Circuit breaker action coil current characteristic diagnosis and analysis system and method

Abstract

The invention provides a circuit breaker action coil current characteristic diagnosis and analysis system and method. Belongs to the technical field of power system automation and substation state monitoring. The method comprises the following steps: carrying out synchronous wave recording on current of a plurality of circuit breaker actuating coils on a direct current feeder screen side to generate multi-channel synchronous wave recording data; meanwhile, SOE message information is collected, a circuit breaker with specific action is automatically matched and positioned by utilizing a timestamp and action identification information in an SOE message, and action circuit breaker positioning data is generated; the current waveform data of the circuit breaker actuating coil are collected in real time, and a high-precision algorithm is used for feature extraction and analysis, so that abnormal fluctuation in current change can be accurately captured, such as overlarge starting current and unstable holding current, the risk of misoperation or rejection of operation of the circuit breaker caused by abnormal current characteristics is reduced, and the service life of the circuit breaker is prolonged. Electric power system accidents caused by equipment faults are avoided, and the operation reliability of the circuit breaker is improved.

Description

TECHNICAL FIELD

[0001] The application provides a circuit breaker operating coil current characteristic diagnosis analysis system and method, and belongs to the technical field of power system automation and substation state monitoring. BACKGROUND

[0002] In power system operation, as a key device, the operating coil current characteristic of the circuit breaker directly reflects the health state of the device, and is crucial to guarantee stable operation of the power grid.

[0003] However, the existing circuit breaker mechanical characteristic testing products have many defects. In terms of installation, a Hall sensor or a vibration sensor is usually installed on the side of the switch body, which requires the installation of sensors and testing devices for each circuit breaker body in the high-voltage chamber, which is not only difficult to wire and requires a large amount of modification, but also leads to high cost. In terms of data, the recorded wave data is disconnected from the background SOE information, and the specific operating circuit breaker cannot be automatically and accurately positioned, forming a data island. In terms of analysis method, only the original waveform is provided, and there is no effective horizontal and vertical comparison, and it is difficult to quantify the health state of the circuit breaker. Moreover, the data needs to be uploaded to the background for analysis, and there is no intuitive display and on-site early warning, so the operation and maintenance response is seriously lagging. In addition, the manufacturers' publicity is mostly at the general level of "online monitoring" and "waveform analysis", and no one has accurately matched the DC screen remote recording wave with SOE, made horizontal and vertical comparisons, and integrated three-level early warning in depth. The essence is still a recording wave instrument with storage, which cannot meet the actual operation and maintenance needs. SUMMARY

[0004] The application provides a circuit breaker operating coil current characteristic diagnosis analysis system and method to solve the problems mentioned in the background.

[0005] The application provides a circuit breaker operating coil current characteristic diagnosis analysis method, which comprises the following steps:

[0006] S1: Synchronously recording the operating coil currents of a plurality of circuit breakers on the DC feeder screen side to generate multi-channel synchronous recording wave data; simultaneously collecting SOE message information, automatically matching and positioning the specific operating circuit breaker by using the time stamp and operation identification information in the SOE message, and generating operating circuit breaker positioning data;

[0007] S2: According to the operating circuit breaker positioning data, the multi-channel synchronous recording wave data is screened to extract the coil current recording wave data corresponding to the operating circuit breaker; the coil current recording wave data is preprocessed to generate preprocessed coil current waveform data;

[0008] S3: Based on the preprocessed coil current waveform data, horizontal waveform comparison is performed to generate horizontal waveform difference data; and vertical waveform comparison is performed to generate vertical waveform change data;

[0009] S4: According to the transverse waveform difference data and the longitudinal waveform change data, a key feature quantity is extracted, the key feature quantity is used, a preset health state evaluation model is combined, a health state of the action circuit breaker is quantitatively evaluated, and circuit breaker health state evaluation data is generated;

[0010] S5: According to the circuit breaker health state evaluation data, three-level early warning thresholds are set, different health state levels are corresponded respectively; the circuit breaker health state evaluation data is compared with the three-level early warning thresholds, risk early warning processing is performed, and three-level early warning data of the circuit breaker action coil current characteristics is generated.

[0011] The circuit breaker action coil current characteristic diagnosis and analysis system provided by the application comprises:

[0012] One or more processors;

[0013] A memory for storing one or more programs;

[0014] When the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of the above.

[0015] The application has the following advantages: by collecting current waveform data of the circuit breaker action coil in real time and using high-precision algorithms for feature extraction and analysis, abnormal fluctuations in current changes, such as excessive starting current and unstable maintenance current, can be accurately captured, the risk of circuit breaker misoperation or refusal caused by abnormal current characteristics is reduced, power system accidents caused by equipment failure are avoided, and the reliability of circuit breaker operation is improved; based on big data analysis technology, historical current data is deeply mined and modeled, potential failure trends of the circuit breaker action coil, such as insulation aging and coil burning, can be predicted in advance, early warning reports can be generated according to the failure risk level and maintenance suggestions can be attached, the equipment downtime caused by sudden failure is reduced, economic losses caused by unplanned maintenance are reduced, and the predictability and initiative of equipment maintenance are enhanced; by comparing current characteristic data of circuit breaker action coils of different types and specifications, circuit breaker selection and configuration schemes can be optimized, current performance of various equipment under specific working conditions can be clearly understood, and reasons for improper selection (such as load mismatch and poor environmental adaptability) can be analyzed, performance degradation problems caused by equipment selection errors are avoided, the cost of later modification and replacement is reduced, and the scientificity and economy of circuit breaker selection are improved; a standardized and automated current characteristic diagnosis process is adopted, combined with a remote monitoring and diagnosis platform, real-time and remote monitoring and analysis of the current characteristics of the circuit breaker action coil are realized, the workload and errors of on-site manual detection are reduced, inaccurate diagnosis results caused by human factors are avoided, the circuit breaker state monitoring and diagnosis work is more efficient and accurate, and the intelligent level of power system operation and maintenance is enhanced. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 for the method of the present application;

[0017] Figure 2 for the waveform and characteristic quantity of the present application;

[0018] Figure 3 for the control loop current waveform comparison diagram of the present application. DETAILED DESCRIPTION

[0019] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings, it should be understood that the preferred embodiments described herein are only used to explain and illustrate the present application, and are not used to limit the present application.

[0020] One embodiment of the present application, as shown in a circuit breaker operating coil current characteristic diagnostic analysis method, the method comprises: Figure 1

[0021] S1: Synchronously record the coil current of a plurality of circuit breakers on the DC feeder screen side to generate multi-channel synchronous recording data; at the same time, collect SOE (Sequence of Events) message information, use the time stamp and action identification information in the SOE message to automatically match and locate the specific action circuit breaker, and generate action circuit breaker positioning data;

[0022] S2: According to the action circuit breaker positioning data, filter the multi-channel synchronous recording data, and extract the coil current recording data corresponding to the action circuit breaker; pre-process the coil current recording data, including filtering, denoising and other operations, to generate pre-processed coil current waveform data;

[0023] S3: Based on the pre-processed coil current waveform data, perform horizontal waveform comparison, that is, compare and analyze the coil current waveform of the current action circuit breaker with the coil current waveform of other normal circuit breakers of the same type to generate horizontal waveform difference data; at the same time, perform vertical waveform comparison, that is, compare and analyze the coil current waveform of the current action circuit breaker with the coil current waveform of the action of the circuit breaker to generate vertical waveform change data;

[0024] S4: According to the horizontal waveform difference data and the vertical waveform change data, extract key characteristic quantities such as current peak value, current rise rate, action time, etc.; use these key characteristic quantities, combine with a pre-set health state evaluation model, quantitatively evaluate the health state of the action circuit breaker, and generate circuit breaker health state evaluation data;

[0025] ​S5: According to the circuit breaker health state evaluation data, set three levels of early warning threshold corresponding to different health state levels; compare the circuit breaker health state evaluation data with the three levels of early warning threshold, perform risk early warning processing, and generate three levels of early warning data of the circuit breaker operating coil current characteristics; at the same time, intuitively display the early warning data locally to realize on-site early warning and provide timely response basis for operation and maintenance personnel.

[0026] The working principle and effect of the above technical solution are:

[0027] Through centralized deployment on the DC feeder screen side and 16-way synchronous recording wave design, there is no need to install sensors and monitoring equipment on each circuit breaker at the high-voltage side, which reduces the complexity of field wiring and installation cost, avoids the maintenance burden of scattered deployment of multiple devices, and improves the monitoring deployment efficiency in the multi-circuit breaker scene; through the IRIG-B code time synchronization technology and SOE message automatic matching function, the recording wave data and the action circuit breaker information are accurately associated, the positioning error caused by time deviation is reduced, the additional wiring work of external auxiliary contact is avoided, and the accuracy and reliability of the circuit breaker action positioning are enhanced; through horizontal and vertical waveform comparison, characteristic parameter quantitative analysis and local automatic diagnosis mechanism, there is no need for manual analysis of original waveforms one by one, which reduces the workload and misjudgment risk of manual analysis, avoids the limitations of single analysis method, and improves the accuracy of circuit breaker health state diagnosis; through local three-level early warning, 10-inch touch screen intuitive display and WEB remote access function, abnormal information is pushed in real time with potential fault prompt, which reduces the probability of fault delay disposal, avoids the dependence of operation and maintenance personnel on on-site duty, enhances the timeliness of fault response and operation flexibility; through the whole-process automatic data acquisition, analysis, early warning and maintenance auxiliary closed loop, accurate decision basis is provided for operation and maintenance, resource waste caused by blind maintenance is reduced, power grid fluctuation caused by circuit breaker fault expansion is avoided, and the stability and safety redundancy of power system power supply are improved. Among them, as shown in Figure 2 According to the research of the research institute, through recording the 10kV circuit breaker opening and closing coil current waveform curve, the working state of the corresponding circuit breaker can be analyzed by the data of several key points; Figure 2 and Figure 3 is the waveform curve of the circuit breaker opening and closing coil operating current, which is usually composed of two wave peaks and one wave trough. According to the time position of the wave peak and wave trough, the waveform can be divided into five stages, representing different movement processes in the opening and closing process. Through comparison and analysis with historical data, potential defects can be found and located, which can intuitively and accurately reflect whether the operation state of each part of the circuit breaker is good or not; as can be seen from the figure, the whole process can be divided into five stages according to the movement of the iron core:

[0028] Phase 1: t = t0 ~ t1; t0 is the moment the high-voltage circuit breaker's opening and closing command is issued. The coil is energized at t0, and the current rises rapidly during the time interval t0 to t1. t1 is the moment the iron core begins to move, when the magnetic flux within the coil rises sufficiently to drive the iron core. The duration of Phase 1 is related to the voltage of the control power supply and the coil resistance, and can reflect the state of the coil. Its characteristic is that the current rises, but the iron core has not yet moved.

[0029] Phase 2: t = t1 ~ t2. At time t1, the iron core, under the influence of electromagnetic force, overcomes the resistance of gravity, spring force, and other forces, and begins to accelerate. The current begins to decrease rapidly until the iron core stops moving; this corresponds to time t2, which indicates that the iron core has triggered the load of the operating mechanism. Phase 2 reflects the motion state of the iron core and indicates whether there are any faults such as jamming or tripping during its movement.

[0030] Stage 3: t = t2 ~ t3. When the iron core hits the closing latch device, locks the door or valve, the iron core stops moving or bounces briefly, the current begins to increase, causing the opening spring to open the circuit.

[0031] Phase 4: t = t3 ~ t4. This phase is a continuation of the previous phase, with the current maintaining a slow increase or stabilization, and the switching process continuing.

[0032] Phase 5: t = t4 ~ t5. This is the current interruption phase, where the current value rapidly drops to zero. During this phase, the auxiliary switch disconnects, generating and elongating an arc between the auxiliary switch contacts. The arc voltage rapidly increases, forcing the coil current to decrease rapidly, the contacts fully disconnect, and the equivalent resistance between the contacts rapidly increases until the arc is completely extinguished.

[0033] The current has two peak points and one trough point. The t0~t1 stage reflects the working state of the coil, and this stage can be used to detect whether the coil resistance is normal. The t1~t2 stage reflects the working state of the iron core, such as the tripping and energy release mechanical load changes, and whether there is any jamming in the structure. The t2~t4 stage is the process of the transmission structure driving the contacts to complete the closing or opening operation. Taking t0 as the zero point of time, eight characteristic parameters can be selected: t1, t2, t3, t4, t5, I1, I2, and I3 to analyze the quality of the power supply voltage of the circuit breaker's operating mechanism, the idle stroke of the iron core, the frictional resistance of the mechanism, and whether there is any jamming, etc. Moreover, the coil current of the circuit breaker has a fixed characteristic curve. The trend of the curve represents the current change trend during the opening and closing process of the circuit breaker. Generally, the opening and closing curves can almost coincide. If there are different faults, this curve trend will not coincide. Figure 3 The three curves in the diagram represent the coil current waveforms during three operation of the same switch, and they almost overlap.

[0034] In one embodiment of the present invention, S1 includes:

[0035] S11. Install a multi-channel waveform recording terminal on the DC feeder panel side of the substation, configure 16 synchronous acquisition channels, set the sampling frequency to 4-10KHZ (adapt to the circuit breaker's action time of tens of milliseconds), and enable the IRIG-B code time synchronization function to ensure the consistency of data timestamps across all channels.

[0036] S12. The waveform recording terminal monitors the total current of the DC feeder circuit in real time and adopts a sudden change triggering mechanism (based on the characteristics of load current superimposed on operating current). When the current sudden change reaches the preset threshold, it automatically starts multi-channel synchronous waveform recording, collects the current data of multiple circuit breaker operating coils, and generates a multi-channel synchronous waveform recording raw dataset.

[0037] S13. Synchronously collect SOE message information sent by the substation backend system, extract key fields in the message, including action timestamp, opening and closing action identifier, and circuit number, to form a set of key information of SOE message;

[0038] S14. Using the IRIG-B code time synchronization result of the waveform recording terminal, the time axis of the original dataset of multi-channel synchronous waveform recording and the key information set of SOE message are calibrated to eliminate time deviation and ensure that the time dimension of the two is completely aligned.

[0039] S15. Based on the calibrated timestamp and action identification information, establish a correlation mapping model between the waveform recording data and SOE messages, automatically identify the circuit and number of the circuit breaker that triggered the waveform recording, and generate accurate positioning data of the circuit breaker.

[0040] The working principle and effects of the above technical solution are as follows:

[0041] The 16-channel synchronous acquisition design eliminates the need for independent acquisition equipment on each circuit breaker, reducing on-site installation and wiring workload, avoiding the cost pressure of distributed deployment of multiple devices, and improving the acquisition deployment efficiency in multi-circuit breaker scenarios. With a 4-10kHz adaptive sampling frequency, it accurately captures the current waveform details of circuit breaker operation over tens of milliseconds, reducing waveform data distortion, avoiding feature loss due to sampling frequency mismatch, and enhancing the integrity of the raw data. IRIG-B code time synchronization and time axis calibration eliminate time deviation between the recorded waveform data and SOE messages, reducing error sources in subsequent correlation analysis, avoiding positioning errors caused by time asynchrony, and improving data time consistency. The sudden change trigger mechanism activates waveform recording only when the current is abnormal, reducing the storage of invalid data, avoiding resource waste caused by continuous waveform recording, and enhancing the targeting of data acquisition.

[0042] By extracting key fields from SOE messages and associating them with a mapping model, the circuit breaker can be automatically located, reducing the tedious manual matching process, avoiding the need for additional wiring of external auxiliary contacts, and improving the accuracy and convenience of circuit breaker location.

[0043] In one embodiment of the present invention, step S12 includes:

[0044] Based on the rated load current of the DC feeder circuit and the current superposition characteristics when the circuit breaker operates, a sudden current start threshold (usually set to 1.2-1.5 times I0) and a trigger delay time (5-10μs to avoid false triggering due to instantaneous interference) are preset. At the same time, the synchronous start logic of 16 acquisition channels is associated to generate a sudden current trigger parameter configuration table.

[0045] The waveform recording terminal continuously and at high frequency samples the total current of the DC feeder circuit according to the preset sampling frequency of 4-10KHZ, and records the current value once every 250-100μs (corresponding to the 4-10KHZ frequency) to form a real-time current sampling sequence.

[0046] The difference between adjacent sampling points in the real-time current sampling sequence is calculated to obtain the current change rate. If the current change rate continuously exceeds the threshold in the sudden change trigger parameter configuration table and the duration reaches the trigger delay time, it is determined that the sudden change trigger condition is met, and a trigger start command is generated.

[0047] After receiving the trigger start command, the waveform recording terminal synchronously activates 16 acquisition channels to ensure that all channels acquire the current of multiple circuit breaker operating coils in parallel under the same timestamp (based on IRIG-B code time synchronization), forming a multi-channel parallel current data stream;

[0048] The multi-channel parallel current data streams are classified by channel number, and a precise timestamp (accurate to the microsecond level) is added to each data stream. Two to three unstable sampling points at the beginning of the acquisition are removed, and the data are integrated to form a multi-channel synchronous waveform recording raw dataset.

[0049] The working principle and effects of the above technical solution are as follows:

[0050] By presetting a sudden change threshold (1.2-1.5 times I0) adapted to the characteristics of current superposition and designing a 5-10μs trigger delay, false triggering caused by instantaneous interference is reduced, invalid waveform recordings are avoided from occupying storage resources, and the accuracy of trigger judgment is enhanced. Through 4-10kHz high-frequency continuous sampling, current values ​​are recorded every 250-100μs, accurately capturing the current waveform details of the circuit breaker's operation over tens of milliseconds, reducing waveform distortion and loss of key features, and improving the integrity of the original sampled data. Through the synchronous activation logic of 16 acquisition channels, combined with the unified timestamp guaranteed by IRIG-B code time synchronization, the sampling time difference between channels is reduced, avoiding the impact of data asynchrony on subsequent correlation analysis, and enhancing the consistency of multi-channel data. By classifying and organizing the data stream by channel, adding microsecond-level timestamps, and removing unstable sampling points at the beginning of the acquisition, invalid data interference is reduced, and the subsequent processing caused by data clutter is avoided, improving the regularity and reliability of the original dataset of multi-channel synchronous waveform recording.

[0051] In one embodiment of the present invention, step S14 includes:

[0052] Read the IRIG-B code time synchronization signal output in real time from the waveform recording terminal, analyze the second pulse, minute pulse and time code information in the signal, and generate a standard timestamp sequence of IRIG-B code with microsecond-level precision as a unified reference for time calibration.

[0053] Retrieve the original timestamps of the multi-channel synchronous waveform recording dataset (generated locally on the terminal during acquisition), compare them point by point with the IRIG-B code standard timestamp sequence, calculate the time deviation value of each waveform recording data (such as the deviation caused by the local clock drift of the terminal), perform linear compensation correction on the original timestamps based on the deviation value, and generate a calibrated multi-channel waveform recording dataset.

[0054] Analyze the action timestamps in the key information set of SOE messages, combine the network transmission delay between the substation backend and the waveform recording terminal (preset delay compensation model, optimized based on historical transmission data), align the SOE timestamps with the IRIG-B code standard timestamp sequence, correct the time deviation caused by transmission delay, and generate a calibrated SOE message information set.

[0055] Select the overlapping time interval between the calibrated multi-channel waveform recording dataset and the calibrated SOE message information set (e.g., 500ms before and after waveform recording trigger), randomly select 20-30 time nodes, calculate the absolute deviation of the timestamps of the two, and if the maximum deviation is ≤5μs, the time alignment is deemed qualified and a time calibration accuracy verification report is generated.

[0056] Based on the time calibration accuracy verification report, the calibrated multi-channel waveform recording dataset and the calibrated SOE message information set are associated and marked according to a unified IRIG-B code time axis to ensure that the time dimension of each waveform recording data is completely matched with the corresponding SOE information, and a time-calibrated associated dataset is generated.

[0057] The working principle and effects of the above technical solution are as follows:

[0058] By parsing IRIG-B codes to generate microsecond-level standard timestamp sequences, a unified time calibration benchmark is established, reducing time errors caused by local clock drift and avoiding calibration chaos due to the lack of a unified benchmark, thus improving the basic accuracy of time calibration. Linear compensation correction of the original timestamps in the waveform recording data specifically eliminates local clock deviations at the terminal, reducing time distortion in the waveform recording data, preventing the cumulative effect of deviations on subsequent analysis, and enhancing the time consistency of multi-channel waveform recording data. SOE message transmission delay compensation and standard timestamp alignment correct for time differences caused by network transmission, reducing time misalignment between waveform recordings and SOE messages, avoiding action positioning errors caused by time mismatches, and improving the accuracy of their correlation. Accuracy verification by randomly selecting 20-30 time nodes ensures a maximum deviation ≤5μs, reducing the influx of unqualified calibration data into subsequent stages, avoiding the risk of blindly accepting calibration results, and enhancing the reliability of time calibration. By associating and marking two types of datasets along a unified time axis, complete time dimension matching is achieved, reducing the tedious manual data association operation, avoiding diagnostic biases caused by data association errors, and providing solid support for accurate positioning of subsequent circuit breaker actions.

[0059] In one embodiment of the present invention, S2 includes:

[0060] S21. Based on the precise positioning data of the circuit breaker, select the coil current recording data of the corresponding channel of the circuit breaker from the original multi-channel synchronous recording dataset, remove other irrelevant channel data, and generate a single-target circuit breaker recording dataset.

[0061] S22. Adaptive Kalman filtering algorithm is used to filter the waveform data of a single target circuit breaker to remove high-frequency interference signals such as power grid harmonics and electromagnetic radiation, while retaining the true trend of coil current change.

[0062] S23. Using wavelet threshold denoising method, noise is suppressed on the filtered dataset, eliminating random noise generated during sampling, repairing minor distortions during data acquisition, and generating a denoised dataset.

[0063] S24. Convert the noise-reduced dataset into the COMTRADE99 format commonly used in the power industry, unify the parameter standards, including data range and time unit, and generate a standardized coil current dataset.

[0064] S25. Perform integrity verification on the standardized coil current dataset to detect issues such as missing data or broken waveforms. Perform interpolation repair on incomplete data to generate preprocessed coil current waveform data.

[0065] The working principle and effects of the above technical solution are as follows:

[0066] By filtering target channel data according to positioning data and removing irrelevant information, redundant data interference to subsequent processing is reduced, avoiding low analysis efficiency caused by mixed data and improving the targeting and convenience of data processing. Adaptive Kalman filtering removes high-frequency interference such as power grid harmonics and electromagnetic radiation, preserving the true trend of current changes, reducing signal distortion caused by the external electromagnetic environment, avoiding misleading diagnostic judgments by interference signals, and enhancing the authenticity of the data. Wavelet threshold denoising suppresses random noise and repairs minor distortions, further purifying data quality, reducing noise interference during sampling, preventing subtle features from being masked by noise, and improving the purity of current waveform data. Converting to the COMTRADE99 universal format and unifying data range and time units reduces processing obstacles caused by format incompatibility, avoids parameter standard confusion affecting cross-system analysis, and enhances the universality and interoperability of the data. Integrity verification and interpolation repair fill in data gaps and repair waveform breaks, reducing the impact of incomplete data on diagnostic results, avoiding analytical bias caused by data defects, and improving the reliability and usability of preprocessed waveform data.

[0067] In one embodiment of the present invention, step S3 includes:

[0068] S31. Collect coil current waveform data of normal circuit breakers of the same model, the same service life and no fault records, and extract their typical characteristic parameters. The typical characteristics include peak value, operating time and rising edge slope. Construct a normal waveform benchmark database of circuit breakers of the same type.

[0069] S32. Align the preprocessed coil current waveform data with the waveforms in the normal waveform reference database using feature points (including key nodes such as the start point of the rising edge, the peak point, and the end point of the falling edge) to ensure consistency in the comparison dimensions.

[0070] S33. Using a dynamic time warping algorithm, calculate the differences between the current waveform and the reference waveform in terms of characteristic parameters and waveform morphology, including peak deviation rate, motion duration deviation rate, and waveform similarity coefficient, and generate a horizontal waveform difference dataset.

[0071] S34. Extract the coil current waveform data of all past effective operations of the circuit breaker, organize them in the order of operation time, construct the circuit breaker's own historical waveform database, and mark key historical waveforms such as the first operation and the most recent normal operation.

[0072] S35. Compare the current waveform with the first action waveform and the average waveform of the three most recent normal actions in the historical waveform database, calculate the time-series change amount and change rate of each characteristic parameter, and generate a longitudinal waveform change trend dataset.

[0073] S36. Integrate the horizontal waveform difference dataset with the vertical waveform change trend dataset to form a complete two-dimensional waveform comparison analysis result set, and clarify the core dimensions of the difference and the key trends of change.

[0074] The working principle and effects of the above technical solution are as follows:

[0075] By constructing a database of normal waveform benchmarks of the same model and operating years, core comparison standards such as peak value and duration of motion are clearly defined, reducing comparison deviations caused by the lack of a unified benchmark, avoiding misjudgments caused by inconsistent benchmarks, and improving the reliability of horizontal comparisons. By aligning key nodes (rising edge start point, peak point, etc.) with feature points, the comparison dimensions are ensured to be consistent, reducing analysis errors caused by waveform misalignment, avoiding the impact of dimensional confusion on difference judgment, and enhancing the accuracy of waveform comparison. By using a dynamic time warping algorithm to calculate feature parameter differences and waveform similarity, subtle morphological differences are accurately captured, reducing the limitations of traditional comparison methods. This approach avoids biased analysis due to missing features and improves the accuracy of identifying horizontal differences. By constructing a dedicated historical waveform database for circuit breakers and marking key historical waveforms, it focuses on their own temporal changes, reducing the blindness of vertical analysis, avoiding trend misjudgments due to lack of historical references, and enhancing the perception of changes in the equipment's own status. By integrating horizontal and vertical comparison results, it clarifies the core dimensions of differences and key trends of change, reducing the bias of single-dimensional analysis, avoiding the problem that isolated data cannot support diagnosis, improving the comprehensiveness and persuasiveness of waveform analysis results, and providing a solid basis for subsequent health status assessment.

[0076] In one embodiment of the present invention, S36 includes:

[0077] S361. Extract the feature parameters from the horizontal waveform difference dataset (including peak deviation rate, action duration deviation rate, and waveform similarity coefficient) and the vertical waveform change trend dataset (including the time-series change amount and change rate of feature parameters), unify the parameter units (e.g., the deviation rate is kept to 2 decimal places, and the change rate is in % / action) and data format, and generate a normalized two-dimensional feature dataset.

[0078] S362. Based on the historical fault case library of circuit breakers, analyze the contribution of horizontal differences (comparison with the same model) and vertical changes (comparison with its own time series) to fault diagnosis. Use the analytic hierarchy process (AHP) to determine the fusion weight of each feature parameter (such as peak deviation rate weight 0.3 and time series change rate weight 0.35) to generate a two-dimensional feature fusion weight matrix.

[0079] S363. Using the normalized two-dimensional feature dataset as input, and combining the fusion weight matrix, perform weighted summation on the same feature dimensions (such as the horizontal peak deviation rate and the vertical peak change), and perform feature splicing on different feature dimensions (such as the waveform similarity coefficient and the rate of change of action duration) to generate a preliminary two-dimensional fusion dataset.

[0080] S364. Establish verification rules (e.g., horizontal peak deviation rate ≤ 5% and vertical peak change rate ≤ 8% / time are judged as logically consistent), mark contradictory data (e.g. horizontal is normal but vertical is abnormal) in the preliminary two-dimensional fusion dataset, and correct outliers by combining the recent operating conditions of the circuit breaker (e.g., ambient temperature and humidity, number of operations), and generate the verified fusion dataset.

[0081] S365. From the fused dataset after verification, select the top 3 feature parameters with the highest weight (such as peak deviation rate, rate of change of action duration, and waveform similarity coefficient) as the core dimensions of difference, extract trends of more than 3 consecutive changes in the time series (such as the action duration increasing one after another) as the key trends of change, and integrate them to form a complete two-dimensional waveform comparison analysis result set.

[0082] The working principle and effects of the above technical solution are as follows:

[0083] By unifying the units and data formats of feature parameters, a normalized two-dimensional feature dataset is generated, reducing the fusion chaos caused by the mixing of data from different formats, avoiding calculation errors caused by inconsistent units, and improving the compatibility and accuracy of data fusion. By using the analytic hierarchy process (AHP) combined with historical failure cases to determine fusion weights, the contribution of horizontal and vertical analysis is quantified, reducing the bias of weight allocation dominated by subjective experience, avoiding the impact of unreasonable weight settings on the fusion effect, and enhancing the scientific rigor and relevance of data fusion. Through a fusion method of weighted summation of features within the same dimension and splicing features from different dimensions, core information in both horizontal and vertical dimensions is comprehensively integrated, reducing the need for single-dimensional fusion. By addressing the limitations of logic, key features were not overlooked, thus improving the comprehensiveness and completeness of the fused data. By marking contradictory data with verification rules and correcting outliers based on operating conditions, the impact of data conflicts and distortions was eliminated, reducing analytical bias caused by contradictory data, avoiding the risk of blindly accepting fusion results, and enhancing the reliability of the fused dataset. By selecting the top three core features by weight and continuous time-series trends, key diagnostic information was focused on, reducing the interference of redundant data on subsequent analysis, avoiding the blurring of key points caused by information overload, improving the efficiency and accuracy of fault diagnosis, and providing high-quality data support for circuit breaker health assessment.

[0084] In one embodiment of the present invention, S363 includes:

[0085] Traverse the normalized two-dimensional feature dataset, identify and extract the same feature dimension parameter pairs corresponding to the horizontal and vertical directions (such as the horizontal peak deviation rate and the vertical peak change, the horizontal action duration deviation rate and the vertical action duration change), group them according to feature categories (peak category, duration category), and generate a set of the same feature dimension parameter pairs.

[0086] For each set of parameters in the same feature dimension parameter pair set, retrieve the weight value of the corresponding parameter in the fusion weight matrix (e.g., horizontal peak deviation rate weight 0.3, vertical peak change weight 0.35), calculate the weighted sum of each set of parameters according to the formula (horizontal parameter value × horizontal weight) + (vertical parameter value × vertical weight), and generate a weighted fusion result set with the same feature dimension.

[0087] The normalized two-dimensional feature dataset is traversed again to filter out independent feature parameters that have no horizontal or vertical correspondence (such as the waveform similarity coefficient unique to the horizontal and the rate of change of action duration unique to the vertical). These parameters are then classified and organized according to horizontal and vertical independent parameters to generate independent parameter sets for different feature dimensions.

[0088] The weighted fusion result set with the same feature dimension and the independent parameter set with different feature dimensions are concatenated according to the fixed feature order of peak class, duration class, similarity class and rate class, and the data field format (including parameter name, weighted value / original value, weight identifier) ​​is unified to generate a preliminary two-dimensional fusion dataset.

[0089] The working principle and effects of the above technical solution are as follows:

[0090] By identifying and extracting pairs of parameters with the same feature dimensions and grouping them by category, the fusion interference caused by the mixing of different types of features is reduced, and the calculation error caused by the confusion of feature classification is avoided, thus improving the targeting and accuracy of data fusion with the same dimension. By retrieving the weight matrix and calculating the weighted sum according to the formula, the contribution of the same features in the horizontal and vertical dimensions is quantified and integrated, reducing the fusion bias dominated by subjective experience, avoiding the one-sidedness of single-dimensional data, and enhancing the rationality of the fusion results with the same feature dimensions. By filtering and classifying independent feature parameters without corresponding relationships, the unique key information in the horizontal and vertical dimensions is completely preserved, reducing the information loss caused by feature omission, avoiding the incompleteness of the fused data, and improving the comprehensiveness of the fused dataset. By splicing the two datasets with fixed feature order and unifying the field format, the data organization is standardized, reducing the subsequent processing obstacles caused by data disorder, avoiding the low analysis efficiency caused by inconsistent formats, enhancing the regularity and usability of the initial fused dataset, and laying a good foundation for subsequent data verification and core feature extraction.

[0091] In one embodiment of the present invention, step S4 includes:

[0092] S41. From the set of dual-dimensional waveform comparison analysis results, key feature quantities are automatically extracted, including peak current, current rise rate, current fall rate, total action duration, peak occurrence time, and duration of each stage (pull-in stage, holding stage), forming a multi-dimensional feature parameter set.

[0093] S42. Based on machine learning algorithms (such as random forest algorithm), combined with the power industry circuit breaker fault diagnosis standards, input historical fault case data and normal operation data to train the model, construct a quantitative assessment model of circuit breaker health status, and set the feature parameter weight coefficients.

[0094] S43. Normalize the multi-dimensional feature parameter set to eliminate the range difference between different parameters, ensure that the weight of each feature in the evaluation model is reasonable, and generate a standardized feature parameter set.

[0095] S44. Input the standardized feature parameter set into the health status quantitative assessment model, and calculate the circuit breaker health status score (score range 0-100 points) through the model. The higher the score, the better the health status.

[0096] S45. Based on the health status score results and combined with the preset level classification rules (e.g., 85 points and above is excellent status, 70-84 points is good status, 50-69 points is average status, and below 50 points is risk status), generate circuit breaker health status assessment data.

[0097] The working principle and effects of the above technical solution are as follows:

[0098] By automatically extracting key features from multiple dimensions such as peak current, rate of rise, and duration of each stage, the system covers all stages of coil operation, reducing the bias of single-feature evaluation, avoiding omissions of key status information, and improving the comprehensiveness of health assessment. Through machine learning algorithms combined with power industry diagnostic standards, the model is trained using historical fault and normal data, quantifying feature parameter weights, reducing biases from subjective experience judgments, avoiding inconsistencies in assessment standards, and enhancing the scientific rigor of health assessment. By normalizing multi-dimensional feature parameters, the system eliminates differences in parameter ranges, ensuring reasonable weight proportions for each feature, reducing model misjudgments caused by parameter range interference, avoiding imbalances in feature importance, and improving the effectiveness of model input data. A quantitative score (0-100 points) intuitively presents the health status, replacing fuzzy qualitative judgments, reducing ambiguity in status definition, avoiding subjectivity in manual judgment, and enhancing the identifiability of health status. By generating assessment data through preset level classification rules, clearly defining excellent, good, average, and risk levels, the system reduces blindness in operation and maintenance decisions, avoids misjudgments and mishandling of equipment status, and improves the targeting of subsequent early warning and maintenance work.

[0099] In one embodiment of the present invention, S42 includes:

[0100] Collect historical fault case data of circuit breakers (including waveform characteristics and diagnostic results of fault types such as coil short circuit and core jamming) and normal operation data (feature parameters without fault records), remove abnormal samples with a data missing rate of >5%, label the samples according to four health levels: high quality / good / average / risk, and generate a labeled model training dataset.

[0101] Based on the size of the training dataset (e.g., sample size > 1000), configure the core parameters of the random forest, set the number of decision trees to 100-200 (to balance model accuracy and computational efficiency), set the maximum tree depth to 8-12 layers (to avoid overfitting), use the Gini coefficient as the feature selection criterion, and generate a random forest algorithm parameter configuration table.

[0102] The labeled model training dataset is divided into training and validation sets in a 7:3 ratio. The training set is used as input, and the model training is started by combining the random forest algorithm parameter configuration table. The model learns the feature parameter mapping relationship corresponding to different health levels and generates an initial assessment model of the circuit breaker health status.

[0103] Retrieve power industry standards such as "DL / T1573-2016 Guidelines for Condition Evaluation of High Voltage Circuit Breakers" to determine the priority of features such as peak current deviation rate and operating time variation for fault diagnosis (e.g., peak deviation has a higher weight than operating time deviation in coil short-circuit faults). Based on the feature importance output by the initial model, adjust the weight coefficients to meet the requirements of industry standards and generate a feature parameter weight coefficient matrix.

[0104] Input the validation set into the initial evaluation model and calculate the model's health level determination accuracy (≥92%). If the accuracy is not up to standard, adjust the number of random forest decision trees and the maximum depth parameter, retrain the model and calibrate the weights until the accuracy requirements are met, and generate a quantitative evaluation model for the circuit breaker's health status.

[0105] The working principle and effects of the above technical solution are as follows:

[0106] By removing outlier samples with a missing data rate >5% and labeling them according to four health levels, the interference of poor-quality data on model training was reduced, and learning bias caused by chaotic sample labels was avoided, thus improving the reliability and effectiveness of the training dataset. By configuring random forest parameters according to sample size (100-200 decision trees, maximum depth 8-12 layers), model accuracy and computational efficiency were balanced, reducing overfitting or underfitting problems caused by unreasonable parameter configuration, avoiding performance loss due to blindly setting parameters, and enhancing the targeted nature of model training. By dividing the training and validation sets in a 7:3 ratio, the model can fully learn feature mapping relationships while retaining an independent validation space. This approach reduces the generalization limitations caused by over-reliance on training data, avoids the bias of training on a single dataset, and improves the adaptability of the initial model. By adjusting feature weights in conjunction with power industry standards and clarifying fault diagnosis priorities, it reduces the deviation between subjective weight allocation and industry norms, avoids the problem of model output not meeting actual operation and maintenance needs, and enhances the compliance and practicality of the evaluation model. By verifying accuracy on the validation set (requiring ≥92%) and iteratively optimizing parameters and weights, it ensures that the model meets diagnostic requirements, reduces the risk of using unqualified models, avoids misjudgments of health status due to insufficient model accuracy, and improves the accuracy and reliability of quantitative assessment of circuit breaker health status.

[0107] In one embodiment of the present invention, step S5 includes:

[0108] S51. Based on the requirements of the circuit breaker equipment manual, industry operation and maintenance standards and historical fault data statistics, set three-level early warning thresholds (normal threshold, attention threshold, and abnormal threshold), and support dynamic adjustment according to the equipment's operating years and environmental conditions.

[0109] S52. Compare the circuit breaker health status assessment data with the three-level early warning threshold. If the health status score is within the normal threshold range, it is determined to be a normal early warning level; if it is below the normal threshold but above the attention threshold, it is determined to be a attention early warning level; if it is below the attention threshold but above the abnormal threshold, it is determined to be an abnormal early warning level; if it is below the abnormal threshold, it is determined to be an emergency early warning level. Generate three-level early warning data for the circuit breaker operating coil current characteristics.

[0110] S53. For different warning levels, automatically generate corresponding warning information, including warning level, triggering reason (such as peak deviation exceeding the standard, extended action time, etc.), possible fault types (refer to mechanical characteristic data analysis criteria), suggested handling measures and priorities, forming a complete warning information package.

[0111] S54. On the 10-inch touch screen of the waveform recording terminal, the corresponding warning information is displayed intuitively. The corresponding warning information includes the warning level (marked by different colors: green - normal, yellow - attention, orange - abnormal, red - emergency), current waveform diagram, horizontal and vertical comparison results, abnormal characteristic parameters, and other corresponding warning information, and supports on-site viewing and operation by maintenance personnel.

[0112] S55: Through the WEB remote access function, the three-level early warning data, early warning information package and raw waveform data are synchronously pushed to the remote operation and maintenance platform, allowing operation and maintenance personnel to view them anytime and anywhere, and at the same time record the early warning historical data.

[0113] The working principle and effects of the above technical solution are as follows:

[0114] By combining equipment manuals, industry standards, and historical data, a three-level early warning threshold is set, supporting dynamic adjustment based on years of operation and working conditions. This reduces the adaptability limitations of fixed thresholds and avoids false or missed warnings caused by unreasonable thresholds, improving the scientific rigor and adaptability of the warning thresholds. A four-level warning system (Normal / Attention / Abnormal / Emergency) accurately determines the equipment's health status, reducing decision-making difficulties caused by ambiguous status definitions, preventing misjudgments of risk levels by maintenance personnel, and enhancing the recognizability of warning levels. The system automatically generates warning information packages containing triggering causes, fault types, handling measures, and priorities, eliminating the need for additional manual analysis and reducing blind inspections. The system reduces repair time costs, prevents the escalation of faults due to inappropriate measures, and improves the targeting and efficiency of maintenance responses. A 10-inch touchscreen displays early warning information locally and intuitively, using color to indicate levels and simultaneously presenting waveforms and abnormal parameters. It supports quick on-site viewing and operation, reducing the difficulty of obtaining information for on-site maintenance, avoiding response delays caused by complex operations, and enhancing the convenience of on-site handling. Web-based remote push of early warning data and historical records allows for viewing anytime, anywhere, reducing reliance on on-site monitoring, avoiding the problem of not being able to promptly detect faults in remote locations, improving the flexibility and coverage of maintenance, and providing complete data support for subsequent traceability and analysis.

[0115] One embodiment of the present invention provides a circuit breaker operating coil current characteristic diagnostic analysis system, comprising:

[0116] One or more processors;

[0117] Memory, used to store one or more programs;

[0118] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method described in any one of the above.

[0119] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for diagnosing and analyzing the current characteristics of a circuit breaker's operating coil, characterized in that, The method includes: S1: Synchronously record the current of multiple circuit breaker operating coils on the DC feeder panel side to generate multi-channel synchronous recording data; at the same time, collect SOE message information, and use the timestamp and operation identification information in the SOE message to automatically match and locate the specific operating circuit breaker and generate operating circuit breaker location data. S2: Based on the circuit breaker positioning data, filter the multi-channel synchronous waveform data and extract the coil current waveform data of the corresponding circuit breaker; preprocess the coil current waveform data to generate preprocessed coil current waveform data; S3: Based on the preprocessed coil current waveform data, perform a horizontal waveform comparison to generate horizontal waveform difference data; at the same time, perform a vertical waveform comparison to generate vertical waveform change data. S4: Based on the transverse waveform difference data and the longitudinal waveform change data, extract key feature quantities. Using the key feature quantities and combined with the preset health status assessment model, quantitatively assess the health status of the operating circuit breaker and generate circuit breaker health status assessment data. S5: Based on the circuit breaker health status assessment data, set three-level early warning thresholds, each corresponding to a different health status level; compare the circuit breaker health status assessment data with the three-level early warning thresholds, perform risk early warning processing, and generate three-level early warning data for the circuit breaker operating coil current characteristics.

2. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 1, characterized in that, S1 includes: S11. Install a multi-channel waveform recording terminal on the DC feeder panel side of the substation, configure 16 synchronous acquisition channels, set the sampling frequency to 4-10KHZ, and enable the IRIG-B code time synchronization function. S12. The waveform recording terminal monitors the total current of the DC feeder circuit in real time. It adopts a sudden change triggering mechanism. When the current sudden change reaches the preset threshold, it automatically starts multi-channel synchronous waveform recording, collects the current data of multiple circuit breaker operating coils, and generates a multi-channel synchronous waveform recording raw dataset. S13. Synchronously collect SOE message information sent by the substation backend system, extract key fields in the message, and form a set of key information of SOE message; S14. Using the IRIG-B code time synchronization result of the waveform recording terminal, perform time axis calibration on the original dataset of multi-channel synchronous waveform recording and the key information set of SOE messages. S15. Based on the calibrated timestamp and action identification information, establish a correlation mapping model between the waveform recording data and SOE messages, automatically identify the circuit and number of the circuit breaker that triggered the waveform recording, and generate accurate positioning data of the circuit breaker.

3. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 2, characterized in that, S14 includes: Read the IRIG-B code time synchronization signal output in real time from the waveform recording terminal, analyze the second pulse, minute pulse and time code information in the signal, and generate a standard timestamp sequence of IRIG-B code with microsecond-level precision as a unified reference for time calibration. Retrieve the original timestamps of the multi-channel synchronous waveform recording dataset, compare them point by point with the IRIG-B code standard timestamp sequence, calculate the time deviation value of each waveform recording data, and perform linear compensation correction on the original timestamps based on the deviation value to generate a calibrated multi-channel waveform recording dataset. Analyze the action timestamps in the key information set of SOE messages, combine the network transmission delay between the substation backend and the waveform recording terminal, align the SOE timestamps with the IRIG-B code standard timestamp sequence, correct the time deviation caused by the transmission delay, and generate a calibrated SOE message information set. Select the overlapping time intervals between the calibrated multi-channel waveform dataset and the calibrated SOE message information set, randomly select 20-30 time nodes, calculate the absolute deviation of the timestamps of the two, and if the maximum deviation is ≤5μs, the time alignment is deemed qualified and a time calibration accuracy verification report is generated. Based on the time calibration accuracy verification report, the calibrated multi-channel waveform recording dataset and the calibrated SOE message information set are associated and marked according to a unified IRIG-B code time axis to generate a time-calibrated associated dataset.

4. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 1, characterized in that, The S2 includes: S21. Based on the precise positioning data of the circuit breaker, select the coil current recording data of the corresponding channel of the circuit breaker from the original multi-channel synchronous recording dataset, remove other irrelevant channel data, and generate a single-target circuit breaker recording dataset. S22. Adaptive Kalman filtering algorithm is used to filter the waveform data of a single-target circuit breaker. S23. Apply wavelet thresholding to suppress noise in the filtered dataset and generate a denoised dataset. S24. Convert the noise-reduced dataset into the COMTRADE99 format commonly used in the power industry, unify parameter standards, and generate a standardized coil current dataset. S25. Perform integrity verification on the standardized coil current dataset, interpolate and repair incomplete data, and generate preprocessed coil current waveform data.

5. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 1, characterized in that, The S3 includes: S31. Collect coil current waveform data of normal circuit breakers of the same type, with the same service life and no fault records, extract their typical characteristic parameters, and construct a normal waveform benchmark database of circuit breakers of the same type. S32. Align the feature points of the preprocessed coil current waveform data with those of the waveform in the normal waveform reference database. S33. Using the dynamic time warping algorithm, calculate the difference values ​​between the current waveform and the reference waveform in terms of characteristic parameters and waveform morphology, and generate a horizontal waveform difference dataset. S34. Extract the coil current waveform data of all past effective operations of the circuit breaker, organize them in the order of operation time, and build a historical waveform database dedicated to the circuit breaker. S35. Compare the current waveform with the first action waveform and the average waveform of the three most recent normal actions in the historical waveform database, calculate the time-series change amount and change rate of each characteristic parameter, and generate a longitudinal waveform change trend dataset. S36. Merge the horizontal waveform difference dataset with the vertical waveform change trend dataset to form a complete two-dimensional waveform comparison analysis result set.

6. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 5, characterized in that, S36 includes: S361. Extract the feature parameters of the horizontal waveform difference dataset and the vertical waveform change trend dataset, unify the parameter units and data format, and generate a normalized two-dimensional feature dataset. S362. Based on the historical fault case library of circuit breakers, analyze the contribution of horizontal differences and vertical changes to fault diagnosis, use the analytic hierarchy process to determine the fusion weight of each feature parameter, and generate a two-dimensional feature fusion weight matrix. S363. Using the normalized two-dimensional feature dataset as input, and combining it with the fusion weight matrix, perform weighted summation on the same feature dimensions and feature splicing on different feature dimensions to generate a preliminary two-dimensional fusion dataset. S364. Establish verification rules, mark contradictory data in the preliminary two-dimensional fusion dataset, correct outliers based on the recent operating conditions of the circuit breaker, and generate a verified fusion dataset. S365. From the fused dataset after verification, the top 3 feature parameters with the highest weights are selected as the core dimensions of difference. The trends of more than 3 consecutive changes in the time series are extracted as the key trends of change. The results are then integrated to form a complete set of two-dimensional waveform comparison analysis results.

7. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 6, characterized in that, S363 includes: Traverse the normalized two-dimensional feature dataset, identify and extract the same feature dimension parameter pairs corresponding in the horizontal and vertical directions, group them by feature category, and generate a set of the same feature dimension parameter pairs. For each set of parameters in the same feature dimension parameter pair set, retrieve the weight value of the corresponding parameter in the fusion weight matrix, calculate the weighted sum of each set of parameters according to the formula (horizontal parameter value × horizontal weight) + (vertical parameter value × vertical weight), and generate a weighted fusion result set with the same feature dimension. The normalized two-dimensional feature dataset is traversed again to filter out independent feature parameters that have no horizontal or vertical correspondence. These parameters are then classified and organized according to their horizontal and vertical independence to generate independent parameter sets for different feature dimensions. The weighted fusion result set with the same feature dimensions and the independent parameter sets with different feature dimensions are concatenated according to the fixed feature order of peak, duration, similarity and rate, and the data field format is unified to generate a preliminary two-dimensional fusion dataset.

8. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 1, characterized in that, The S4 includes: S41. From the set of dual-dimensional waveform comparison analysis results, key feature quantities are automatically extracted to form a multi-dimensional feature parameter set; S42. Based on machine learning algorithms and combined with the power industry circuit breaker fault diagnosis standards, input historical fault case data and normal operation data to train the model, construct a quantitative assessment model of circuit breaker health status, and set the feature parameter weight coefficients. S43. Normalize the multi-dimensional feature parameter set to eliminate the range differences between different parameters and generate a standardized feature parameter set. S44. Input the standardized feature parameter set into the health status quantitative assessment model, and obtain the circuit breaker health status score through model calculation. The higher the score, the better the health status. S45. Based on the health status score results and combined with the preset level classification rules, generate circuit breaker health status assessment data.

9. The method for diagnosing and analyzing the current characteristics of a circuit breaker operating coil according to claim 1, characterized in that, The S5 includes: S51. Set three-level early warning thresholds and support dynamic adjustment based on equipment operating years and environmental conditions; S52. Compare the circuit breaker health status assessment data with the three-level early warning threshold. If the health status score is within the normal threshold range, it is determined to be a normal early warning level; if it is below the normal threshold but above the attention threshold, it is determined to be a attention early warning level; if it is below the attention threshold but above the abnormal threshold, it is determined to be an abnormal early warning level; if it is below the abnormal threshold, it is determined to be an emergency early warning level. Generate three-level early warning data for the circuit breaker operating coil current characteristics. S53. For different warning levels, automatically generate corresponding warning information, including the warning level and triggering reason, to form a complete warning information package; S54. On the 10-inch touch screen of the waveform recording terminal, the corresponding warning information is displayed intuitively, including the warning level; S55. Through the WEB remote access function, the three-level early warning data, early warning information package, and raw waveform data are synchronously pushed to the remote operation and maintenance platform.

10. A circuit breaker operating coil current characteristic diagnostic analysis system, comprising: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1 to 9.

Images