High-voltage circuit breaker and mechanical characteristic health diagnosis method and system thereof

Abstract

The invention belongs to the technical field of circuit breaker fault diagnosis, and particularly relates to a high-voltage circuit breaker and a mechanical characteristic health diagnosis method and system thereof. The method comprises the following steps: S1, acquiring operation state information of the high-voltage circuit breaker in a switching-off / switching-on process; s2, inputting the running state information into a pre-trained fault diagnosis model to obtain a fault state type of the mechanical characteristics of the high-voltage circuit breaker; the fault diagnosis model comprises a strong classifier, and the strong classifier is a classifier obtained by combining at least two weak classifiers through an AdaBoost algorithm. According to the method, the classification model is constructed by adopting a multi-algorithm fusion technology through the AdaBoost algorithm, and the AdaBoost algorithm enables the model to pay more attention to samples which are difficult to classify by adjusting the weight of data points, so that the classification effect of the model is enhanced. The technical problem of low fault diagnosis precision caused by low classification precision of fault data difficult to classify in the prior art is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of circuit breaker fault diagnosis, and in particular relates to a high-voltage circuit breaker and a method and system for diagnosing the health of its mechanical characteristics. Background Art

[0002] High-voltage switchgear refers to electrical equipment such as circuit breakers, disconnectors, and load switches with a rated voltage of 3kV or above. High-voltage switchgear is crucial infrastructure in power grid construction, and clusters of high-voltage switchgear form the physical backbone of modern power grids. These devices play a dual role in smart grid architectures: they provide isolation for network topology reconstruction and precisely dispatch power flows. When abnormal system conditions occur, high-voltage switches can strategically disconnect faulty circuits, protecting the stable operation of healthy grid segments while enabling intelligent regulation of power supply quality. As a core device, circuit breakers are often considered synonymous with high-voltage switchgear due to their exceptional circuit-breaking performance.

[0003] The vacuum circuit breaker consists of six core components: a support frame, a vacuum arc extinguishing device, a conductive system, a mechanical transmission unit, an insulating support, and an actuator. The opening / closing electromagnetic coil, a key sensing unit for control, uses Hall-effect-based current sensing technology to digitally capture its dynamic current characteristics. By analyzing the coil current characteristic parameters, potential equipment failure modes can be effectively predicted. This diagnostic method enables rapid and accurate diagnosis and early warning of major operational anomalies such as refusal to operate or malfunction. The actuator, the core drive unit of the vacuum circuit breaker, directly determines the accuracy of the opening and closing sequence control. The spring-loaded operating system features a modular design, consisting of an independent closing energy storage unit and an opening release unit. These two units are symmetrically arranged, and energy transfer is achieved through a transient collision mechanism on a four-bar linkage shaft. A unitized maintenance approach allows for rapid replacement of individual functional modules if they experience an anomaly, significantly shortening equipment offline maintenance cycles and reducing operational costs.

[0004] Current circuit breaker fault diagnosis methods primarily include offline manual testing and offline sensor monitoring. Offline manual testing involves periodic manual inspection of the circuit breaker's mechanical condition, determining whether it is functioning properly by observing light bulbs or vibrations. Offline sensor monitoring involves installing sensor devices, such as vibration sensors and accelerometers, to detect changes in the circuit breaker's mechanical properties. However, offline manual testing cannot monitor the circuit breaker's mechanical condition in real time, potentially preventing faults from occurring before they occur, impacting the stable operation of the power system. Offline sensor monitoring requires specialized sensor equipment, increasing system costs and maintenance workload.

[0005] The Chinese invention patent application with the existing application publication number CN117192347A and the application publication date of December 8, 2023, discloses a high-voltage circuit breaker online monitoring and fault diagnosis system and method. The system obtains the circuit breaker's opening and closing coil current signal and contact travel signal through a current sensor and an angular displacement sensor, respectively. After the current signal and the travel signal are median filtered, characteristic parameters are extracted, including the amplitude of each peak / trough in the current signal and the iron core contact duration, iron core movement duration, opening / closing duration, etc., as well as the contact starting position, maximum displacement, stable position, total travel and corresponding time information in the travel signal. Based on the extracted characteristic parameters, LSSVM (Least Square Support Vector Machine) is used to perform fault classification to obtain the circuit breaker fault type. The above scheme uses LSSVM to perform fault diagnosis based on the mechanical characteristics of the circuit breaker, simplifying the calculation and improving the fault diagnosis efficiency. However, the classification accuracy of LSSVM is low, resulting in low fault diagnosis accuracy. Summary of the Invention

[0006] The object of the present invention is to provide a high-voltage circuit breaker and a mechanical characteristic health diagnosis method and system thereof, so as to solve the technical problem of low fault diagnosis accuracy caused by low classification accuracy in the prior art.

[0007] To solve the above technical problems, the present invention provides a technical solution for a high-voltage circuit breaker mechanical characteristic health diagnosis method: a high-voltage circuit breaker mechanical characteristic health diagnosis method, the method comprising:

[0008] S1. Obtaining the operating status information of the high-voltage circuit breaker during the opening / closing process;

[0009] S2. Inputting the operating status information into a pre-trained fault diagnosis model to obtain a fault status type of the mechanical characteristics of the high-voltage circuit breaker;

[0010] The fault diagnosis model includes a strong classifier, which is a classifier obtained by combining at least two weak classifiers through the AdaBoost algorithm.

[0011] The beneficial effect of the above technical solution is that the technical solution of the present invention, a method for diagnosing the mechanical characteristics of a high-voltage circuit breaker, is an improved invention. Unlike traditional single algorithms, the present invention uses the AdaBoost algorithm to construct a classification model using multi-algorithm fusion technology. Through training, a mapping relationship between circuit breaker faults and operating status information is established. The AdaBoost algorithm adjusts the weights of data points, making the model pay more attention to difficult-to-classify samples, thereby enhancing the model's classification effect. The present invention solves the technical problem of low fault diagnosis accuracy caused by low classification accuracy for difficult-to-classify fault data in the prior art.

[0012] Furthermore, the weak classifier is a random forest model.

[0013] Furthermore, the operating status includes coil current information of the opening / closing electromagnetic coil and contact stroke information of the operating actuator.

[0014] Furthermore, the contact stroke information includes a time-frequency representation of a stroke characteristic obtained by Hilbert-Huang transforming a stroke-time characteristic waveform.

[0015] Furthermore, the coil current information includes current characteristic parameters extracted from the current-time characteristic waveform, and the current characteristic parameters include the moment when the moving iron core starts to move and its corresponding current amplitude, the moment when the moving iron core moves to its position and its corresponding current amplitude, the moment when the moving contact starts to move and its corresponding current amplitude, the moment when the coil circuit power supply is cut off and its corresponding current amplitude, and the moment when the coil current disappears after the coil circuit power supply is cut off.

[0016] Furthermore, the contact stroke information includes stroke characteristic parameters extracted from a stroke-time characteristic waveform, and the stroke characteristic parameters include the moment when the contacts are just opened / closed and the corresponding stroke and opening / closing speed.

[0017] Furthermore, the current-time characteristic waveform is obtained by fitting the original current-time signal of the opening / closing electromagnetic coil collected by the least squares method into a polynomial curve, and obtaining the current-time characteristic waveform according to the fitted polynomial curve.

[0018] Furthermore, the stroke-time characteristic waveform is obtained by fitting the collected original stroke-time signal of the contact into a polynomial curve, and obtaining the stroke-time characteristic waveform according to the fitted polynomial curve;

[0019] The method of fitting the collected original travel-time signal of the contact into a polynomial curve specifically includes: dividing the collected original travel-time signal into at least two intervals in time; fitting a polynomial curve in each interval respectively; fitting or interpolating at the segmentation point between the two intervals to make the fitting curve continuous; and combining the fitting curves in each interval to obtain a polynomial curve corresponding to the travel characteristic.

[0020] The present invention also provides a technical solution for a high-voltage circuit breaker mechanical characteristic health diagnosis system: a high-voltage circuit breaker mechanical characteristic health diagnosis system, comprising a processor, which is used to execute a computer program to implement the steps of the high-voltage circuit breaker mechanical characteristic health diagnosis method as described above.

[0021] The present invention also provides a technical solution for a high-voltage circuit breaker: a high-voltage circuit breaker, comprising a health diagnosis module and a data acquisition module for obtaining operating status information of the high-voltage circuit breaker during the opening / closing process, the health diagnosis module comprising a processor, the processor being used to execute a computer program to implement the steps of the high-voltage circuit breaker mechanical characteristic health diagnosis method as described above.

[0022] The beneficial effect of the above technical solution is that the technical solution for a high-voltage circuit breaker of the present invention is an improved invention. Unlike traditional single algorithms, the present invention uses the AdaBoost algorithm to build a classification model using multi-algorithm fusion technology. Through training, a mapping relationship between circuit breaker faults and operating status information is established. The AdaBoost algorithm adjusts the weights of data points, making the model pay more attention to difficult-to-classify samples, thereby enhancing the model's classification effect. This solves the technical problem of low fault diagnosis accuracy caused by low classification accuracy for difficult-to-classify fault data in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a schematic diagram of the overall system framework of an embodiment of the high-voltage circuit breaker mechanical characteristic health diagnosis system of the present invention;

[0024] Figure 2 A health diagnosis flow chart of an embodiment of a high-voltage circuit breaker mechanical characteristic health diagnosis system of the present invention;

[0025] Figure 3 This is the stroke-time characteristic waveform of the high-voltage circuit breaker mechanical characteristic health diagnosis system during closing of the embodiment of the present invention;

[0026] Figure 4 The current-time characteristic waveform of the high-voltage circuit breaker mechanical characteristic health diagnosis system during closing of the present invention;

[0027] Figure 5 This is a HHT conversion flow chart of an embodiment of the high-voltage circuit breaker mechanical characteristic health diagnosis system of the present invention;

[0028] Figure 6 This is a wavelet denoising flow chart of an implementation scheme of the high-voltage circuit breaker mechanical characteristic health diagnosis system of the present invention. DETAILED DESCRIPTION

[0029] Unlike traditional single algorithms, this invention uses the AdaBoost algorithm to build a classification model using multi-algorithm fusion technology. Through training, a mapping relationship between circuit breaker faults and operating status information is established. The AdaBoost algorithm adjusts the weights of data points, allowing the model to focus more on difficult-to-classify samples, thereby enhancing the model's classification effectiveness. This invention solves the technical problem of low fault diagnosis accuracy caused by low classification accuracy for difficult-to-classify fault data in the prior art.

[0030] Implementation method of high-voltage circuit breaker mechanical characteristics health diagnosis system:

[0031] like Figure 1 As shown, the high-voltage circuit breaker mechanical characteristic health diagnosis system includes a data layer for collecting circuit breaker operation data and a fault diagnosis layer for performing data processing.

[0032] The data layer uses Hall current sensors and angular displacement sensors to capture the current signal and contact travel signal of the electromagnetic coil in real time during the opening (closing) process. The Hall current sensor collects the coil current waveform at a sampling frequency of 4.8kHz. The sensor node encapsulates the data into a COMTRADE file that complies with the IEC 60255-24 standard via the FTP protocol through industrial Ethernet. The file contains two parts: the CFG configuration file records metadata such as channel mapping and sampling rate; and the DAT data file stores the actual current and displacement signals. After the file is sent to the host computer, the system first parses the CFG file to obtain key information such as the sampling rate, timestamp, and channel mapping relationship: Channel 1 is the current signal, Channel 2 is the displacement signal. Then, the binary data in the DAT file is read, converted into a floating-point array, and cached in the Redis memory database to reduce the delay of subsequent reading. According to the channel mapping information in the CFG file, the corresponding current and displacement data are extracted from the cache and data analysis begins. The main steps of data analysis are as follows: (1) First, determine whether the operation is opening or closing; (2) Calculate the characteristic parameters of the coil current and the characteristic parameters of the contact travel. (3) These characteristic parameters are stored in the mechanical property data table of the MySQL database in JSON format. This completes the work of the data layer and provides a high-quality data foundation for subsequent health diagnosis and evaluation.

[0033] The system uses a scheduled task scheduling mechanism to periodically scan new operation records in the database to ensure data integrity. The scanning task is based on the data information table and uses the analysis_type field to filter out all unprocessed records after the most recent scan timestamp. A single scan can process up to 1,000 data items to avoid excessive single-task load. At the same time, the system supports breakpoint resumption, recording the timestamp of the most recent successful scan and persistently storing it to ensure that it can recover from the interruption point after a network interruption or system anomaly, avoiding data loss or duplicate processing. The processor of the fault diagnosis layer uses the AdaBoost-RF model (i.e., the fault diagnosis model) to perform fault analysis on the acquired data to implement a high-voltage circuit breaker mechanical characteristic health diagnosis system method. After the diagnosis is completed, the diagnosis results, including the fault type, the timestamp of the original feature data, and the operation and maintenance strategy dynamically generated based on historical fault cases and expert experience, are written back to the database. At the same time, the analysis_type field of the corresponding record in the data information table is updated, completing a complete closed loop from data scanning to health diagnosis.

[0034] In this embodiment, the fault type status (i.e., the output result of the above-mentioned AdaBoost-RF model) includes a stuck release; poor contact of the auxiliary switch; decreased switching capacity of the auxiliary switch; opening / closing speed higher than the design range; opening / closing speed lower than the design range; abnormal contact bounce; and low coil voltage.

[0035] The specific diagnostic process of the high-voltage circuit breaker mechanical characteristics health diagnosis system method is as follows: Figure 2 As shown in the figure, after the system is initialized and the system self-check is completed, the collected original current-time signal and original travel-time signal data are preprocessed. The parameters such as the opening and closing current-time characteristic waveform and the travel-time characteristic waveform obtained based on the processed data are used as input parameters, and the algorithm is used to output the waveform characteristic value for further health diagnosis. If the diagnosis is abnormal, an alarm is triggered.

[0036] Specifically, check the sample number, feature number, data type and other information of the dataset (including the original current-time signal collected by the Hall current sensor and the original travel-time signal collected by the angular displacement sensor), then check whether the dataset has missing values, duplicate values ​​and outliers, use interpolation methods to fill in missing values, select and delete the rows where duplicate values ​​are located to process duplicate values, and after completing data cleaning, save the cleaned dataset.

[0037] The data in the dataset is further processed using the least squares method to obtain the current-time characteristic waveform and the travel-time characteristic waveform:

[0038] Assume that the current fitting polynomial curve is:

[0039]

[0040] Where: m is the number of curve fitting; a0, a1, ..., a m is the regression coefficient to be determined, x represents time, and n is the number of samples.

[0041] According to the least squares fitting principle, the sum of square errors Q is minimized:

[0042]

[0043] The problem is transformed into solving Q=p(a0,a1,…,a m ), for a j (j=0,1,2,..,m) find the partial derivative and set it equal to 0:

[0044]

[0045] Arrangement can be obtained as follows:

[0046]

[0047] Convert the above equations into matrix representation:

[0048]

[0049] Further simplification yields:

[0050]

[0051] make

[0052] We can get X×A=Y, and get the coefficient matrix of the fitting curve A=(X T X) -1 X T Y, the current fitting polynomial f(x) can be further obtained as the travel-time characteristic waveform.

[0053] The trend of the circuit breaker opening / closing travel curve (travel-time characteristic waveform) is similar to a trapezoid, and the points immediately after opening and closing contain important information about the circuit breaker's motion characteristics. A piecewise curve fitting method can be used to perform local least squares fitting within each interval to obtain a more accurate circuit breaker opening / closing travel time curve. However, data point jumps may occur near the boundaries of adjacent intervals, resulting in curve discontinuity. To address this issue, a continuous fitting method based on least squares is adopted. The online monitored travel time data is piecewise fitted, and refitting and interpolation are performed at the piecewise points to achieve continuity of the fitting curve and reduce fitting errors.

[0054] The circuit breaker opening and closing characteristic waveforms (including the travel-time characteristic waveform and the travel-time characteristic waveform) are obtained by the least squares method. Furthermore, the waveforms are processed using the wavelet noise reduction method. Figure 6 Because the scaling factor and translation factor in the continuous wavelet transform are both continuously changing real numbers, in practical applications, the collected data are all discrete. Therefore, a discrete wavelet transform (DWT) is required. The discretization of the scaling factor a and the translation factor b is usually done as follows:

[0055]

[0056] The corresponding discrete wavelet function is:

[0057]

[0058] The discrete wavelet transform of the signal to be processed is:

[0059]

[0060] When a0=2, b0=1, we can get the orthogonal binary wavelet:

[0061]

[0062] In summary, the characteristic waveform related to the mechanical characteristics of the circuit breaker is obtained.

[0063] Further, feature points are extracted from the feature waveform, referring to Figure 3 and Figure 4 The structural characteristics of the high-voltage vacuum circuit breaker are analyzed. According to the working principle of the opening and closing coils and the opening and closing current curves, T1 to T5 are selected as the current characteristic points.

[0064] t0~t1 stage: At t0, the closing coil is energized. At this time, the air gap in the magnetic circuit is the largest, the air gap magnetic resistance is the largest, and the inductance is the smallest.

[0065] Stage t1 to 12: At time t1, the moving iron core starts to move, the air gap in the magnetic circuit decreases, the air gap magnetic resistance decreases, the coil inductance increases, and the coil current gradually decreases.

[0066] Stage t2~t3: At t2, the moving iron core moves into place, the striker hits the closing trigger to stop moving, the energy storage mechanism releases the stored energy, and the moving contact starts to move.

[0067] Phase t3-t4: At t3, the moving contact begins to operate, and the coil current is approximately steady-state. The magnitude of the coil current I3 during this phase reflects the magnitude of the coil voltage and the control loop resistance.

[0068] Stage t4~t5: At t4, the DC power supply of the coil circuit is cut off, an arc is generated between the auxiliary contacts and is rapidly elongated. The arc voltage increases, causing the current to decrease until the arc is extinguished and the coil current is reduced to zero.

[0069] The Hilbert-Huang (HHT) transform is used to process the denoised opening and closing stroke curve, successfully extracting the opening and closing points, and further calculating characteristic parameters such as the opening and closing speed. Figure 5 .

[0070] The core idea of ​​the HHT transform is to decompose the signal through EMD (Empirical Mode Decomposotion) into a series of IMFs (Intrinsic Mode Functions), and then perform a Hilbert transform on each IMF to obtain the time-frequency representation of the signal. The EMD method is adaptive and can effectively overcome the difficulties in selecting basis functions and the energy leakage problems in the signal processing process in traditional methods. Therefore, it has become an efficient tool for analyzing nonlinear and non-stationary vibration signals. EMD is the core step of HHT, and its goal is to decompose the signal into several IMFs. The HHT transform process is shown in the figure, and the specific steps are as follows:

[0071] (1) Let the original signal be x(t) and the residual be r(t). The signal x(t) can be expressed as the sum of several IMFs and residuals, as shown in the following formula:

[0072]

[0073] Among them, the IMF i (t) is the i-th IMF; r n (t) is the final residual, n is the number of IMFs;

[0074] (2) Find all local maxima and local minima of the signal. Use the differential threshold method to identify the strictly local maximum points in the signal waveform that meet the conditions. Connect the extreme points using the cubic spline interpolation algorithm to form the upper and lower envelopes. Based on the envelope pair, calculate the instantaneous mean m(t). Subtract the mean from the signal to obtain the candidate IMF, as shown in the following formula:

[0075] h(t)=r(t)-m(t)

[0076] (3) Check whether h(t) satisfies the IMF condition. If not, repeat step (2) until the condition is satisfied. If so, h(t) is an IMF.

[0077] (4) Update the residual as follows:

[0078] r(t)=r(t)-h(t)

[0079] Repeat steps (2) to (3) for the residual r(t) until the residual is a monotonic function or no more IMFs can be extracted;

[0080] (5) Perform Hilbert transform on each IMFc(t), and its mathematical expression can be expressed as follows:

[0081]

[0082] Where PV represents the Cauchy principal value integral; the Hilbert transform can be regarded as a filter that delays the signal phase by 90 degrees;

[0083] (6) Construct the analytical signal of IMF and calculate the instantaneous frequency and instantaneous amplitude through Hilbert transform:

[0084]

[0085] Where A(t) is the instantaneous amplitude envelope; is the instantaneous phase function; A(t) and The mathematical expression is as follows:

[0086]

[0087] The instantaneous frequency is defined as the derivative of the phase, and the mathematical expression can be expressed as follows:

[0088]

[0089] (7) Combine the instantaneous frequency and instantaneous amplitude of each IMF to obtain the time-frequency representation of the signal:

[0090]

[0091] Where n is the number of IMFs, A i (t) is the instantaneous amplitude of the ith IMF, f i (t) is the instantaneous frequency of the i-th IMF; f is the natural frequency of the i-th IMF.

[0092] In summary, the characteristic parameters related to the mechanical characteristics of the circuit breaker are obtained.

[0093] Furthermore, health diagnosis is performed based on characteristic parameters. Aiming at the stroke characteristic parameters and current characteristic parameters of high-voltage vacuum circuit breaker, a high-voltage circuit breaker health diagnosis algorithm based on AdaBoost optimized random forest is proposed.

[0094] Adaptive Boosting (AdaBoost) is a boosting algorithm that aims to improve the accuracy of weak classifiers or weak learners by combining multiple weak classifiers into a strong classifier. AdaBoost was proposed by Yoav Freund and Robert Schapire in 1996 and has performed well in many machine learning tasks, especially classification problems. AdaBoost improves classification results by iteratively training a series of weak classifiers. Each weak classifier is assigned a different importance weight in the previous round of training. Specifically, the core idea of ​​AdaBoost is to make the model pay more attention to samples that are difficult to classify by adjusting the weights of data points. This weighting mechanism makes AdaBoost particularly outstanding when dealing with imbalanced datasets.

[0095] The AdaBoost process can be described in the following steps:

[0096] 1) Initialize weights:

[0097] Given a training set S = {(x1,y1),(x2,y2),...,(x n ,y n )}, each data point (x i ,y i ) have the same initial weights, namely:

[0098]

[0099] Where n is the number of samples in the training set, and D1(i) represents the weight of sample i. (x i ,y i ) i Represents the sample feature values ​​input to the classifier, including the above-mentioned current feature parameters, travel feature parameters and the time-frequency representation of the travel-time waveform obtained by HHT; y i Represents the eigenvalue x i The labels include no fault and seven fault types (respectively, stuck release, poor auxiliary switch contact, reduced auxiliary switch switching capacity, opening / closing speed higher than the design range, opening / closing speed lower than the design range, abnormal contact bounce and low coil voltage).

[0100] 2) Training weak classifiers:

[0101] In each iteration, AdaBoost trains a weak classifier h t (x), whose goal is to minimize the weighted classification error:

[0102]

[0103] where ht (x i ) is the prediction result of the classifier for sample i obtained in the tth round of iteration, y i is the true label of sample i, I is the indicator function, which is 1 if the prediction is wrong (i.e. the inequality holds), and 0 if the prediction is correct (i.e. the inequality does not hold).

[0104] 3) Update classifier weights:

[0105] For the classifier h in round t t (x), its weight α t Calculated by the following formula:

[0106]

[0107] Among them, ε t is the weighted error rate of the classifier in round t. t The role of is to adjust the importance of the classifier in the final combination.

[0108] 4) Update sample weights:

[0109] After each round of training, AdaBoost adjusts the weights of the samples based on the performance of the current classifier. The weights of incorrectly classified samples will be increased, while the weights of correctly classified samples will be reduced. The new sample weights are updated as follows:

[0110] D t+1 (i) = D t (i)*exp(-α t y i h t (x i ))

[0111] In this way, more misclassified samples will receive more attention in the next round of training

[0112] 5) Final classifier:

[0113] The final strong classifier is a weighted combination of all weak classifiers:

[0114]

[0115] Where T is the total number of iterations, α t is the weight of each weak classifier, h t (x) is the weak classifier obtained in the tth round of iteration.

[0116] In this embodiment, each weak classifier is a random forest model. In other embodiments, the weak classifier can select other classification models based on machine learning, such as support vector machines, long short-term memory neural networks, or decision trees, or different types of weak classifiers can be selected to form the final strong classifier, such as multiple weak classifiers based on support vector machines + multiple weak classifiers based on long short-term memory neural networks to form the final strong classifier.

[0117] Implementation method of high-voltage circuit breaker mechanical characteristics health diagnosis method:

[0118] A method for diagnosing the mechanical health characteristics of a high-voltage circuit breaker comprises: S1. Obtaining operational status information of the high-voltage circuit breaker during its opening and closing processes; S2. Inputting the operational status information into a pre-trained fault diagnosis model to determine the fault status type of the high-voltage circuit breaker's mechanical characteristics; the fault diagnosis model includes a strong classifier, which is a classifier obtained by combining at least two weak classifiers using the AdaBoost algorithm. The specific method for diagnosing the mechanical health characteristics of a high-voltage circuit breaker has been described in sufficient detail in the aforementioned embodiment of the high-voltage circuit breaker mechanical health diagnosis system and will not be repeated here.

[0119] High voltage circuit breaker implementation method:

[0120] A high-voltage circuit breaker includes a health diagnostic module and a data acquisition module for acquiring operating status information during the high-voltage circuit breaker opening and closing process. The health diagnostic module includes a processor configured to execute a computer program to implement the steps of the above-described method for diagnosing the mechanical characteristics of a high-voltage circuit breaker. The specific method for diagnosing the mechanical characteristics of a high-voltage circuit breaker has been described in sufficient detail in the above-described embodiment of the high-voltage circuit breaker mechanical characteristics health diagnostic system and will not be repeated here.

[0121] Furthermore, the operating status includes coil current information of the opening / closing electromagnetic coil and contact stroke information of the operating actuator; the data acquisition module includes a current sensor for collecting coil current information of the opening / closing electromagnetic coil and an angular displacement sensor for collecting contact stroke information of the operating actuator.

[0122] Specifically, the processor may be a CPU, or other general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor may also be a processor that supports the Advanced RISC Machine (ARM) architecture.

[0123] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments without inventive effort, or replace some of the technical features therein with equivalents. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for diagnosing the mechanical characteristics of a high-voltage circuit breaker, characterized in that: The method includes: S1. Obtaining the operating status information of the high-voltage circuit breaker during the opening / closing process; S2. Inputting the operating status information into a pre-trained fault diagnosis model to obtain a fault status type of the mechanical characteristics of the high-voltage circuit breaker; The fault diagnosis model includes a strong classifier, which is a classifier obtained by combining at least two weak classifiers through the AdaBoost algorithm.

2. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 1, characterized in that: The weak classifier is a random forest model.

3. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 1, characterized in that: The operating status includes coil current information of the opening / closing electromagnetic coil and contact stroke information of the operating actuator.

4. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 3, characterized in that: The contact stroke information includes a time-frequency representation of a stroke characteristic obtained by Hilbert-Huang transforming a stroke-time characteristic waveform.

5. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 3, characterized in that: The coil current information includes current characteristic parameters extracted from the current-time characteristic waveform, and the current characteristic parameters include the moment when the moving iron core starts to move and its corresponding current amplitude, the moment when the moving iron core moves to its position and its corresponding current amplitude, the moment when the moving contact starts to move and its corresponding current amplitude, the moment when the coil circuit power supply is cut off and its corresponding current amplitude, and the moment when the coil current disappears after the coil circuit power supply is cut off.

6. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 3, 4 or 5, characterized in that: The contact stroke information includes stroke characteristic parameters extracted from the stroke-time characteristic waveform, and the stroke characteristic parameters include the moment when the contacts are just opened / closed and the corresponding stroke and opening / closing speed.

7. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 5, characterized in that: The current-time characteristic waveform is obtained by fitting the original current-time signal of the opening / closing electromagnetic coil collected by the least squares method into a polynomial curve, and obtaining the current-time characteristic waveform according to the fitted polynomial curve.

8. The high-voltage circuit breaker mechanical characteristic health diagnosis method according to claim 4, characterized in that: The stroke-time characteristic waveform is obtained by fitting the collected original stroke-time signal of the contact into a polynomial curve, and obtaining the stroke-time characteristic waveform according to the fitted polynomial curve; The method of fitting the collected original travel-time signal of the contact into a polynomial curve specifically includes: dividing the collected original travel-time signal into at least two intervals in time; fitting a polynomial curve in each interval respectively; fitting or interpolating at the segmentation point between the two intervals to make the fitting curve continuous; and combining the fitting curves in each interval to obtain a polynomial curve corresponding to the travel characteristic.

9. A high-voltage circuit breaker mechanical characteristic health diagnosis system, comprising a processor, characterized in that: The processor is configured to execute a computer program to implement the steps of the high-voltage circuit breaker mechanical characteristic health diagnosis method according to any one of claims 1 to 8.

10. A high-voltage circuit breaker, comprising a health diagnosis module and a data acquisition module for obtaining operating status information of the high-voltage circuit breaker during opening / closing, wherein the health diagnosis module comprises a processor, characterized in that: The processor is configured to execute a computer program to implement the steps of the high-voltage circuit breaker mechanical characteristic health diagnosis method according to any one of claims 1 to 8.

Images