transformer vibration characteristic signal that has undergone time synchronization and interpolation processing and precisely corresponds to the circuit breaker operation time, and is used for subsequent collaborative analysis in the digital twin model.

[0056] The target current signal refers to the circuit breaker operating current signal segment that corresponds precisely in time to the transformer vibration characteristic signal, and is used for subsequent collaborative analysis in the digital twin model.

[0057] Clock information refers to the timestamp data recorded by the data acquisition system when collecting the first and second monitoring data, which indicates the time of data generation. It is used to determine the temporal relationship of data from different sources and is obtained by the clock module built into the data acquisition system during data collection.

[0058] The first time point refers to the specific moment when the spectral amplitude of the vibration signal corresponding to the first monitoring data changes significantly. It is used to identify the key time point when the transformer may be subjected to electrical operation shocks. The abrupt change point is identified by analyzing the curve of the spectral amplitude changing with time in the first monitoring data.

[0059] The second time point refers to the specific moment in the current signal corresponding to the second monitoring data when the circuit breaker operating current begins to rise. It is used to identify the starting time point of the circuit breaker operation. This is obtained by analyzing the original current waveform data corresponding to the second monitoring data to identify the starting point of the current increase from zero.

[0060] The initial time difference refers to the absolute value of the time interval between the first and second time points. It is used to measure the consistency between the sudden change in transformer vibration and the operation of the circuit breaker in time. It is obtained by subtracting the value of the first time point from the value of the second time point and taking the absolute value.

[0061] The preset time difference threshold refers to the maximum allowable time interval set in advance to determine whether two events can be considered to occur synchronously in time. It is used to filter out unrelated data pairs with weak time correlation and is set in advance according to the physical laws of stress wave propagation speed and response time in the power system.

[0062] Synchronous data groups refer to the data pairs in the first and second monitoring data whose initial time difference is less than or equal to a preset time difference threshold. This means that these data originate from a collaborative event that may have a causal relationship.

[0063] In this step, the first and second monitoring data are received first, and both sets of data are accompanied by precise clock information at the time of acquisition. Next, this precise clock information is used to analyze the first monitoring data, specifically by examining the sequence of its spectral amplitude changes over time, identifying the moment when the amplitude suddenly increases, and recognizing this as the first time point. Simultaneously, the original current waveform corresponding to the second monitoring data is analyzed to find the starting moment when the current significantly increases from zero or a steady-state value, and this is recognized as the second time point. Then, the value at the second time point is subtracted from the value at the first time point, and the absolute difference is calculated to obtain the initial time difference. Then, the calculated initial time difference is compared with a pre-set time difference threshold. If the initial time difference is less than or equal to the threshold, it is determined that the pair of first and second monitoring data being processed are related, reflecting a potentially related event, and thus they are marked as a synchronous data group. Finally, for the pair of data determined to be synchronous data groups, the spectral amplitude time series in the first monitoring data is interpolated, such as by using a linear interpolation algorithm, to make the time interval between its data points uniform, or to align it with the number of data points of the target current signal. The processed signal is the target acoustic signal. At the same time, the original current waveform data corresponding to the current rising phase in the second monitoring data is directly marked as the target current signal.

[0064] For example, following the specific implementation of the previous step, firstly, a first monitoring data 100-500Hz spectrum amplitude sequence marked with time T1 and a second monitoring data marked with time T2, containing information on the current rising phase and the stable phase, are received; secondly, the first monitoring data is analyzed, and a peak in the spectrum amplitude is found at T1+0.015 seconds, which is recorded as the first time point; then, the current waveform corresponding to the second monitoring data is analyzed, and the current begins to rise at T2+0.010 seconds, which is recorded as the second time point; then, the initial time difference is calculated to be 0.005 seconds. If the preset time difference threshold is assumed to be 0.02 seconds, since 0.005 seconds is less than 0.02 seconds, this pair of data is determined to be a synchronous data group; subsequently, the spectrum amplitude sequence in the first monitoring data is interpolated to generate the target acoustic signal, and the data segment in the current waveform that lasts for 0.1 seconds starting from T2+0.010 seconds is marked as the target current signal.

[0065] This step effectively filters out highly correlated and potentially causally related data pairs from the massive asynchronously acquired data by comparing and analyzing the time difference between the abrupt change in transformer vibration characteristics and the start of circuit breaker operation. The filtered vibration characteristic data is then interpolated and shaped to ensure precise time matching with the current signal. This process ensures that the acoustic and current signals used in subsequent digital twin model analysis originate from the same event, providing a strict time synchronization guarantee for accurately analyzing the fault correlation between the transformer and the circuit breaker, and avoiding misjudgments caused by time asynchrony.

[0066] Step 103: Input the target acoustic signal and target current signal into the digital twin model, wherein the digital twin model includes a structural dynamics model of the transformer and an operating mechanism model of the circuit breaker.

[0067] In this step, the digital twin model refers to a high-fidelity simulation model built in virtual space that can dynamically reflect the status and behavior of physical entities such as high-voltage substations and their equipment. It is used to simulate and analyze the operation process and fault evolution of physical entities and is constructed by integrating geometric models, physical laws, operational data, and historical data.

[0068] Structural dynamics model refers to a sub-model in digital twin model that specifically simulates the vibration response characteristics of the internal mechanical structure of a transformer, such as the core and windings, under the action of electromagnetic and mechanical forces. It is used to deduce its vibration behavior based on the input electrical state and is achieved by establishing mass, stiffness, and damping parameters and solving dynamic differential equations.

[0069] The operating mechanism model refers to a sub-model in the digital twin model that specifically simulates the mechanical motion characteristics of the operating mechanism, such as coils, connecting rods, and contacts, and their relationship with current during the opening and closing operation of a circuit breaker. It is used to deduce the contact motion state based on the input coil current and is achieved by establishing electromagnetic-mechanical coupling equations.

[0070] In this step, the processed, time-synchronized target acoustic and current signals are first used as a set of related input data and prepared for analysis in a pre-built digital twin model. Next, the digital twin model is accessed, and based on the characteristics of the input data, the target acoustic signal is routed to the sub-model specifically responsible for simulating the mechanical vibration characteristics of the transformer (structural dynamics model), while the target current signal is routed to the sub-model specifically responsible for simulating the operating mechanical characteristics of the circuit breaker (operating mechanism model). Finally, through this step, the target acoustic signal, representing the actual vibration characteristics of the transformer, is assigned to the structural dynamics model capable of analyzing its internal mechanical state, while the target current signal, representing the actual operating current of the circuit breaker, is assigned to the operating mechanism model capable of analyzing its mechanism's operational state. This prepares the data for the next step of in-depth, targeted fault feature identification within the two specialized sub-models.

[0071] For example, following the specific implementation of the previous step, the obtained target acoustic signal, i.e., the 100-500Hz spectrum amplitude sequence after interpolation, and the target current signal, i.e., the current waveform data within 0.1 seconds during the closing operation of circuit breaker B, are first input into the digital twin model established for the high-voltage substation. Secondly, the digital twin model identifies that the target acoustic signal corresponds to the vibration characteristics of transformer A, and guides it to the transformer A structural dynamics model embedded in the model. At the same time, it identifies that the target current signal corresponds to the operating current of circuit breaker B, and guides it to the circuit breaker B operating mechanism model embedded in the model. Finally, the real-time data collected on site has been accurately allocated to the corresponding components in the virtual model for subsequent simulation analysis.

[0072] This step serves as a crucial bridge connecting data acquisition and processing with model simulation analysis, enabling the precise import of multi-source heterogeneous monitoring data collected synchronously on-site into the corresponding virtual simulation model. It ensures that characteristic data representing the transformer's state are processed by a structural dynamics model that reveals its mechanical fault mechanism, while characteristic data representing the circuit breaker's state are processed by a model that reflects the performance of its operating mechanism. This targeted data distribution mechanism lays a solid foundation for subsequent high-fidelity state deduction and accurate fault feature identification within their respective professional models, making in-depth analysis based on physical mechanisms possible.

[0073] Step 104: In the digital twin model, the acoustic signal is compared with the reference acoustic pattern pre-stored in the voiceprint database to identify the first abnormal feature corresponding to the transformer mechanical fault, and the current signal is compared with the theoretical current curve of the operating mechanism model to identify the second abnormal feature corresponding to the circuit breaker electrical fault.

[0074] Optionally, step 104 may specifically include: Step 1041: Input the target acoustic signal into the structural dynamics model in the digital twin model to obtain the multi-point simulated vibration response corresponding to the transformer core and windings.

[0075] Step 1042: Input the target current signal into the operating mechanism model in the digital twin model to obtain the simulated motion trajectory of the circuit breaker moving contact.

[0076] Step 1043: Retrieve the target reference acoustic mode that matches the current transformer model from the acoustic signature database. The reference acoustic mode contains the reference spectrum distribution of the core vibration under normal conditions.

[0077] Step 1044: The multi-point simulated vibration response is synthesized according to spatial location to obtain the overall vibration characteristic spectrum of the transformer tank, and the amplitude deviation between the overall vibration characteristic spectrum and the reference spectrum at the preset characteristic frequency point is calculated.

[0078] Step 1045: Compare the simulated motion trajectory with the theoretical motion trajectory pre-stored in the operating mechanism model to obtain the time delay of the moving contact during the opening and closing process.

[0079] Step 1046: Mark the feature frequency point corresponding to the amplitude deviation exceeding the first preset threshold as the first abnormal feature.

[0080] Step 1047: Mark the motion phase in which the time delay exceeds the second preset threshold as the second abnormal feature.

[0081] In this step, the acoustic signature database refers to a database that stores standard vibration characteristic data of transformers of various types and states during normal operation. It is used as a comparison benchmark to identify anomalies in the current equipment status. It is constructed by collecting a large amount of vibration data of transformers in a healthy state and extracting features.

[0082] The reference acoustic pattern refers to the vibration characteristic data that the device with the same model as the transformer being analyzed should have under normal and fault-free conditions, which is stored in the acoustic pattern database. It is used as a benchmark to determine whether the current device is in an abnormal state, and is obtained by querying and calling the acoustic pattern database according to the transformer model.

[0083] The first abnormal feature refers to the abnormal vibration characteristics of the transformer that are discovered through comparative analysis, indicating that the transformer may have mechanical faults. These are used to calculate the probability of the fault.

[0084] The theoretical current curve refers to the standard curve stored inside the circuit breaker operating mechanism model, which shows the change of coil current over time when the circuit breaker of this model is opened or closed under ideal fault-free conditions. It is used as a benchmark to judge whether the actual operating current is abnormal and is the inherent theoretical reference data of the operating mechanism model.

[0085] The second abnormal feature refers to the abnormal action timing indicators discovered through comparative analysis, which indicate that the circuit breaker operating mechanism may have a fault. These indicators are used to calculate the fault probability in subsequent calculations. The fault probability is obtained by identifying the time delay between the simulated motion trajectory and the theoretical motion trajectory at a specific stage when they exceed a threshold.

[0086] The transformer core and windings are the core components inside the transformer that realize the conversion of electromagnetic energy. Their mechanical state directly affects the vibration characteristics of the transformer and is the main object of simulation in the structural dynamics model.

[0087] Multi-point simulated vibration response refers to the vibration data at multiple key locations inside the transformer calculated by the structural dynamics model after receiving the target acoustic signal. This data reflects the internal state and is obtained through simulation calculations by running the structural dynamics model.

[0088] The moving contact of a circuit breaker refers to the movable conductive part in a circuit breaker that is directly responsible for connecting or disconnecting the circuit. Its motion state directly reflects the health of the operating mechanism and is the main object of simulation in the operating mechanism model.

[0089] The simulated motion trajectory refers to the data on the displacement of the circuit breaker moving contact during the opening and closing process calculated by the operating mechanism model after receiving the target current signal. It is used to reflect the action state of the mechanism and is obtained by simulation calculation through running the operating mechanism model.

[0090] The reference spectrum distribution refers to the standard intensity distribution data of the vibration energy of a normal transformer core at different frequency points contained in the reference acoustic mode. It is used for spectrum comparison and is the core content of the reference acoustic mode.

[0091] The overall vibration characteristic spectrum refers to the spectral data representing the vibration of the entire transformer tank surface, which is obtained by comprehensively considering the multi-point simulated vibration response data and the transformer tank structure. It is used to compare with the reference spectrum and is obtained by processing the multi-point simulated vibration response using a spatial synthesis algorithm.

[0092] Preset characteristic frequency points refer to several specific frequency points that are highly sensitive to specific mechanical faults in transformers and are used to focus on fault diagnosis. They are determined in advance based on knowledge of transformer fault mechanisms.

[0093] Amplitude deviation refers to the difference between the amplitude of the overall vibration characteristic spectrum and the amplitude of the reference spectrum distribution at the same preset characteristic frequency point. It is used to quantify the degree of anomaly and is obtained by calculating the amplitude difference between the two spectra at a specific frequency point.

[0094] The pre-stored theoretical motion trajectory refers to the displacement-time relationship curve that the moving contact of this type of circuit breaker should have under ideal conditions, which is stored in the operating mechanism model and is used as a comparison benchmark. It is the standard data built into the model.

[0095] The time delay refers to the lag time between the position of the moving contact in the simulated motion trajectory and the position of the moving contact in the pre-stored theoretical motion trajectory at the same moment during the opening and closing process. It is used to quantify the degree of sluggishness of the action and is obtained by comparing the difference between the two trajectories on the time axis.

[0096] The first preset threshold is a pre-set limit value used to determine whether the amplitude deviation constitutes a significant anomaly. If the amplitude deviation exceeds this value, the frequency point is considered abnormal. It is set based on historical data and experience.

[0097] The second preset threshold is a pre-set boundary value used to determine whether the time delay constitutes a significant anomaly. If the time delay exceeds this value, the motion phase is considered abnormal. It is set based on historical data and experience.

[0098] The operation phase refers to the key intervals into which the circuit breaker's opening or closing operation process is divided in chronological order, and is used to evaluate the operation performance in stages.

[0099] In this step, the digital twin model first receives the target acoustic signal and the target current signal, and inputs the target acoustic signal into its internal transformer structural dynamics model. This structural dynamics model, based on the transformer's physical parameters and dynamic equations, simulates the vibration of key components such as the transformer core and windings at different spatial locations under the excitation of the input signal, and outputs the simulated vibration response at these locations. Simultaneously, the digital twin model inputs the target current signal into its internal circuit breaker operating mechanism model. This operating mechanism model, based on the electromagnetic-mechanical coupling equations of the circuit breaker, simulates the movement process of the moving contact under the drive of the input current, and outputs the simulated motion trajectory. Secondly, the digital twin model queries and calls the corresponding reference acoustic mode from the acoustic signature database according to the specific model of the transformer being monitored. This reference acoustic mode contains the reference spectrum distribution of the core vibration of that model of transformer under normal conditions. Next, the digital twin model processes the obtained multi-point simulated vibration responses, employing algorithms such as spatial superposition and weighted averaging to combine these responses distributed at different points inside the transformer into a comprehensive vibration characteristic that represents the overall external performance of the entire transformer tank, i.e., the overall vibration characteristic spectrum. This overall vibration characteristic spectrum is then compared with a reference spectrum distribution retrieved from the acoustic signature database, specifically calculating the amplitude deviation at several pre-selected fault-sensitive characteristic frequency points. Simultaneously, the digital twin model compares the simulated motion trajectory calculated by the operating mechanism model with the theoretical motion representing the ideal state stored within that operating mechanism model. The trajectories are compared, and the time delay is obtained by calculating the time lag of the simulated trajectory relative to the theoretical trajectory at each key motion stage in the opening or closing process. Finally, the digital twin model compares the amplitude deviation at each characteristic frequency point with a first preset threshold, and marks the characteristic frequency points whose amplitude deviation exceeds the first preset threshold. The set of these points is the first abnormal feature. Similarly, the time delay of each motion stage is compared with a second preset threshold, and the motion stages whose time delay exceeds the second preset threshold are marked. The set of these stages is the second abnormal feature.

[0100] For example, following the specific implementation of the previous step, firstly, the digital twin model inputs the target acoustic signal into the structural dynamics model of transformer A and calculates the vibration data of the core and windings at three key points, i.e., the multi-point simulated vibration response; simultaneously, the target current signal is input into the operating mechanism model of circuit breaker B, and the displacement curve of the moving contact from the start of movement to the completion of closure is calculated, i.e., the simulated motion trajectory; secondly, a reference acoustic mode with the same model as transformer A is retrieved from the acoustic signature database to obtain its reference spectrum distribution; then, the simulated vibration responses of the three points are synthesized into a line representing the transformer... The overall vibration characteristic spectrum of the enclosure A is calculated, and the amplitude deviation of this overall vibration characteristic spectrum from the reference spectrum at several preset characteristic frequency points of 100Hz, 200Hz, and 400Hz is calculated, for example, assuming it is +3dB, +5dB, and +8dB respectively. Then, the simulated motion trajectory and the theoretical trajectory are compared, and it is found that the completion time of the moving contact in the closing phase is delayed by 3 milliseconds compared with the theoretical value. Finally, if the first preset threshold is 4dB, then 200Hz and 400Hz are marked as the first abnormal feature, and if the second preset threshold is 2 milliseconds, then the closing motion phase is marked as the second abnormal feature.

[0101] This step transforms the input signal into anomaly indicators with clear physical meaning by performing high-fidelity physical simulation and precise comparison with the baseline state within the digital twin model. This not only detects the existence of anomalies but also locates the location and stage of the anomaly through specific dimensions such as characteristic frequency points and motion phases. This provides precise and quantitative feature inputs for subsequent fault type determination, fault correlation analysis, and fault source location, greatly improving the accuracy and interpretability of fault diagnosis.

[0102] Step 105: Calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature.

[0103] Optionally, step 105 may specifically include: Step 1051: Obtain the set of feature frequency points corresponding to the first abnormal feature and the set of motion stages corresponding to the second abnormal feature.

[0104] Step 1052: Calculate the transformer core condition score based on the amplitude deviation corresponding to each characteristic frequency point in the set of characteristic frequency points.

[0105] Step 1053: Calculate the circuit breaker contact status score based on the time delay corresponding to each motion stage in the motion stage set.

[0106] Step 1054: Based on the transformer core condition score and the circuit breaker contact condition score, calculate the initial correlation probability value through a weighted fusion method.

[0107] Optionally, step 1054 may include the following steps: based on the connection relationship between the transformer and the circuit breaker in the power grid topology, determine the first weighting coefficient corresponding to the transformer core condition score and the second weighting coefficient corresponding to the circuit breaker contact condition score; multiply the transformer core condition score by the first weighting coefficient to obtain a first weighted score; multiply the circuit breaker contact condition score by the second weighting coefficient to obtain a second weighted score; sum the first weighted score and the second weighted score to obtain a comprehensive score value; and input the comprehensive score value into a preset probability transformation function to output an initial correlation probability value.

[0108] Step 1055: Obtain the electrical distance information between the transformer and the circuit breaker recorded in the digital twin model.

[0109] Step 1056: Correct the initial association probability value based on the electrical distance information to obtain the association fault probability value.

[0110] In this step, the associated fault probability value is a quantified numerical index used to represent the likelihood that the transformer and circuit breaker, as associated devices, will jointly cause a system fault due to the fault coupling effect. It is calculated by comprehensively evaluating the independent state scores of the two devices and taking into account the influence of their electrical distance.

[0111] The characteristic frequency point set refers to the group consisting of all characteristic frequency points marked as the first abnormal feature. It is used to comprehensively evaluate the overall abnormality of the transformer core and is obtained by collecting all characteristic frequency points whose amplitude deviation exceeds the first preset threshold.

[0112] The motion stage set refers to the group consisting of all motion stages marked as the second abnormal feature. It is used to comprehensively evaluate the overall abnormality of the circuit breaker contacts and is obtained by collecting all motion stages whose time delay exceeds the second preset threshold.

[0113] The transformer core condition score is a quantitative score that comprehensively reflects the mechanical condition of the transformer core. The higher the score, the more serious the abnormality. It is used to calculate the correlation probability.

[0114] The circuit breaker contact condition score is a quantitative score that comprehensively reflects the operational performance of the circuit breaker contacts. The higher the score, the more severe the abnormality, and it is used to calculate the correlation probability.

[0115] The initial correlation probability value refers to the preliminary correlation fault probability value calculated without considering the electrical distance between devices. It is obtained by weighting and fusing the transformer core condition score and the circuit breaker contact condition score, and then through a preset conversion relationship.

[0116] The first weighting coefficient refers to a proportional coefficient assigned to the transformer core condition score during weighted fusion. It is used to reflect the importance of the transformer condition to the associated faults and is determined based on the connection relationship and influence of the transformer and circuit breaker in the power grid topology.

[0117] The second weighting coefficient refers to a proportional coefficient assigned to the circuit breaker contact status score during weighted fusion. It is used to reflect the importance of the circuit breaker status to the associated faults and is determined based on the connection relationship and influence of the transformer and circuit breaker in the power grid topology.

[0118] The first weighted score refers to the result of multiplying the transformer core condition score by the first weighting coefficient, and is used in the comprehensive score calculation.

[0119] The second weighted score refers to the result of multiplying the circuit breaker contact condition score by the second weighting coefficient, and is used in the comprehensive score calculation.

[0120] The overall score is the sum of the first weighted score and the second weighted score, which is used to reflect the overall level of abnormality in the status of the two devices.

[0121] Electrical distance information refers to the quantitative data of electrical connection impedance or equivalent distance between a transformer and a circuit breaker recorded in a digital twin model, used to measure the ease with which faults propagate.

[0122] In this step, firstly, based on the results generated from the first and second abnormal features, all feature frequency points marked as abnormal are obtained to form a feature frequency point set, and simultaneously, all motion stages marked as abnormal are obtained to form a motion stage set. Secondly, the feature frequency point set is processed by first reading the amplitude deviation value corresponding to each feature frequency point in the feature frequency point set, and then using an algorithm, such as directly adding all these amplitude deviation values, or assigning different weights according to the importance of each frequency point before adding them, to finally calculate a single value to represent the overall abnormality level of the transformer core. This value is the transformer core condition score. Similarly, the motion stage set is processed by first reading the time delay value corresponding to each motion stage in the motion stage set, and then using the same algorithm, such as weighted summation, to calculate a single value to represent the overall abnormality level of the circuit breaker contacts. This value is the circuit breaker contact condition score. Next, weighted fusion is performed. Based on the closeness of the electrical connection between the transformer and the circuit breaker in the actual power grid, that is, their connection relationship in the power grid topology, two weighting coefficients are set: a first weighting coefficient representing the importance of the transformer's state and a second weighting coefficient representing the importance of the circuit breaker's state. Multiplication then begins, multiplying the transformer core state score by the first weighting coefficient to obtain the first weighted score, and simultaneously multiplying the circuit breaker contact state score by the second weighting coefficient to obtain the second weighted score. These two weighted scores are then added together to obtain a comprehensive score. This comprehensive score is then input into a pre-defined probability transformation function, which describes the mapping relationship between the comprehensive score and the fault probability. By querying or calculating this probability transformation function, an initial correlation probability value is output. Next, a key parameter, the electrical distance information between the transformer and the circuit breaker, is queried from the digital twin model. This electrical distance information reflects the electrical proximity of the two devices. Finally, based on the retrieved electrical distance information, the calculated initial correlation probability value is corrected. If the electrical distance is very close, the initial correlation probability value may be increased because the fault is more easily propagated directly. If the electrical distance is far, the initial correlation probability value may be reduced, and the final corrected value is the correlation failure probability value.

[0123] For example, following the specific implementation of the previous step, firstly, a set of characteristic frequency points is obtained, including two points at 200Hz and 400Hz; and the transformer core condition score is calculated based on the amplitude deviation of these points; secondly, a set of motion stages is obtained, including the closing stage, and the circuit breaker contact condition score is calculated based on the time delay of the closing stage; then, since transformer A and circuit breaker B are directly connected main equipment in the station, a higher weight coefficient is assigned to the transformer condition, and a relatively lower weight coefficient is assigned to the circuit breaker condition; then, a comprehensive score is obtained after weighted calculation, and an initial associated probability value is obtained according to the preset score-probability comparison relationship; subsequently, it is confirmed from the digital twin model that the electrical distance between transformer A and circuit breaker B is very close, belonging to the same electrical node; based on this close proximity relationship, the initial probability value is adjusted upward, and finally a higher associated fault probability value is generated.

[0124] This step integrates and quantifies the identified abnormal features scattered across different devices and dimensions into a unified, easily understood, and comparable associated fault probability value. It not only considers the severity of the anomaly of each device itself, but also reflects the relative importance of different devices in associated faults through weighting coefficients, and further incorporates consideration of the ease of fault propagation through electrical distance information. This comprehensive evaluation method enables the output results to more comprehensively and accurately reflect the overall risk level of associated devices, providing a scientific and quantitative basis for the final decision on whether to issue an early warning.

[0125] Step 106: When the associated fault probability value is greater than the preset probability threshold, output a fault warning command.

[0126] Optionally, step 106 may specifically include: Step 1061: When the associated fault probability value is greater than the preset probability threshold, determine the corresponding warning level according to the different numerical ranges in which the associated fault probability value is located.

[0127] Step 1062: Determine the potential fault area of ​​the transformer core based on the set of characteristic frequency points corresponding to the first abnormal feature.

[0128] Step 1063: Determine the potential fault type of the circuit breaker contacts based on the set of motion stages corresponding to the second abnormal feature.

[0129] Step 1064: Based on the electrical connection relationship between the transformer and the circuit breaker recorded in the digital twin model, generate fault impact range information.

[0130] Step 1065: Combine the information on the warning level, potential fault area, potential fault type, and fault impact range, and output a fault warning command.

[0131] In this step, the preset probability threshold refers to a pre-set probability threshold value used to determine whether the associated fault risk has reached the level that requires issuing an early warning. It is set by combining historical operating data and safety criteria.

[0132] Fault warning instructions refer to a complete set of instructions containing specific warning information and handling suggestions, used to guide operation and maintenance personnel to take targeted measures.

[0133] Different numerical ranges refer to multiple consecutive probability value ranges divided above a preset probability threshold, used to distinguish the severity of risks, and are determined according to operation and maintenance procedures and risk level classification standards.

[0134] The warning level refers to a level indicator that represents the urgency and risk level of a fault, and is used to determine the response priority and timeliness requirements.

[0135] Potential fault areas refer to the locations of components inside the transformer most likely to fail, inferred from abnormal characteristics. They are used to guide on-site inspection and location, and are determined by analyzing the correspondence between the characteristic frequency points corresponding to the first abnormal characteristic and the internal structure of the transformer.

[0136] Potential fault types refer to the nature of possible faults in the circuit breaker operating mechanism inferred from abnormal characteristics. They are used to guide the formulation of maintenance strategies and are determined by analyzing the correspondence between the movement stage corresponding to the second abnormal characteristic and typical fault modes.

[0137] Fault impact range information refers to the description of the range of electrical equipment that a predicted fault may affect. It is used to assess the consequences of an accident and to develop isolation plans. It is derived by analyzing the electrical connection relationship between transformers and circuit breakers in a digital twin model.

[0138] In this step, the calculated associated fault probability value is first compared with a preset probability threshold. When the associated fault probability value is greater than the preset probability threshold, an early warning generation process is triggered. Based on the magnitude of the associated fault probability value, it is categorized into one of the preset different numerical ranges to determine the corresponding early warning level. For example, the higher the probability value, the higher the corresponding early warning level, indicating a more urgent risk. Secondly, the set of characteristic frequency points corresponding to the first abnormal feature is analyzed. Based on the pre-established mapping knowledge between characteristic frequencies and the locations of internal transformer components, the internal transformer components most likely corresponding to these abnormal frequency points are inferred, thereby determining the potential fault area of ​​the transformer core. Simultaneously, the set of characteristic frequency points corresponding to the second abnormal feature is analyzed. The process involves first identifying the most likely causes of abnormal movement phases based on a pre-established correspondence between abnormal movement phases and typical circuit breaker fault types, thereby determining the potential fault types of the circuit breaker contacts. Next, it accesses the power grid topology information recorded in the digital twin model, analyzes the electrical connection relationship between the transformer and the circuit breaker, and, based on this relationship, deduces the potential impact on downstream or other related equipment if a fault occurs at this node, thus generating fault impact range information. Finally, the obtained warning level, potential fault area, potential fault type, and fault impact range information are combined and encapsulated according to a predetermined format specification to form a complete fault warning instruction and output it.

[0139] For example, following the specific implementation of the previous step, firstly, the calculated associated fault probability value is compared with a preset probability threshold, and it is found that the associated fault probability value has exceeded the preset threshold; secondly, based on the high-risk value range where the associated fault probability value is located, the warning level is determined to be high; then, the set of characteristic frequency points corresponding to the first abnormal feature is analyzed, and it is found that the abnormal frequency points mainly correspond to the clamping area on the transformer core, thus determining that the potential fault area is the clamping area on the core; then, the set of motion stages corresponding to the second abnormal feature is analyzed, and it is found that there is an abnormal delay at the end of the closing process, thus determining that the potential fault type is circuit breaker contact erosion; subsequently, the digital twin model is queried to confirm that transformer A and circuit breaker B are directly connected and are important outgoing lines in the station, thus generating fault impact range information, indicating that the fault may affect the load of the entire outgoing line; finally, the high warning level, the fault area of ​​the clamping area on the core, the contact erosion fault type, and the range information affecting the outgoing line load are combined to generate and output a detailed fault warning instruction.

[0140] This step transforms the quantified failure probability into specific and actionable early warning information. It not only provides an assessment of the risk level, but more importantly, by linking it with the results of previous analysis, it gives the potential location, type, and possible scope of impact of the failure. This makes the output early warning instructions highly targeted and instructive, greatly facilitating maintenance personnel to quickly understand the nature of the risk, locate the failure point, and formulate effective response strategies, thus realizing a closed loop from risk early warning to precise maintenance.

[0141] Figure 2 This application provides a schematic diagram of the structure of a power fault prediction system based on digital twins, as shown below. Figure 2 As shown, the system includes: The acquisition module 21 is used to acquire the first monitoring data and the second monitoring data; Processing module 22 is used to perform time synchronization processing on the first monitoring data and the second monitoring data to obtain the target acoustic signal and the target current signal; Input module 23 is used to input the target acoustic signal and target current signal into the digital twin model, wherein the digital twin model includes a structural dynamics model of the transformer and an operating mechanism model of the circuit breaker; The identification module 24 is used to compare the acoustic signal with the reference acoustic pattern pre-stored in the voiceprint database in the digital twin model to identify the first abnormal feature corresponding to the transformer mechanical fault, and to compare the current signal with the theoretical current curve of the operating mechanism model to identify the second abnormal feature corresponding to the circuit breaker electrical fault. Calculation module 25 is used to calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature; The output module 26 is used to output a fault warning command when the associated fault probability value is greater than a preset probability threshold.

[0142] Figure 2 The aforementioned power fault prediction system based on digital twins can perform... Figure 1 The implementation principle and technical effects of the power fault prediction method based on digital twins described in the illustrated embodiment will not be repeated here. The specific methods by which each module and unit of the power fault prediction system based on digital twins in the above embodiments perform operations have been described in detail in the embodiments related to this method, and will not be elaborated upon here.

[0143] In one possible design, Figure 2 The power fault prediction system based on digital twins in the illustrated embodiment can be implemented as a computing device, such as... Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32; The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are invoked and executed by the processing component 32.

[0144] The processing component 32 is used for the above Figure 1 The embodiment describes a power fault prediction method based on digital twins.

[0145] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above-described method. Alternatively, the processing component may be implemented as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described method.

[0146] Storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0147] Of course, computing devices may also include other components, such as input / output interfaces, display components, communication components, etc.

[0148] Input / output interfaces provide interfaces between processing components and peripheral interface modules, which can be output devices, input devices, etc.

[0149] The communication components are configured to facilitate wired or wireless communication between computing devices and other devices.

[0150] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server, and the aforementioned processing components, storage components, etc., can be basic server resources rented or purchased from the cloud computing platform.

[0151] This application also provides a computer storage medium storing a computer program, which, when executed by a computer, can perform the above-described functions. Figure 1 The embodiment shown is a power fault prediction method based on digital twin.

[0152] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0153] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0154] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A power fault prediction method based on digital twins, characterized in that, include: Acquire the first and second monitoring data; The first and second monitoring data are time-synchronized to obtain the target acoustic signal and the target current signal. The target acoustic signal and target current signal are input into a digital twin model, wherein the digital twin model includes a structural dynamics model of a transformer and an operating mechanism model of a circuit breaker; In the digital twin model, the acoustic signal is compared with the reference acoustic pattern pre-stored in the voiceprint database to identify the first abnormal feature corresponding to the transformer mechanical fault, and the current signal is compared with the theoretical current curve of the operating mechanism model to identify the second abnormal feature corresponding to the circuit breaker electrical fault. Calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature; When the associated fault probability value is greater than a preset probability threshold, a fault warning command is output.

2. The method according to claim 1, characterized in that, Acquire first and second monitoring data, including: The original vibration waveform data of the transformer is collected by multiple vibration sensors arranged on the wall of the transformer tank at a first preset sampling frequency. The original current waveform data of the circuit breaker is collected at a second preset sampling frequency by a current sensor connected in series in the circuit breaker's opening and closing coil circuit. The original vibration waveform data is transformed in the frequency domain to obtain vibration spectrum data, and the spectral amplitude within a preset frequency band is extracted from the vibration spectrum data. The initial raw current waveform data is segmented to divide the current rise phase data and current stabilization phase data during circuit breaker operation. The spectral amplitude within the preset frequency band is marked as the first monitoring data; The duration of the current rising phase data and the average current value of the current stabilizing phase data are jointly labeled as the second monitoring data.

3. The method according to claim 1, characterized in that, The first and second monitoring data are time-synchronized to obtain the target acoustic signal and the target current signal, including: Receive the first monitoring data and the second monitoring data, wherein both the first monitoring data and the second monitoring data are accompanied by clock information; Based on the clock information, the first time point in the first monitoring data where the spectrum amplitude suddenly changes is identified, and the second time point in the second monitoring data where the current rise phase begins is identified; Calculate the initial time difference between the first time point and the second time point; The initial time difference is compared with a preset time difference threshold. If the initial time difference is less than or equal to the time difference threshold, the first monitoring data and the second monitoring data are determined to be a synchronized data group. The spectral amplitude of the first monitoring data identified as a synchronous data group is interpolated to obtain the target acoustic signal. At the same time, the current waveform data corresponding to the duration of the current rise phase is marked as the target current signal.

4. The method according to claim 1, characterized in that, In the digital twin model, the acoustic signal is compared with a reference acoustic pattern pre-stored in the voiceprint database to identify a first abnormal feature corresponding to a transformer mechanical fault, and the current signal is compared with the theoretical current curve of the operating mechanism model to identify a second abnormal feature corresponding to a circuit breaker electrical fault, including: The target acoustic signal is input into the structural dynamics model in the digital twin model to obtain the multi-point simulated vibration response corresponding to the transformer core and windings; The target current signal is input into the operating mechanism model in the digital twin model to obtain the simulated motion trajectory of the circuit breaker moving contact; The target reference acoustic pattern matching the current transformer model is retrieved from the acoustic pattern database. The reference acoustic pattern contains the reference spectral distribution of core vibration under normal conditions. The multi-point simulated vibration response is synthesized according to spatial location to obtain the overall vibration characteristic spectrum of the transformer tank, and the amplitude deviation between the overall vibration characteristic spectrum and the reference spectrum at the preset characteristic frequency point is calculated. The simulated motion trajectory is compared with the theoretical motion trajectory pre-stored in the operating mechanism model to obtain the time delay of the moving contact during the opening and closing process. The characteristic frequency point corresponding to the amplitude deviation exceeding the first preset threshold is marked as the first abnormal feature; The motion phase in which the time delay exceeds the second preset threshold is marked as the second abnormal feature.

5. The method according to claim 1, characterized in that, Based on the first abnormal feature and the second abnormal feature, the associated fault probability value is calculated, including: Obtain the set of feature frequency points corresponding to the first abnormal feature and the set of motion stages corresponding to the second abnormal feature; The transformer core condition score is calculated based on the amplitude deviation corresponding to each characteristic frequency point in the set of characteristic frequency points. The circuit breaker contact status score is calculated based on the time delay corresponding to each motion stage in the set of motion stages. Based on the transformer core condition score and the circuit breaker contact condition score, an initial correlation probability value is calculated using a weighted fusion method. Obtain the electrical distance information between the transformer and the circuit breaker recorded in the digital twin model; The initial association probability value is corrected based on the electrical distance information to obtain the associated fault probability value.

6. The method according to claim 5, characterized in that, Based on the transformer core condition score and the circuit breaker contact condition score, an initial correlation probability value is calculated using a weighted fusion method, including: Based on the connection relationship between transformers and circuit breakers in the power grid topology, the first weighting coefficient corresponding to the transformer core condition score and the second weighting coefficient corresponding to the circuit breaker contact condition score are determined. The transformer core condition score is multiplied by the first weighting coefficient to obtain the first weighted score; The circuit breaker contact status score is multiplied by the second weighting coefficient to obtain the second weighted score; The first weighted score and the second weighted score are summed to obtain the comprehensive score. The comprehensive score value is input into a preset probability conversion function to output an initial correlation probability value.

7. The method according to claim 1, characterized in that, When the associated fault probability value is greater than a preset probability threshold, a fault warning instruction is output, including: When the associated fault probability value is greater than the preset probability threshold, the corresponding warning level is determined according to the different numerical ranges in which the associated fault probability value is located. Based on the set of characteristic frequency points corresponding to the first abnormal feature, the potential fault area of ​​the transformer core is determined. Based on the set of motion stages corresponding to the second abnormal feature, the potential fault type of the circuit breaker contacts is determined; Based on the electrical connection relationship between the transformer and the circuit breaker recorded in the digital twin model, information on the scope of the fault impact is generated; The warning level, potential fault area, potential fault type, and fault impact range information are combined to output a fault warning command.

8. A power fault prediction system based on digital twins, characterized in that, include: The acquisition module is used to acquire the first monitoring data and the second monitoring data; The processing module is used to perform time synchronization processing on the first monitoring data and the second monitoring data to obtain the target acoustic signal and the target current signal; An input module is used to input the target acoustic signal and the target current signal into a digital twin model, wherein the digital twin model includes a structural dynamics model of a transformer and an operating mechanism model of a circuit breaker; The identification module is used to compare the acoustic signal with a reference acoustic pattern pre-stored in the voiceprint database in the digital twin model to identify a first abnormal feature corresponding to a transformer mechanical fault, and to compare the current signal with the theoretical current curve of the operating mechanism model to identify a second abnormal feature corresponding to a circuit breaker electrical fault. The calculation module is used to calculate the associated fault probability value based on the first abnormal feature and the second abnormal feature; The output module is used to output a fault warning command when the associated fault probability value is greater than a preset probability threshold.

9. A computing device, characterized in that, It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are invoked and executed by the processing component to implement a power fault prediction method based on digital twins as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, The device contains a computer program that, when executed by a computer, implements a power fault prediction method based on digital twins as described in any one of claims 1 to 7.

Images