Fault detection method and device of vacuum circuit breaker and computer equipment

Abstract

The invention relates to a fault detection method and device of a vacuum circuit breaker, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps that in the working process of the vacuum circuit breaker, initial sensor data of a sensor are obtained, signal noise reduction is conducted on the initial sensor data to obtain target sensor data, the sensor is used for monitoring the working state of the vacuum circuit breaker, feature extraction is conducted on the target sensor data to obtain feature data, and the feature data are sent to a server; and constructing a feature sequence based on the feature data, and predicting the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result. By adopting the method, the timeliness of fault detection of the vacuum circuit breaker can be improved.

Description

Technical Field

[0001] The present application relates to the field of electric power technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for detecting a vacuum circuit breaker fault. Background Art

[0002] With the development of power technology, various types of power equipment have gradually increased and improved. Various types of power equipment can cooperate in the production and transmission of electric energy. Vacuum circuit breakers, as a type of power equipment, can safely and efficiently open and close circuits. In order to ensure the safety of vacuum circuit breakers, it is necessary to perform fault detection on vacuum circuit breakers.

[0003] In the prior art, a maintenance time is usually preset, and the vacuum circuit breaker is shut down for maintenance after the maintenance time arrives, which causes a problem of timeliness in fault detection of the vacuum circuit breaker. Summary of the Invention

[0004] Based on this, it is necessary to provide a vacuum circuit breaker fault detection method, device, computer equipment, computer-readable storage medium and computer program product that can improve the timeliness of vacuum circuit breaker fault detection in order to address the above technical problems.

[0005] In a first aspect, the present application provides a fault detection method for a vacuum circuit breaker, comprising: during the operation of the vacuum circuit breaker, obtaining initial sensor data of a sensor, and performing signal noise reduction on the initial sensor data to obtain target sensor data, wherein the sensor is used to monitor the operating status of the vacuum circuit breaker; performing feature extraction on the target sensor data to obtain feature data, and constructing a feature sequence based on the feature data; predicting the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result.

[0006] In the second aspect, the present application also provides a fault detection device for a vacuum circuit breaker, including: a sensor data acquisition module, used to acquire the initial sensor data of the sensor during the operation of the vacuum circuit breaker, and perform signal noise reduction on the initial sensor data to obtain target sensor data, and the sensor is used to monitor the working status of the vacuum circuit breaker; a feature sequence construction module, used to perform feature extraction on the target sensor data to obtain feature data, and construct a feature sequence based on the feature data; a fault state prediction module, used to predict the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result.

[0007] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described in the first aspect when executing the computer program.

[0008] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described in the first aspect when executed by a processor.

[0009] In a fifth aspect, the present application also provides a computer program product, comprising a computer program, which implements the steps of the method described in the first aspect when executed by a processor.

[0010] The above-mentioned vacuum circuit breaker fault detection method, device, computer equipment, computer-readable storage medium and computer program product, by setting a sensor on the vacuum circuit breaker, obtains the working status of the vacuum circuit breaker in real time through the sensor to monitor the working status of the vacuum circuit breaker; during the operation of the vacuum circuit breaker, obtains initial sensor data of the sensor and performs signal noise reduction on the sensor data to obtain target sensor data to predict the fault status of the vacuum circuit breaker; extracts features from the target sensor data to obtain feature data and constructs a feature sequence based on the feature data; represents the working status of the vacuum circuit breaker by the feature sequence; predicts the fault status of the vacuum circuit breaker based on the feature sequence to obtain a fault status prediction result; by obtaining the working status of the vacuum circuit breaker in real time and performing prediction, it is avoided to perform scheduled power outage maintenance on the vacuum circuit breaker when the fault condition of the vacuum circuit breaker is unknown, thereby improving the timeliness of fault detection of the vacuum circuit breaker. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following briefly introduces the drawings required for use in the embodiments of the present application or related technical descriptions. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A diagram illustrating an application environment of a vacuum circuit breaker fault detection method according to an embodiment;

[0013] Figure 2 1 is a flow chart of a method for detecting a fault of a vacuum circuit breaker according to an embodiment;

[0014] Figure 3 1 is a flow chart of a signal noise reduction operation in one embodiment;

[0015] Figure 4 is a schematic diagram of wavelet decomposition in one embodiment;

[0016] Figure 5 This is a flowchart of a first target feature extraction operation according to an embodiment;

[0017] Figure 6 This is a flow chart of the second target feature extraction operation according to an embodiment;

[0018] Figure 7 A schematic diagram of a time curve for opening and closing the circuit breaker in one embodiment;

[0019] Figure 8 This is a flow chart of the third target feature extraction operation according to an embodiment;

[0020] Figure 9 A schematic diagram of a current waveform of an opening and closing release during opening and closing operation in one embodiment;

[0021] Figure 10 This is a flowchart of the fourth target feature extraction operation according to an embodiment;

[0022] Figure 11 A schematic diagram of a current waveform of an energy storage motor during opening and closing operations in one embodiment;

[0023] Figure 12 A schematic diagram of a flow chart of a fault state prediction operation according to an embodiment;

[0024] Figure 13 A schematic flow chart of another fault state prediction operation according to an embodiment;

[0025] Figure 14 A schematic diagram of the network structure of a probabilistic neural network in one embodiment;

[0026] Figure 15 is a flow chart of a vacuum circuit breaker fault detection method according to another embodiment;

[0027] Figure 16 FIG. 1 is another application environment diagram of a vacuum circuit breaker fault detection method according to an embodiment;

[0028] Figure 17 is a structural block diagram of a fault detection device for a vacuum circuit breaker in one embodiment;

[0029] Figure 18 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0031] It should be noted that the terms "first", "second", etc. used in this application may be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "including" and "having" used in this application and any variations thereof are intended to cover non-exclusive inclusions. The term "plurality" used in this application refers to two or more. The term "and / or" used in this application refers to one of the solutions or any combination of multiple solutions.

[0032] The vacuum circuit breaker fault detection method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown, the application environment at least includes: a server 101 , a vacuum circuit breaker 102 , and a sensor group 103 configured on the vacuum circuit breaker 102 .

[0033] The server 101 is configured to obtain initial sensor data from the sensor group 103, perform signal noise reduction on the initial sensor data to obtain target sensor data, perform feature extraction on the target sensor data to obtain feature data, construct a feature sequence based on the feature data, and predict the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result. The server 101 may be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0034] The vacuum circuit breaker 102 is used to operate circuit switches in the distribution network. The vacuum circuit breaker 102 can be applied to high-voltage power transmission environments. The vacuum circuit breaker 102 may include a switching release, an energy storage motor, and an insulating pull rod; the sensor group 103 is used to monitor the working status of the vacuum circuit breaker 102. The sensor group 103 may include a Hall current sensor, a vibration sensor, an angular displacement sensor, and a pressure sensor.

[0035] In an exemplary embodiment, Figure 2 As shown, a vacuum circuit breaker fault detection method is provided, which is applied to Figure 1 The server in FIG. 1 is used as an example to illustrate the method, including the following steps 201 to 203. Among them:

[0036] Step 201 : During the operation of the vacuum circuit breaker, initial sensor data of a sensor is acquired, and signal noise reduction is performed on the initial sensor data to obtain target sensor data.

[0037] In actual applications, vacuum circuit breakers, as common circuit-breaking devices in power systems, are usually subject to post-maintenance, planned maintenance, or condition-based maintenance, and it is impossible to detect the fault status of the vacuum circuit breaker in a timely manner. To address this, the present application configures sensors on the vacuum circuit breaker to monitor the working status of the vacuum circuit breaker in real time. When the vacuum circuit breaker is working, that is, when performing switching operations, the working data of each component of the vacuum circuit breaker is collected to predict the fault status, thereby realizing real-time fault status prediction for the vacuum circuit breaker.

[0038] Among them, the sensor is used to monitor the working status of the vacuum circuit breaker; further, the vacuum circuit breaker may generate large mechanical vibrations when working, which will cause deviations in the data collected by the sensor. To this end, non-contact sensors or sensors with stable physical states can be used to collect data. The circuit data of the vacuum circuit breaker can be collected through a Hall current sensor, and the displacement data of the vacuum circuit breaker can be collected through an angular displacement sensor based on magnetic field changes. For mechanical data such as pressure and vibration, pressure sensors and vibration sensors can be used. By monitoring the working status of the vacuum circuit breaker through non-contact sensors, the interference of mechanical characteristics on sensor data collection is avoided, and the reliability and effectiveness of sensor data are improved.

[0039] During the implementation process, when the vacuum circuit breaker is working, the sensor can collect data of the vacuum circuit breaker in real time and send it to the server. The server receives the initial sensor data sent by the sensor and performs signal noise reduction processing on the initial sensor data to obtain target sensor data.

[0040] During the initial sensor data acquisition process, the sensors can be configured in the form of sensor groups, and the sensors can also be configured with a sensor data integration system. The sensor data integration system integrates the types of each sensor and the data collected by each sensor and sends them to the server; optionally, the initial sensor data includes sensor data of multiple sensors, and the sensor data includes sensor type, data collection time and collection data.

[0041] During the signal noise reduction process, the server can perform data noise reduction on the initial sensor data through wavelet transform, Fourier transform, or adaptive filtering. In addition, data noise reduction can also be performed through hardware means. A filter can be set on the sensor side to remove noise.

[0042] Step 202 : extracting features from the target sensor data to obtain feature data, and constructing a feature sequence based on the feature data.

[0043] The present application processes sensor data by characterizing the sensor data, and characterizes the working status of the vacuum circuit breaker through the characterized feature sequence, making the data in the data processing process more concise, avoiding the separate detection of various sensor data, and improving the fault detection efficiency.

[0044] During the implementation process, the server first extracts features from the target sensor data corresponding to each sensor to obtain feature data, and then constructs a feature sequence based on the feature data. The target sensor data includes at least one sensor data; the sensor data corresponds to the sensor type.

[0045] During the feature extraction process, the target sensor data may be converted into a feature vector, which represents the operating state of the vacuum circuit breaker monitored by the sensor in the form of the feature vector. Alternatively, the target sensor data may be converted into a feature value, which represents the operating state of the vacuum circuit breaker monitored by the sensor in the form of the feature value. Optionally, the feature data includes a feature vector and / or a feature value.

[0046] During the feature sequence construction process, the order of feature data in the feature sequence can be preset, and the feature data can be sorted according to the order of the feature data and then combined into a feature sequence. For example, the feature data can be sorted according to the preset sensor type. In addition, the feature data can also be constructed into a feature matrix, and the feature matrix can be represented as a feature sequence. The feature data in the feature matrix can also be sorted according to the predicted order.

[0047] For example, the feature sequence can be represented as X obs (t) = [x1, x2, x3…x n ] T , where x1, x2, x3 and x n It is the feature data after feature extraction of the target sensor data; the feature data corresponds to the type of sensor.

[0048] For another example, the feature sequence can be represented as an n×m feature matrix, where each row is feature data corresponding to a type of sensor.

[0049] Step 203: predicting the fault state of the vacuum circuit breaker based on the characteristic sequence to obtain a fault state prediction result.

[0050] During the implementation process, the server predicts the fault state of the vacuum circuit breaker according to the feature sequence, that is, determines whether the working state of the vacuum circuit breaker is normal according to the feature sequence, and the prediction result is the fault state prediction result.

[0051] During the execution process, fault status prediction can be performed in a variety of ways. The server can determine the feature sequence and / or feature matrix under normal conditions in advance based on the sensor data of the vacuum circuit breaker under normal conditions, and perform feature comparison between the feature sequence and / or feature matrix under normal conditions and the feature matrix constructed based on the feature data. If the difference is within the threshold range, it indicates that the vacuum circuit breaker is not faulty; if the difference is outside the threshold range, it indicates that the vacuum circuit breaker is faulty. Alternatively, fault detection processing can be performed through a pre-trained model. The model can perform multi-layer mapping of the feature sequence through the neural network inside the model to output the prediction result. The prediction result is the fault status prediction result of the vacuum circuit breaker.

[0052] In the above-mentioned fault detection method for vacuum circuit breaker, the working state of the vacuum circuit breaker is monitored in real time by sensors, and initial sensor data for the vacuum circuit breaker is collected during the working process of the vacuum circuit breaker, thereby improving the timeliness of the fault detection of the vacuum circuit breaker. The server performs signal noise reduction on the initial sensor data to obtain target sensor data, and improves the accuracy of the fault state prediction result by data noise reduction. The server performs feature extraction on the target sensor data to obtain feature data, and constructs a feature sequence based on the feature data. The working state of the vacuum circuit breaker is characterized by the feature sequence, avoiding direct processing of multiple sensor data, and improving the efficiency of fault state prediction. The fault state of the vacuum circuit breaker is predicted based on the feature sequence to obtain a fault detection result. Through real-time prediction and accurate judgment, the timeliness of the fault state prediction of the vacuum circuit breaker is improved while improving the accuracy and efficiency of the fault state prediction.

[0053] Based on the above exemplary embodiment, the following provides a vacuum circuit breaker fault detection method in one or more exemplary embodiments, wherein the method is applied to Figure 1 The server in is used as an example to illustrate the details, including the following.

[0054] In practical applications, due to environmental interference, the data measured by the sensor usually carries noise interference and / or random clutter. To address this, the present application performs signal denoising on the initial sensor data to obtain target sensor data, thereby reducing noise interference while enhancing effective characteristic signals. In an optional implementation scheme provided by the present application, Figure 3 As shown, signal noise reduction is performed, including steps 301 to 302:

[0055] Step 301 : performing wavelet decomposition on initial sensor data according to a wavelet function to obtain first sensor data.

[0056] This application uses wavelet analysis to perform data noise reduction, and improves the data noise reduction effect through the good time-frequency localization and multi-resolution analysis capabilities of wavelet analysis.

[0057] During implementation, the server scales and translates the initial sensor data using the translation factor and the scaling factor in the wavelet function to obtain the first sensor data.

[0058] For example, the initial sensor data can be subjected to wavelet analysis using the wavelet mother function Ψ, as shown in formula (1):

[0059] Formula (1);

[0060] Among them, a is the scaling factor and b is the translation factor.

[0061] During the wavelet decomposition process, since the sensor data collected by the sensor are all discrete signals, the translation factor and scaling factor of the wavelet function can also be discretized, and the initial sensor data can be wavelet decomposition based on the discretized translation factor and scaling factor. The scaling factor can be discretized in a power series, and the translation factor can be uniformly discretized. That is: the initial sensor data is wavelet decomposed according to the discretized wavelet function to obtain the first sensor data; the discretized wavelet function includes the discretized scaling factor and the discretized translation factor.

[0062] For example, if the scaling factor is discretized into a power series and the translation factor is discretized uniformly, the wavelet discretization basis function is as shown in formula (2):

[0063] Formula (2);

[0064] Considering the initial sensor data as a square integrable function f(n), the corresponding discrete wavelet changes are shown in formula (3):

[0065] Formula (3).

[0066] Step 302 : reconstruct the first sensor data according to the control function to obtain target sensor data.

[0067] During implementation, after the wavelet decomposition is completed, the server may further remove the decomposed noise and reconstruct the target sensor data based on the removed sensor data.

[0068] During the data reconstruction process, the noise contained in the sensor data can be denoised, and the wavelet coefficients can be processed by adjusting the threshold function and the threshold strategy function. Finally, the signal is reconstructed to obtain the target sensor data after denoising the initial sensor data.

[0069] For example, Figure 4As shown, wavelet basis functions with good orthogonality, compact support and symmetry and high vanishing matrix order are selected to perform multi-layer decomposition of the signal (such as three-layer decomposition, the decomposition process is shown in the figure). The noise part is usually contained in D1, DA2, and DAA3. The wavelet coefficients are processed by adjusting the threshold function and the threshold strategy function, and finally the signal is reconstructed to achieve the purpose of signal denoising.

[0070] Furthermore, after wavelet denoising, a wavelet denoising quality value may be calculated based on the root mean square error (RMES) and / or the signal-to-noise ratio (SNR). If the denoising quality value is less than a threshold, secondary denoising processing is performed.

[0071] For example, after wavelet denoising, the quality of wavelet denoising is evaluated based on the root mean square error and signal-to-noise ratio. The root mean square error refers to the mean square error between the denoised signal and the original signal, reflecting the difference between the original signal and the denoised signal. The smaller its value, the better the denoising effect. The signal-to-noise ratio refers to the ratio of the original signal energy to the noise energy. It is generally believed that the larger the signal-to-noise ratio, the better the denoising effect. The calculation process of the root mean square error is shown in formula (4), and the calculation process of the signal-to-noise ratio is shown in formula (5):

[0072] Formula (4);

[0073] Formula (5).

[0074] In terms of the final denoising effect, the sym5 wavelet basis, heuristic threshold, and three-layer decomposition denoising can be selected for the trip unit current signal. The RMSE is 0.0051, the SNR is 39.1385, and the denoising effect is the best. The sym5 wavelet basis, unbiased likelihood estimation, and three-layer decomposition denoising can be selected for the vibration signal. The RMSE is 4.9721, the SNR is 5.8178, and the denoising effect is the best. The sym5 wavelet basis, unbiased likelihood estimation, and three-layer decomposition denoising can be selected for the pressure signal. The RMSE is 3.8818, the SNR is 24.2675, and the denoising effect is the best.

[0075] This application uses wavelet functions to perform signal noise reduction on initial sensor data, thereby improving the noise reduction quality of initial sensor data in time domain and multi-resolution scenarios. By performing signal noise reduction on initial sensor data, the influence of environmental noise on the fault state prediction of the vacuum circuit breaker is removed, thereby improving the accuracy of the fault state prediction of the vacuum circuit breaker.

[0076] In actual scenarios, the types of sensors configured for vacuum circuit breakers can be various to accurately detect the working status of vacuum circuit breakers from multiple dimensions. The types of sensors can be vibration sensors, angular displacement sensors, pressure sensors and / or Hall current sensors to monitor different components of the vacuum circuit breaker. Optionally, the sensors include vibration sensors, angular displacement sensors, pressure sensors and / or Hall current sensors. Optionally, the target sensor data includes: target vibration data, target angular displacement data, target pressure data, first Hall current data and / or second Hall current data. Correspondingly, the feature data includes: vibration feature data, mechanical feature data, first current feature value, second current feature value, tripping time feature value, starting current feature value, working current feature value, energy storage time feature value, first sensitive feature, second sensitive feature and third sensitive feature. The following describes the feature extraction process of each sensor respectively.

[0077] (1) Vibration sensor

[0078] In actual scenarios, the vibration signal generated during the operation of the vacuum circuit breaker is a very important signal in the transmission module, and the signal contains a large amount of status information of the vacuum circuit breaker. To this end, the present application analyzes the vibration signal through time domain statistics, the complete empirical mode decomposition method with adaptive noise and autocorrelation (CEEMDANA), and the wavelet packet energy spectrum entropy method to extract the vibration characteristics of the transmission module. The target vibration data can be subjected to time domain feature extraction to obtain time domain vibration characteristics, the time domain vibration characteristics can be subjected to feature denoising to obtain the first time domain vibration characteristics, and the first time domain vibration characteristics can be subjected to feature change to obtain vibration feature data. In the first optional implementation manner provided by the present application, if Figure 5 As shown, feature extraction is performed on target sensor data, including steps 501 to 503.

[0079] Step 501: extract time domain features from target vibration data to obtain time domain vibration features.

[0080] During the implementation process, the server extracts features of the target vibration data in the time domain dimension to obtain time domain vibration features.

[0081] During the execution process, when a fault occurs in the vacuum circuit breaker transmission module, the most intuitive manifestation is the change in the time domain signal amplitude and probability distribution. Considering the influence of the time domain characteristic parameters, the root mean square value (RMS), skewness index (SKE) and kurtosis index (KUR) can be selected as the time domain characteristics of the vibration signal to improve the accuracy of fault state prediction.

[0082] For example, the calculation of the root mean square value can be shown in formula (6), the calculation of the skewness index can be shown in formula (7), and the calculation of the kurtosis index can be shown in formula (8):

[0083] Formula (6);

[0084] Formula (7);

[0085] Formula (8).

[0086] Step 502 : performing feature cyclic denoising on the time-domain vibration feature according to a preset number of cycles to obtain a first time-domain vibration feature after feature cyclic denoising.

[0087] During the implementation process, the time domain characteristics can roughly reflect the working state of the vacuum circuit breaker, and can be further decomposed to remove noise. The time domain vibration characteristics are cycled through a preset number of cycles to obtain the first time domain vibration characteristics.

[0088] During the execution process, the Hilbert-Huang transform (HHT) can be used to perform feature denoising, and the number of loop iterations can be set to perform feature decomposition, and the decomposed noise can be removed to obtain the first time domain vibration feature.

[0089] For example, at each stage of signal decomposition, adaptive Gaussian white noise is inserted, and each IMF component is obtained by the difference between the local mean of the signal and the mixed signal. The Spearman correlation coefficient SR is introduced to judge the IMF component, and the IMF components that are strongly correlated with the original vibration signal and have actual physical significance are screened out. Not only does it not need to pre-set the number of components, but it can also eliminate false modal components. The first IMF component can be calculated first, and then the SR of the IMF component is calculated. The algorithm ends when the SR is less than 0.2. Then the second IMF component is calculated, and the kth IMF component is calculated. The above operation is repeated. If the SR of IMF(N) is less than 0.2 in the Nth loop iteration, the decomposition is stopped, and the final signal to be decomposed is as shown in formula (9):

[0090] Formula (9).

[0091] Step 503: Perform feature transformation on the first time-domain vibration feature to obtain vibration feature data.

[0092] During the implementation process, when the vacuum circuit breaker is in operation, the vibration generated by the impact is mainly a high-frequency signal, and under different states, the change in impact intensity affects the relative energy distribution in the high-frequency band. In view of this, the first time domain vibration characteristics can be transformed into the frequency domain to obtain vibration characteristic data.

[0093] During execution, the first time-domain vibration feature can be transformed using the wavelet packet energy spectrum entropy method to obtain vibration feature data. Entropy refers to the average amount of information after excluding redundant information. It represents the complexity and uncertainty of the average amount of information per character. Furthermore, during the operation of a vacuum circuit breaker, the vibration generated by the impact is primarily a high-frequency signal. Changes in impact intensity under different conditions affect the relative energy distribution in the high-frequency band. Therefore, the high-frequency relative energy of the vibration signal is very important characteristic information. The high-frequency modal (Hilbert) marginal spectrum energy is the square integral of the Hilbert marginal spectrum across frequency bands. Its practical physical significance is to weaken the noise mixed in the vibration signal and amplify the actual frequency components of the vibration signal, which can better reflect the state of the vibration signal.

[0094] For example, using a wavelet packet decomposition algorithm to decompose the vibration signals of a vacuum circuit breaker under normal conditions, mechanical interference, trip spring fatigue, buffer fault, and overtravel fault conditions, we can obtain the transformation coefficients of different components. The wavelet packet transformation coefficients can be roughly viewed as the distribution of the signal in a specific frequency band. Under different fault conditions, the transformation coefficients of each node are significantly different. The wavelet energy spectrum entropy of each scale under different fault conditions of the vacuum circuit breaker is extracted as a vibration feature.

[0095] This application reflects the state of the vibration signal by extracting features from the vibration data of the vacuum circuit breaker, and characterizes the mechanical state, trip spring state, buffer state and other working states of the vacuum circuit breaker through vibration feature data, thereby improving the reliability and comprehensiveness of the fault state prediction of the vacuum circuit breaker.

[0096] (2) Angular displacement sensor and pressure sensor

[0097] In actual scenarios, the stroke-time characteristic curve is also an important online monitoring content in the vacuum circuit breaker transmission module. From the stroke-time characteristic curve, important mechanical characteristic parameters of the vacuum circuit breaker can be obtained, such as opening distance, stroke, overstroke, opening / closing time, opening / closing speed and opening rebound amplitude. Among them, opening / closing speed, opening / closing time and overstroke are particularly critical mechanical characteristic parameters of the vacuum circuit breaker. In view of this, the displacement data of the vacuum circuit breaker contact can be calculated through the target angular displacement data, and the opening and closing moments of the vacuum circuit breaker can be determined through the target pressure data. The mechanical characteristic data of the vacuum circuit breaker can be calculated based on the displacement data, opening moment and closing moment. In the second optional implementation manner provided by the present application, if Figure 6 As shown, feature extraction is performed on target sensor data, including steps 601 to 603.

[0098] Step 601: Determine displacement data of contacts included in a vacuum circuit breaker according to target angular displacement data.

[0099] During the implementation process, when the vacuum circuit breaker is in operation, the mechanism drives the insulating pull rod connected to the main shaft to move linearly through the rotation of the main shaft. In response to this, the server calculates the displacement data of the vacuum circuit breaker contacts based on the target angular displacement data and the travel time characteristic curve of the vacuum circuit breaker.

[0100] Step 602: performing wavelet decomposition on the target pressure data to obtain decomposed pressure data, and performing data transformation on the decomposed pressure data to obtain the closing time and opening time of the vacuum circuit breaker.

[0101] During the implementation process, the pressure sensor is installed at the connection between the four-link and the insulating link. The insulating pull rod pressure of the vacuum circuit breaker is indirectly equal to the contact pressure of the vacuum circuit breaker. The mutation information of the target pressure data is often reflected in the envelope of the target pressure data. The local extreme point of the signal envelope corresponds to the mutation point of the target pressure data. In view of this, the server can perform wavelet decomposition on the envelope of the target pressure data through wavelet transformation, and perform secondary high-frequency modal transformation on the decomposed data to obtain the closing and opening times of the vacuum circuit breaker.

[0102] Step 603: Calculate mechanical characteristic data of the vacuum circuit breaker according to the closing moment, the opening moment and the displacement data.

[0103] Among them, the characteristic data includes mechanical characteristic data, and the mechanical characteristic data includes at least one of the opening distance characteristic value, stroke characteristic value, overtravel characteristic value, opening and closing time characteristic value, opening and closing speed characteristic value and opening rebound characteristic value of the vacuum circuit breaker. Overtravel refers to the distance that the moving contact continues to move after contacting the static contact during the operation of the circuit breaker. When the vacuum circuit breaker performs a closing operation, once the moving contact contacts the static contact, in order to ensure a good electrical connection and sufficient contact pressure, the moving contact will continue to move forward for a distance. This additional movement distance is called overtravel. The opening distance refers to the minimum distance between the moving contact and the static contact when the circuit breaker is in the opening state. This distance is an important parameter to ensure that the circuit breaker can reliably cut off the circuit under high voltage conditions without arc reignition.

[0104] During implementation, if the overtravel is too small, it will affect the breaking capacity and dynamic stability of the vacuum circuit breaker. If it is too large, it will increase the closing work of the operating mechanism, making the closing extremely unreliable. If the opening / closing speed is too high, it will cause increased bounce and repeated collisions between the moving and static contacts, causing arc reignition and contact wear. If the opening / closing speed is too low, the fault current cannot be cut off, causing over-tripping or explosion. To address this, the server can calculate the overtravel and opening and closing speed of the vacuum circuit breaker based on the closing time, opening time and contact displacement data. Optionally, the mechanical characteristic data includes overtravel characteristic values ​​and opening and closing speed characteristic values.

[0105] For example, Figure 7As shown in the figure, closing time refers to the moment when the vacuum circuit breaker is just closed T2 (determined by the pressure signal); opening time refers to the inherent opening time of the vacuum circuit breaker, that is, the moment when it is just opened Y2 (determined by the pressure signal); overtravel refers to the contact spring compression distance h from the moment when the moving and static contacts are just closed to the moment when they are fully closed during the closing process; H1 is the displacement corresponding to the moment when the closing stroke curve is just closed T2, that is, the opening distance of the vacuum circuit breaker; H2 is the displacement corresponding to the moment when the opening stroke curve is just opened Y2; H is the total stroke.

[0106] (3) First Hall current sensor

[0107] In actual scenarios, the opening and closing release is a key component of the vacuum circuit breaker auxiliary module. As the first-level control component, it controls the vacuum circuit breaker to complete the opening / closing operation, and reflects certain faults of the release by comparing the current waveform during the closing / opening action with the current waveform under normal conditions. For this purpose, the first Hall current data of the opening and closing release can be monitored by a non-contact Hall current sensor, and the local maximum and minimum values ​​of the opening and closing release during the working process can be traversed to determine the characteristic value of the opening and closing release working time point. In the third optional implementation manner provided by the present application, if Figure 8 As shown, feature extraction is performed on target sensor data to obtain feature data, including steps 801 to 803.

[0108] Step 801: Perform local maximum traversal on first Hall current data based on a first time period to obtain a first current characteristic value of an opening and closing release.

[0109] The opening and closing release is responsible for controlling the circuit breaker's opening (opening the circuit) and closing (closing) operations. Based on their functionality, releases can be divided into two main types: opening releases and closing releases. The first time period represents the time period when the opening and closing release's core is halfway open, including the time period when the core is halfway open. The second time period represents the time period when the opening and closing release is tripped, including the time period when the opening and closing release is tripped. The first Hall effect current sensor is used to monitor the vacuum circuit breaker's opening and closing release.

[0110] During implementation, the current change during the closing and opening action reflects the details of the iron core movement and contains time and state information. To this end, the server can collect the current of the opening and closing release through the first Hall current sensor and construct the current waveform of the opening and closing release. By traversing the local maximum value in the current waveform, the time period when the iron core of the opening and closing release is half-opened and the corresponding current value are determined. That is, step 801 can be replaced by: traversing the local maximum value of the first Hall current data based on a preset time step to obtain the first current characteristic value of the opening and closing release; the moment corresponding to the first current characteristic value is the moment when the iron core of the opening and closing release is half-opened.

[0111] Step 802: Perform a local minimum traversal on the first Hall current data based on the second time period to obtain a second current characteristic value of the opening and closing release.

[0112] During implementation, the server may perform a local minimum traversal on the waveform corresponding to the first Hall effect current data in the second time period to obtain a second current characteristic value used to characterize the tripping of the opening and closing release. Step 802 may be replaced by: performing a local minimum traversal on the first Hall effect current data based on a preset time step to obtain a second current characteristic value of the opening and closing release; the moment corresponding to the second current characteristic value is the moment when the opening and closing release is completely tripped.

[0113] Step 803 : Calculate the tripping time characteristic value of the opening and closing release according to the time corresponding to the first current characteristic value and the time corresponding to the second current characteristic value.

[0114] The characteristic data includes at least one of a first current characteristic value, a second current characteristic value, and a tripping time characteristic value.

[0115] During the implementation process, the moment corresponding to the first current characteristic value can be queried in the first Hall current data, that is, the moment when the iron core pushes up to open the opening and closing half-axis, and the moment corresponding to the second current characteristic value can be queried in the first Hall current data, that is, the moment when the opening and closing release is completely tripped. The moment when the iron core pushes up to open the opening and closing half-axis and the moment when the opening and closing release is completely tripped are the tripping time characteristic values.

[0116] For example, Figure 9 As shown in the figure, according to the change of current waveform, the trip action can be divided into four stages:

[0117] (a) t0-t1 Triggering Phase: At time t0, the opening / closing release is energized. According to the switching rule, due to the coil inductance, the coil current i cannot change suddenly. Before reaching I1 (the triggering current), the electromagnetic attraction of the moving iron core is less than the reaction force, and the moving iron core remains stationary. At this time, δ is maximum, the air gap reluctance is maximum, and L(δ) is minimum. During this phase, the current rises exponentially.

[0118] (b) t1-t2 closing movement stage: At t1, the current i reaches I1, the electromagnetic attraction of the moving iron core begins to exceed the reaction force, the moving iron core begins to move, δ begins to decrease, the magnetic resistance decreases, and the current increases. Even though a moving back EMF ivdL / d is generated that hinders the increase in current, due to the small v, the coil current continues to increase overall until it reaches I2. At this moment t2, the iron core pushes the opening / closing half-axis to open. As v continues to increase, the iron core pushes the opening / closing half-axis to open and continue to move, δ continues to decrease, and the moving back EMF becomes dominant, and the current begins to decrease. Until t3, the iron core is limited and stops moving, completing the opening / closing trip.

[0119] (c) t3-t4 electromagnetic transition stage: Although the mechanical motion process ends, due to the existence of the working power supply, the magnetic flux continues to exist and the coil current continues to increase until it reaches its respective steady-state value;

[0120] (d) t4-t5 auxiliary opening and closing stage: the closing / opening is in place, and the auxiliary switch of the vacuum circuit breaker cuts off the circuit, causing the current to decrease rapidly;

[0121] The time t0 can be used as the reference point, and the 1.5ms before and after it are regions. When this point is equal to the maximum value in the region, it is the local maximum, and when this point is equal to the minimum value in the region, it is the local minimum. If it does not exist, the reference point is increased by 1 and the iteration is continued until the local maximum point is found. By extracting the current characteristic values ​​of the opening / closing release (the current I2 when the half-shaft is opened, the current I3 when tripping, and the tripping time T), the normal voltage, high voltage start-up, low voltage start-up and stuck working states can be distinguished.

[0122] An optional implementation method provided by the present application conveniently and efficiently obtains the key time and corresponding current characteristic value of the vacuum circuit breaker during the opening and closing process by traversing the first Hall current data, thereby improving the efficiency of feature extraction. At the same time, by obtaining the working data of the trip unit, the fault detection dimension of the vacuum circuit breaker is increased, thereby improving the accuracy of the fault detection of the vacuum circuit breaker.

[0123] (4) Second Hall current sensor

[0124] In actual scenarios, the energy storage motor is also a key component of the vacuum circuit breaker auxiliary module. If there is a problem with the energy storage motor, opening / closing when the energy storage is too little or the energy storage is not completed will inevitably lead to opening / closing failure. In response to this, the present application extracts the current waveform characteristics of the energy storage motor through the motor current profile method and the segmented integration method, and can traverse the second Hall current data to obtain the starting current characteristic value, the working current characteristic value and the energy storage time characteristic value, and integrate the starting current characteristic value, the working current characteristic value and the energy storage time characteristic value based on the corresponding time to obtain sensitive characteristics. In the fourth optional implementation manner provided by the present application, if Figure 10 As shown, feature extraction is performed on target sensor data to obtain feature data, including: step 1001 to step 1002:

[0125] Step 1001: traverse the second Hall current data based on the third time period to obtain a starting current characteristic value, a working current characteristic value, and an energy storage time characteristic value.

[0126] The energy storage motor of a vacuum circuit breaker is a component used to provide energy to the circuit breaker's operating mechanism. Its primary function is to charge a spring or other energy storage device to provide the mechanical energy required for the circuit breaker's switching operation. The second Hall effect current sensor is used to monitor the energy storage motor of the vacuum circuit breaker. The third time period includes the motor startup period, the motor idling period, and the motor energy storage period.

[0127] During implementation, the working state of the energy storage motor can be divided into the motor starting stage, the idle stroke stage, and the working stage. The server can traverse the local maximum value to obtain the starting current characteristic value in the motor starting state, and obtain the starting working current value in the motor idling state by traversing the local minimum value, and calculate the average current value in the motor working state. The time from the energy storage circuit connection moment to the time corresponding to the starting current characteristic value is the motor starting time period, the time corresponding to the starting current characteristic value to the time corresponding to the starting working current value is the motor idling time period, and the time from the starting working current value to the time when the energy storage circuit is disconnected is the motor working time period; the average current value is the average current of the motor working time period. Optionally, the working current characteristic value includes the starting working current value and the working average current value.

[0128] For example, Figure 11 As shown in the figure, the waveform of the energy storage motor current can be divided into three stages:

[0129] (a) t0-t1 Motor Startup Phase: At time t0, the motor begins to be energized. The current trend satisfies Equation (5-7). The peak current, Imax, is related to the operating power supply and the armature resistance. Here, Ua is the motor armature voltage, Ra is the motor armature resistance, and TM is the electromechanical time constant.

[0130] (b) t1-t2 idle stroke stage: Due to the unequal cam radii, there is a certain amount of idle stroke between the ratchet and the cam. After the energy storage motor is started, the motor idles within the time t1-t2, causing the current to drop to Imin.

[0131] (c) t2-t3 motor working stage: When the ratchet contacts the cam, the energy storage motor starts to compress the spring to perform work. As the torque increases, the current continues to increase. After that, the motor runs at a stable current until the energy storage circuit is disconnected after the energy storage is completed.

[0132] The energy storage motor Imax represents the peak current when the motor starts, Imin represents the current when the motor starts to work, time t2 represents the beginning of energy storage of the motor, time t3 represents the end of energy storage, T1=t3-t2 represents the energy storage time of the motor, and Imean represents the average current when the energy storage motor works.

[0133] By selecting currents Imax, Imin, Imean, and time T1 as the energy storage motor's characteristic values, the feature extraction principle is similar to the trip current profile method. This feature extraction can identify different operating states of the energy storage motor, such as normal state, overvoltage state, undervoltage state, and gear wear.

[0134] Step 1002: Integrate the second Hall current data based on the motor startup time period to obtain a first sensitive feature, integrate the second Hall current data based on the motor idling time period to obtain a second sensitive feature, and integrate the second Hall current data based on the motor energy storage time period to obtain a third sensitive feature.

[0135] Optionally, the characteristic data includes at least one of a starting current characteristic value, a working current characteristic value, an energy storage time characteristic value, a first sensitive characteristic, a second sensitive characteristic, and a third sensitive characteristic.

[0136] During the implementation process, the server can integrate the current characteristic values ​​according to the three stages of the energy storage motor, and use the integration processing results as feature extraction results. It can integrate the second Hall current data based on the motor startup time period to obtain the first sensitive feature, and integrate the second Hall current data based on the motor idling time period to obtain the second sensitive feature, and integrate the second Hall current data based on the motor energy storage time period to obtain the third sensitive feature.

[0137] For example, an energy storage motor can be divided into three phases: starting, idling, and power generation. Based on this phenomenon, the energy storage current is characterized using a piecewise integration method. The curve is integrated over the time periods t0-t1, t1-t2, and t2-t3. Finally, using distance feature evaluation technology, the trip unit's I2, I3, T, sample entropy SE1, and the energy storage motor's Imax, T1, and the third-segment integral value are selected as sensitive features for the auxiliary module.

[0138] In the process of fault state prediction, a specific algorithm can be used to calculate the difference between the feature sequence and the preset feature matrix, and the fault state prediction result can be determined based on the relationship between the difference and the threshold value; in an optional embodiment provided by the present application, if Figure 12 As shown, predicting the fault state of the vacuum circuit breaker includes steps 1201 to 1203:

[0139] Step 1201: Perform matrix calculation on the characteristic sequence according to a preset memory matrix to obtain the state characteristic value of the vacuum circuit breaker.

[0140] The memory matrix is ​​constructed based on characteristic data of the vacuum circuit breaker in a normal state.

[0141] During the implementation process, a memory matrix constructed by the characteristic data of each sensor of the vacuum circuit breaker in the normal state can be calculated in advance to characterize the characteristics of the vacuum circuit breaker in the normal state. The server performs matrix calculation on the characteristic sequence according to the preset memory matrix to obtain the current state characteristic value of the vacuum circuit breaker.

[0142] During the matrix calculation process, the state characteristic value of the vacuum circuit breaker can be obtained by performing weighted calculation through the memory matrix and the characteristic data.

[0143] For example, the NSET nonlinear state estimation technique can be used to calculate the characteristic sequence X obs The expected output vector of the model can be determined as X est , that is, X est =D w =w1X obs (1) +……+w m X obs (m), that is, for any set of input observation vectors X obs , NSET will generate an m-dimensional weight vector w=[w1,w2……w m ] T , that is, the output of the NEST model is a linear combination of m observation vectors in the memory matrix D. First, the process memory matrix D needs to be determined. For vacuum circuit breakers, the process memory matrix D is composed of the eigenvalues ​​of normal fault-free samples, where each column of observation vectors represents the module eigenvalue of a normal operation of the vacuum circuit breaker. The variables in each column of data meet the requirements of simultaneity and integrity, that is, they must be sampled values ​​at the same time or in the same operation process. Therefore, the essence of constructing the process memory matrix is ​​the learning and memory process of the equipment operation characteristics. Its form is as shown in formula (10):

[0144] Formula (10);

[0145] Then, the weight vector w is determined as shown in formula (11):

[0146] Formula (11);

[0147] ε is the residual between input and output. After ε is obtained, it is minimized to obtain the weight vector w as shown in formula (12):

[0148] Formula (12).

[0149] Step 1202: Obtain the eigenvalue corresponding to the eigenvalue sequence, and detect whether the difference between the state eigenvalue and the eigenvalue is greater than a threshold.

[0150] During implementation, the distance between the characteristic state value and the characteristic value can be obtained by matrix calculation, and the distance is used as a difference to detect whether the distance is greater than a threshold.

[0151] For example, the nonlinear operator ⊗ can be used to replace the multiplication operation in ordinary matrix operations, which can be Mahalanobis distance, Euclidean distance, Chebyshev distance, etc. After consulting many documents and repeated verification, this paper finally decided to select the Euclidean distance between two vectors as the nonlinear operator of this paper, as shown in formula (13):

[0152] Formula (13);

[0153] The Euclidean distance reflects the similarity between vectors. The closer the distance, the more similar the two samples are. Conversely, the greater the difference.

[0154] Furthermore, the influence of dimension can be considered and the features can be normalized. For example, due to the influence of dimension, the features need to be normalized before calculation. The final result of the NSET algorithm for the action process or device prediction is shown in formula (14):

[0155] Formula (14);

[0156] The accuracy of the prediction is determined by the residual between the input observation and the expected output. If the device is in a normal state, w is close to 0, and the closer the expected output Xest is to the observation vector Xobs, the smaller the residual will be. At this time, the accuracy of the expected output Xest is very high. If the device suddenly enters a fault state, w changes drastically, and the expected output Xest deviates greatly from the observation vector Xobs, the residual will become larger, and the accuracy of the expected output Xest decreases. In order to diagnose vacuum circuit breaker faults in a timely manner, the threshold corresponding to each fault must be determined first. In this paper, various fault samples are input into the NSET model, and the ratio of the expected output value to the observed value is used as the similarity. The similarity represents the distance between the expected output value of the model and the observed value. The minimum similarity value is taken as the fault threshold, as shown in formula (15):

[0157] Formula (15).

[0158] Step 1203: If the difference is less than the threshold, the fault state prediction result is determined to be that the vacuum circuit breaker is not faulty; if the difference is greater than or equal to the threshold, the fault state prediction result is determined to be that the vacuum circuit breaker is faulty.

[0159] During the implementation process, the server determines the fault detection state based on the gap between the difference and the threshold, that is: the fault detection state is determined based on the calculated geometric distance and the threshold. When the difference is less than the threshold, it indicates that the gap between the current feature sequence of the vacuum circuit breaker and the memory matrix in the normal state is within the threshold range, indicating that the vacuum circuit breaker is working normally, and the fault state prediction result is determined to be that the vacuum circuit breaker is not faulty; when the difference is greater than the threshold, it indicates that the gap between the current feature sequence of the vacuum circuit breaker and the memory matrix in the normal state exceeds the threshold range, indicating that the vacuum circuit breaker has a fault, and the fault state prediction result is determined to be that the vacuum circuit breaker is faulty.

[0160] The present application improves the fault detection efficiency by performing matrix calculation on the characteristic sequence of the vacuum circuit breaker in the current state and the memory matrix of the vacuum circuit breaker in the normal state, and calculates the characteristic sequence that characterizes the overall working state of the vacuum circuit breaker. At the same time, it can comprehensively judge the working state of the vacuum circuit breaker, improve the comprehensiveness of fault detection, and thus improve the reliability of the fault detection results. The accuracy and reliability of the fault detection results are also improved through matrix calculation.

[0161] In the process of fault state prediction, a pre-trained model can also be used to predict the fault state through the neural network included in the model; in another optional embodiment provided by the present application, Figure 13 As shown, predicting the fault state of the vacuum circuit breaker includes step 1301:

[0162] Step 1301: Input the feature sequence into the fault state prediction model to perform fault state prediction and obtain a fault state prediction result.

[0163] Among them, the fault state prediction model performs feature matching through the pattern layer of the probabilistic neural network included in the fault state prediction model to obtain a feature matching result, and performs probability calculation based on the feature matching result through the summation layer of the probabilistic neural network to obtain a probability calculation result, and filters the probability calculation result through the output layer of the probabilistic neural network to obtain a fault state prediction result.

[0164] During implementation, a probabilistic neural network (PNN) can be used to build a fault state prediction model. The server can input the feature sequence into the fault state prediction model, and then perform operations through the pattern layer, summation layer, and output layer of the probabilistic neural network of the fault state prediction model to output the fault state prediction result. The fault state prediction model can be obtained through training samples or based on historical real data of vacuum circuit breakers.

[0165] For example, Figure 14As shown in Figure 1, probabilistic neural networks (PNNs) are a type of artificial neural network with a simple structure, simple training, and widespread application. They use linear learning algorithms to accomplish the work of nonlinear learning algorithms while maintaining the high precision of nonlinear algorithms. PNNs are a feedforward neural network developed from radial basis function neural networks. They integrate density function estimation and Bayesian decision theory. They do not require connection weights for feature sequences, and the hidden layer is directly constructed from given samples, resulting in strong classification capabilities. Suitable for fault diagnosis, its basic structure consists of an input layer, a pattern layer, a summation layer, and an output layer. The input layer accepts feature vectors from a feature sequence and passes them to all pattern units. The number of neurons in the input layer equals the feature dimension of the feature sequence. The number of neurons in the pattern layer is the sum of the feature sequences of each state. A nonlinear operator is used to calculate the matching relationship between the input feature vector and each state in the feature sequence. The summation layer simply accumulates the probability of belonging to a particular fault state to obtain the maximum probability that the input sample belongs to that fault state. Each state has only one summation layer unit, which is connected to the pattern layer belonging to that state. The neurons in the output layer are competitive neurons, each corresponding to a fault state. The number of neurons in the output layer is equal to the number of fault state types. It accepts various probability density functions output by the summation layer. The neuron with the highest probability density outputs 1, corresponding to the fault state to be identified, while all other neurons output 0. This algorithm always converges to a Bayesian optimal solution, with simple convergence, high stability, and strong sample addition capabilities, which can tolerate individual erroneous samples. The PNN method has the best diagnostic effect in transmission module fault diagnosis, with a detection rate of 97.59% and a false detection rate of 2.41%. Therefore, PNN is selected as the optimal fault diagnosis algorithm for the transmission module.

[0166] An optional implementation provided by the present application is to perform prediction processing on the feature sequence through a pre-trained fault state prediction model. The pre-trained fault state prediction model improves the efficiency of fault state prediction. At the same time, the pre-trained fault state prediction model improves the accuracy of fault state prediction.

[0167] In one embodiment, see Figure 15 , which shows a flow chart of a vacuum circuit breaker fault detection method provided by an embodiment of the present application, the vacuum circuit breaker fault detection method can be applied to Figure 1 As shown in the server. Figure 15 As shown, the vacuum circuit breaker fault detection method may include the following steps:

[0168] Step 1501: During the operation of the vacuum circuit breaker, initial sensor data of the sensor is acquired, and signal noise reduction is performed on the initial sensor data to obtain target sensor data.

[0169] Optionally, the sensors include a Hall current sensor, a vibration sensor, an angular displacement sensor and a pressure sensor; the target sensing data includes the first Hall current data corresponding to the opening and closing release, the second Hall current data corresponding to the energy storage motor, the target vibration data, the target angular displacement data and the target pressure data.

[0170] Step 1502 , performing time domain feature extraction on the target vibration data to obtain a time domain vibration feature, performing feature cycle denoising on the time domain vibration feature according to a preset number of cycles to obtain a first time domain vibration feature, and performing feature transformation on the first time domain vibration feature to obtain vibration feature data.

[0171] Step 1503: Determine the displacement data of the contacts included in the vacuum circuit breaker according to the target angular displacement data, and perform wavelet decomposition on the target pressure data to obtain decomposed pressure data.

[0172] Step 1504 , performing data transformation on the decomposed pressure data to obtain the closing and opening times of the vacuum circuit breaker, and calculating the overtravel characteristic value and the closing and opening speed characteristic value of the vacuum circuit breaker based on the closing and opening times and the displacement data.

[0173] Step 1505: perform a local maximum traversal on the first Hall current data based on a preset time step to obtain a first current characteristic value of the opening and closing release; perform a local minimum traversal on the first Hall current data based on a preset time step to obtain a second current characteristic value of the opening and closing release.

[0174] Step 1506: Calculate the tripping time characteristic value of the opening and closing release according to the time corresponding to the first current characteristic value and the time corresponding to the second current characteristic value.

[0175] Optionally, the tripping time characteristic value includes a time characteristic value of the iron core of the opening and closing release pushing open the half axis and a time characteristic value of the opening and closing release tripping.

[0176] Step 1507: traverse the second Hall current data to obtain a starting current characteristic value, a working current characteristic value, and an energy storage time characteristic value, and integrate the second Hall current data based on the motor starting time period and the motor energy storage time period to obtain multiple sensitive features.

[0177] Step 1508, construct a feature sequence based on vibration characteristic data, overtravel characteristic value, opening and closing speed characteristic value, first current characteristic value, tripping time characteristic value, second current characteristic value, time characteristic value of the opening and closing releaser's core pushing open half-axis, starting current characteristic value, working current characteristic value, energy storage time characteristic value and multiple sensitive features.

[0178] Step 1509 , performing matrix calculation on the characteristic sequence according to the preset memory matrix to obtain the state characteristic value of the vacuum circuit breaker, obtaining the characteristic value corresponding to the characteristic sequence, and detecting whether the difference between the state characteristic value and the characteristic value is greater than a threshold.

[0179] Step 1510: If the difference is greater than or equal to the threshold, determine that the fault state prediction result is a vacuum circuit breaker fault.

[0180] It should be noted that any one of steps 1501 to 1510 or any combination of multiple steps can be selected from steps 201 to 203 provided in the above embodiment to form a new implementation method according to the needs of implementation deployment; and any one or multiple technical features in the technical scheme composed of steps 1501 to 1510 can also be selected from any one or multiple technical features in the technical scheme composed of steps 201 to 203 to form a new implementation method according to the needs of actual deployment, or the technical features in one or more optional implementation methods provided in one or more embodiments above can be selected to form a new implementation method, which will not be repeated here.

[0181] It should also be noted that one or more of the above-mentioned vacuum circuit breaker fault detection methods can also be applied to Figure 16 In the application environment shown, one or more of the above-mentioned vacuum circuit breaker fault detection methods are executed by a signal analysis and processing system, and initial sensor data is collected by a collector and uploaded to the signal analysis and processing system.

[0182] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily performed in sequence in the order indicated by the arrows. Unless clearly stated herein, the execution of these steps is not strictly limited in order, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of the steps or stages in other steps or other steps. It is understandable that the various steps in different embodiments can be freely combined as needed, and the various non-contradictory schemes formed by the combination all fall within the scope of protection of this application.

[0183] Based on the same inventive concept, embodiments of the present application also provide a vacuum circuit breaker fault detection device for implementing the aforementioned vacuum circuit breaker fault detection method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of one or more vacuum circuit breaker fault detection device embodiments provided below can be found in the aforementioned limitations of the vacuum circuit breaker fault detection method, and will not be further elaborated here.

[0184] In an exemplary embodiment, Figure 17 As shown, a fault detection device for a vacuum circuit breaker is provided, comprising: a sensor data acquisition module 1701, a feature sequence construction module 1702, and a fault state prediction module 1703, wherein: the sensor data acquisition module 1701 is used to acquire initial sensor data of a sensor during the operation of the vacuum circuit breaker, and perform signal noise reduction on the initial sensor data to obtain target sensor data, and the sensor is used to monitor the working state of the vacuum circuit breaker; the feature sequence construction module 1702 is used to perform feature extraction on the target sensor data to obtain feature data, and construct a feature sequence based on the feature data; the fault state prediction module 1703 is used to predict the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result.

[0185] In one embodiment, the sensor data acquisition module 1701 includes a wavelet decomposition unit and a data reconstruction unit, wherein: the wavelet decomposition unit is used to perform wavelet decomposition on the sensor data according to the wavelet function to obtain first sensor data; the data reconstruction unit is used to reconstruct the first sensor data according to the control function to obtain target sensor data.

[0186] In one embodiment, the feature sequence construction module 1702 includes a time domain feature extraction unit, a feature cycle denoising unit and a feature transformation unit, wherein: the time domain feature extraction unit block is used to perform time domain feature extraction on the target vibration data to obtain a time domain vibration feature; the feature cycle denoising unit is used to perform feature cycle denoising on the time domain vibration feature according to a preset number of cycles to obtain a first time domain vibration feature after feature cycle denoising; the feature transformation unit is used to perform feature transformation on the first time domain vibration feature to obtain vibration feature data, and the feature data includes vibration feature data.

[0187] In one embodiment, the feature sequence construction module 1702 includes a displacement data acquisition unit, a pressure data decomposition unit, and a mechanical feature data calculation unit, wherein: the displacement data acquisition unit is used to determine the displacement data of the contacts included in the vacuum circuit breaker based on the target angular displacement data; the pressure data decomposition unit is used to perform wavelet decomposition on the target pressure data to obtain decomposed pressure data, and perform data transformation on the decomposed pressure data to obtain the closing time and opening time of the vacuum circuit breaker; the mechanical feature data calculation unit is used to calculate the mechanical feature data of the vacuum circuit breaker based on the closing time, opening time and displacement data, and the feature data includes mechanical feature data.

[0188] In one embodiment, the feature sequence construction module 1702 includes a maximum value traversal unit, a minimum value traversal unit and a tripping time characteristic value calculation unit, wherein: the maximum value traversal unit is used to perform a local maximum value traversal on the first Hall current data based on the first time period to obtain the first current characteristic value of the opening and closing release; the minimum value traversal unit is used to perform a local minimum value traversal on the first Hall current data based on the second time period to obtain the second current characteristic value of the opening and closing release; the tripping time characteristic value calculation unit is used to calculate the tripping time characteristic value of the opening and closing release based on the time corresponding to the first current characteristic value and the time corresponding to the second current characteristic value.

[0189] In one embodiment, the feature sequence construction module 1702 includes a feature value calculation unit and a sensitive feature calculation unit, wherein: the feature value calculation unit includes traversing the second Hall current data based on the third time period to obtain a starting current feature value, a working current feature value and an energy storage time feature value; the sensitive feature calculation unit includes integrating the second Hall current data based on the motor starting time period to obtain a first sensitive feature, integrating the second Hall current data based on the motor idling time period to obtain a second sensitive feature, and integrating the second Hall current data based on the motor energy storage time period to obtain a third sensitive feature.

[0190] In one embodiment, the fault state prediction module 1703 includes a state eigenvalue calculation unit and a fault state judgment unit, wherein: the state eigenvalue calculation unit is used to perform matrix calculation on the feature sequence according to a preset memory matrix to obtain the state eigenvalue of the vacuum circuit breaker; the fault state judgment unit is used to obtain the eigenvalue corresponding to the feature sequence, and detect whether the difference between the state eigenvalue and the eigenvalue is greater than a threshold; if the difference is less than the threshold, it is determined that the fault state prediction result is that the vacuum circuit breaker is not faulty; if the difference is greater than or equal to the threshold, it is determined that the fault state prediction result is that the vacuum circuit breaker is faulty.

[0191] In one embodiment, the fault state prediction module 1703 includes a fault state prediction unit, which: the fault state prediction unit includes inputting a feature sequence into a fault state prediction model to perform fault state prediction to obtain a fault state prediction result, the fault state prediction model performs feature matching through the pattern layer of the probabilistic neural network included in the fault state prediction model to obtain a feature matching result, and performs probability calculation based on the feature matching result through the summation layer of the probabilistic neural network to obtain a probability calculation result, and filters the probability calculation result through the output layer of the probabilistic neural network to obtain a fault state prediction result.

[0192] Each module in the above-mentioned vacuum circuit breaker fault detection device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in the computer device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0193] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as shown in FIG. Figure 18 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O) and a communication interface. The processor, memory and input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store fault detection data of the vacuum circuit breaker. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a fault detection method for a vacuum circuit breaker is implemented.

[0194] Those skilled in the art will understand that Figure 18 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0195] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps described in the above method embodiments when executing the computer program.

[0196] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps described in the above method embodiments are implemented.

[0197] In one embodiment, a computer program product is provided, including a computer program, which implements the steps described in the above method embodiments when executed by a processor.

[0198] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0199] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments. Among them, any reference to memory, database or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), data processing logic devices based on quantum computing, artificial intelligence (AI) processors, and the like.

[0200] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0201] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A vacuum circuit breaker fault detection method, characterized in that: The method comprises: During the operation of the vacuum circuit breaker, initial sensor data of a sensor is acquired, and signal noise reduction is performed on the initial sensor data to obtain target sensor data, wherein the sensor is used to monitor the operating state of the vacuum circuit breaker; Extracting features from the target sensor data to obtain feature data, and constructing a feature sequence based on the feature data; The fault state of the vacuum circuit breaker is predicted based on the characteristic sequence to obtain a fault state prediction result.

2. The method according to claim 1, characterized in that The performing signal noise reduction on the initial sensor data to obtain target sensor data includes: Performing wavelet decomposition on the initial sensor data according to a wavelet function to obtain first sensor data; The first sensor data is reconstructed according to the control function to obtain the target sensor data.

3. The method according to claim 1, characterized in that The sensor includes a vibration sensor, and the target sensor data includes target vibration data; and extracting features from the target sensor data to obtain feature data includes: Performing time domain feature extraction on the target vibration data to obtain time domain vibration features; Performing feature cyclic denoising on the time-domain vibration feature according to a preset number of cycles to obtain a first time-domain vibration feature after feature cyclic denoising; Performing feature transformation on the first time-domain vibration feature to obtain vibration feature data, the feature data including the vibration feature data.

4. The method according to claim 1, wherein The sensor includes an angular displacement sensor and a pressure sensor, and the target sensor data includes target angular displacement data and target pressure data; and extracting features from the target sensor data to obtain feature data includes: determining displacement data of a contact included in the vacuum circuit breaker according to the target angular displacement data; Performing wavelet decomposition on the target pressure data to obtain decomposed pressure data, and performing data transformation on the decomposed pressure data to obtain the closing moment and the opening moment of the vacuum circuit breaker; The mechanical characteristic data of the vacuum circuit breaker is calculated according to the closing moment, the opening moment and the displacement data, wherein the characteristic data includes mechanical characteristic data, and the mechanical characteristic data includes at least one of the opening distance characteristic value, stroke characteristic value, overstroke characteristic value, opening and closing time characteristic value, opening and closing speed characteristic value and opening rebound characteristic value of the vacuum circuit breaker.

5. The method according to claim 1, wherein The sensor includes a first Hall current sensor, which is used to monitor the opening and closing release of the vacuum circuit breaker. The target sensor data includes first Hall current data. The feature extraction of the target sensor data to obtain feature data includes: Performing a local maximum traversal on the first Hall current data based on a first time period to obtain a first current characteristic value of the opening and closing release, wherein the first time period is used to represent a time period in which the iron core of the opening and closing release is half-opened; Performing a local minimum traversal on the first Hall current data based on a second time period to obtain a second current characteristic value of the opening and closing release, wherein the second time period is used to represent a time period in which the opening and closing release is tripped; The tripping time characteristic value of the opening and closing release is calculated according to the time corresponding to the first current characteristic value and the time corresponding to the second current characteristic value, and the characteristic data includes at least one of the first current characteristic value, the second current characteristic value and the tripping time characteristic value.

6. The method according to claim 1, characterized in that: The sensor includes a second Hall current sensor, which is used to monitor the energy storage motor of the vacuum circuit breaker. The target sensor data includes the second Hall current data. The feature extraction of the target sensor data to obtain the feature data includes: Traversing the second Hall current data based on a third time period to obtain a starting current characteristic value, a working current characteristic value, and an energy storage time characteristic value, wherein the third time period includes a motor starting time period, a motor idling time period, and a motor energy storage time period; The first sensitive feature is obtained by integrating the second Hall current data based on the motor starting time period, the second sensitive feature is obtained by integrating the second Hall current data based on the motor idling time period, and the third sensitive feature is obtained by integrating the second Hall current data based on the motor energy storage time period. The characteristic data includes at least one of the starting current characteristic value, the working current characteristic value, the energy storage time characteristic value, the first sensitive feature, the second sensitive feature and the third sensitive feature.

7. The method according to any one of claims 1 to 6, characterized in that The predicting the fault state of the vacuum circuit breaker based on the characteristic sequence to obtain a fault state prediction result includes: performing matrix calculation on the characteristic sequence according to a preset memory matrix to obtain a state characteristic value of the vacuum circuit breaker, wherein the memory matrix is ​​constructed based on characteristic data of the vacuum circuit breaker in a normal state; Obtaining a characteristic value corresponding to the characteristic sequence, and detecting whether a difference between the state characteristic value and the characteristic value is greater than a threshold; If the difference is less than the threshold, determining that the fault state prediction result is that the vacuum circuit breaker is not faulty; If the difference is greater than or equal to the threshold, it is determined that the fault state prediction result is the vacuum circuit breaker fault.

8. The method according to any one of claims 1 to 6, characterized in that The predicting the fault state of the vacuum circuit breaker based on the characteristic sequence to obtain a fault state prediction result includes: The feature sequence is input into a fault state prediction model to perform fault state prediction to obtain the fault state prediction result. The fault state prediction model performs feature matching through a pattern layer of a probabilistic neural network included in the fault state prediction model to obtain a feature matching result, and performs probability calculation based on the feature matching result through a summation layer of the probabilistic neural network to obtain a probability calculation result, and the probability calculation result is filtered through an output layer of the probabilistic neural network to obtain the fault state prediction result.

9. A vacuum circuit breaker fault detection device, characterized in that: The device comprises: A sensor data acquisition module is used to acquire initial sensor data of a sensor during the operation of the vacuum circuit breaker, and to perform signal noise reduction on the initial sensor data to obtain target sensor data. The sensor is used to monitor the operating status of the vacuum circuit breaker. A feature sequence construction module is used to extract features from the target sensor data to obtain feature data, and to construct a feature sequence based on the feature data; A fault state prediction module is used to predict the fault state of the vacuum circuit breaker based on the feature sequence to obtain a fault state prediction result.

10. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 8 are implemented.

Images