Fault monitoring system and method for vacuum circuit breaker

Abstract

The invention relates to the technical field of fault monitoring, and particularly discloses a fault monitoring system and method for a vacuum circuit breaker, and the system and method achieve the fault monitoring of the vacuum circuit breaker through the frequency and amplitude analysis of vibration data during the operation of vacuum circuit breaker equipment. Meanwhile, the operation environment data of the vacuum circuit breaker are monitored in real time, and the data processing technology based on deep learning is adopted to carry out interactive analysis on the equipment operation environment data and the currently set fault detection threshold value, so that the current fault detection sensitivity is identified, and then the fault detection threshold value is dynamically adjusted. Therefore, the accuracy and timeliness of fault detection are improved. Through the mode, the probability of misjudgment and missed judgment can be effectively reduced, intelligent monitoring of faults of the vacuum circuit breaker is realized, and the stability and the safety of a power system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault monitoring, and more particularly, to a fault monitoring system and method for a vacuum circuit breaker. BACKGROUND

[0002] As a crucial power distribution and protection device in power systems, the core function of a vacuum circuit breaker lies in rapidly breaking current through the contacts in the vacuum arc chamber in a vacuum environment and achieving arc extinction by utilizing the medium characteristics of vacuum, thereby effectively protecting electrical equipment and controlling the opening and closing state of the power system. With the rapid development of power systems and the popularization of smart grids, higher requirements are placed on the performance, reliability, and fault monitoring capabilities of vacuum circuit breakers.

[0003] Traditional fault monitoring methods for vacuum circuit breakers mainly rely on periodic mechanical inspection, hardware detection, and insulation inspection, etc. Although these methods can to some extent discover potential faults of the equipment, they usually require a lot of time and manpower, and the accuracy of judging the damage degree of the circuit breaker is not high.

[0004] Since vacuum circuit breakers produce specific vibration signals during operation, these vibration signals are closely related to the operating state of the circuit breaker. Therefore, by collecting, analyzing, and processing the vibration signals, effective monitoring of the faults of the vacuum circuit breaker can be achieved. For example, when the equipment has faults such as poor contact, loose or worn springs, etc., the vibration signals will change, showing frequency shift, amplitude increase, or waveform distortion, etc. However, existing vibration signal analysis methods are mostly based on a fixed threshold comparison method, which considers that the equipment has a fault when the amplitude of the vibration signal exceeds the preset threshold. This method lacks adaptability to changes in the operating environment. Fluctuations in different operating environments, such as temperature, humidity, etc., can all affect the operating characteristics of the vacuum circuit breaker, and the fixed fault detection threshold cannot be adjusted in real time according to these changes, which can easily lead to misjudgment or missed judgment, reducing the accuracy and timeliness of fault detection.

[0005] Therefore, an optimized fault monitoring system and method for a vacuum circuit breaker are expected. SUMMARY

[0006] In order to solve the above technical problems, the present application is proposed. Embodiments of the present application provide a fault monitoring system and method for a vacuum circuit breaker, which realizes fault monitoring of the vacuum circuit breaker by frequency and amplitude analysis on vibration data during operation of the vacuum circuit breaker device. At the same time, by monitoring the operation environment data of the vacuum circuit breaker in real time and using deep learning-based data processing technology to interactively analyze the device operation environment data and the currently set fault detection threshold, the current fault detection sensitivity is identified, and the fault detection threshold is dynamically adjusted to improve the accuracy and timeliness of fault detection. In this way, the probability of misjudgment and missed judgment can be effectively reduced, intelligent monitoring of the vacuum circuit breaker fault is realized, and the stability and safety of the power system are improved.

[0007] According to one aspect of the present application, a fault monitoring method for a vacuum circuit breaker is provided, which comprises:

[0008] collecting vibration data during operation of a vacuum circuit breaker device;

[0009] obtaining device operation environment data of the vacuum circuit breaker;

[0010] extracting a currently set fault detection threshold, wherein the fault detection threshold comprises a vibration frequency threshold and a vibration amplitude threshold;

[0011] based on the device operation environment data of the vacuum circuit breaker, performing fault detection sensitivity analysis and adjustment on the currently set fault detection threshold to obtain an adjusted fault detection threshold;

[0012] based on the adjusted fault detection threshold, performing fault detection on the vibration data to determine whether the vacuum circuit breaker has a fault.

[0013] According to another aspect of the present application, a fault monitoring system for a vacuum circuit breaker is provided, which comprises:

[0014] a vibration data collection module for collecting vibration data during operation of a vacuum circuit breaker device;

[0015] an operation environment data acquisition module for obtaining device operation environment data of the vacuum circuit breaker;

[0016] a fault detection threshold extraction module for extracting a currently set fault detection threshold, wherein the fault detection threshold comprises a vibration frequency threshold and a vibration amplitude threshold;

[0017] a fault detection threshold adjustment module for performing fault detection sensitivity analysis and adjustment on the currently set fault detection threshold based on the device operation environment data of the vacuum circuit breaker to obtain an adjusted fault detection threshold;

[0018] The vacuum circuit breaker fault judgment module is configured to perform fault detection on the vibration data based on the adjusted fault detection threshold to determine whether the vacuum circuit breaker has a fault.

[0019] Compared with the prior art, the fault monitoring system and method of the vacuum circuit breaker provided by the present application can realize fault monitoring of the vacuum circuit breaker by analyzing the frequency and amplitude of the vibration data of the vacuum circuit breaker during operation. At the same time, by monitoring the operation environment data of the vacuum circuit breaker in real time and using the deep learning-based data processing technology to interactively analyze the device operation environment data and the currently set fault detection threshold, the current fault detection sensitivity can be identified, and the fault detection threshold can be dynamically adjusted to improve the accuracy and timeliness of fault detection. In this way, the probability of misjudgment and missed judgment can be effectively reduced, intelligent monitoring of the vacuum circuit breaker fault can be realized, and the stability and safety of the power system can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description of embodiments of the present application, taken in conjunction with the accompanying drawings. The drawings provided below are for illustrating the embodiments of the present application and a part of the specification and are used to explain the present application together with the embodiments of the present application, but do not constitute limitations on the present application. In the drawings, the same reference numerals generally denote the same components or steps.

[0021] Figure 1 A flowchart of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0022] Figure 2 A flowchart of the substep S4 of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0023] Figure 3 A data flow schematic diagram of the substep S4 of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0024] Figure 4 A flowchart of the substep S43 of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0025] Figure 5 A flowchart of the substep S431 of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0026] Figure 6 A flowchart of the substep S432 of the fault monitoring method of the vacuum circuit breaker according to the embodiments of the present application.

[0027] Figure 7A block diagram of a fault monitoring system for a vacuum circuit breaker according to embodiments of the present application. DETAILED DESCRIPTION

[0028] As shown in the present application and claims, unless the context clearly indicates otherwise, the words "a," "an," "the," and / or "said" are not limited in scope to the singular, but include the plural. Generally, the terms "include," "including," and / or "comprising" are used in this document to indicate the inclusion of the recited elements, but do not preclude the presence of other elements.

[0029] Although the present application makes various references to certain modules in the system according to embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or server. The modules are merely illustrative, and different aspects of the system and method can use different modules.

[0030] Flowcharts are used in the present application to illustrate the operations performed by the system according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in sequence. Instead, various steps can be processed in reverse order or simultaneously, as desired. Other operations can also be added to or removed from these processes.

[0031] In the following, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It is obvious that the described embodiments are only a part of the embodiments of the present application, and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.

[0032] It is worth noting that in the present application, all actions of obtaining data are carried out in compliance with the corresponding data protection regulations and policies of the country where the device is located, and with the authorization given by the owner of the corresponding device.

[0033] In view of the technical problems described in the above background art, the present application proposes an optimized fault monitoring method for a vacuum circuit breaker, which realizes fault monitoring of the vacuum circuit breaker by analyzing the frequency and amplitude of vibration data of the vacuum circuit breaker device during operation. At the same time, by monitoring the operating environment data of the vacuum circuit breaker in real time, and using deep learning-based data processing technology to interactively analyze the device operating environment data and the currently set fault detection threshold, the current fault detection sensitivity is identified, and the fault detection threshold is dynamically adjusted to improve the accuracy and timeliness of fault detection. In this way, the probability of misjudgment and missed judgment can be effectively reduced, intelligent monitoring of the vacuum circuit breaker fault is realized, and the stability and safety of the power system are improved.

[0034] Figure 1A flowchart of a fault monitoring method for a vacuum circuit breaker according to an embodiment of the present application. As shown in Figure 1 the fault monitoring method for a vacuum circuit breaker, comprising steps of: S1, collecting vibration data of the vacuum circuit breaker device during operation; S2, obtaining device operating environment data of the vacuum circuit breaker; S3, extracting the current set fault detection threshold, wherein the fault detection threshold includes vibration frequency threshold and vibration amplitude threshold; S4, based on the device operating environment data of the vacuum circuit breaker, performing fault detection sensitivity analysis and adjustment on the current set fault detection threshold to obtain an adjusted fault detection threshold; S5, based on the adjusted fault detection threshold, performing fault detection on the vibration data to determine whether the vacuum circuit breaker has a fault.

[0035] In the above-mentioned fault monitoring method for a vacuum circuit breaker, the step S1, collecting vibration data of the vacuum circuit breaker device during operation. It should be understood that when the contacts, springs and other key components inside the vacuum circuit breaker wear, loosen or have other faults, the frequency and amplitude of their vibration will change accordingly. For example, poor contact of the contacts will cause additional impact vibration during closing or opening, making the vibration frequency and amplitude exceed the normal range. By collecting vibration data of the vacuum circuit breaker device during operation, key information of the device operating state can be obtained, so as to identify abnormal features of the vibration data.

[0036] Specifically, before collecting vibration data, a suitable sensor needs to be selected. Considering the working environment and characteristics of the vacuum circuit breaker, an acceleration sensor or a piezoelectric sensor with high sensitivity and wide frequency response range is usually selected. These sensors can accurately capture the small vibration changes caused by mechanical movement and convert them into electrical signals. Depending on different monitoring needs, the installation position and number of sensors may also need to be considered. For example, installing sensors at key parts of the circuit breaker such as the contact operating mechanism and spring mechanism can more comprehensively obtain device operating state information. In addition, attention should be paid to the contact method between the sensor and the measured object, and appropriate coupling agent or fastening device should be used to ensure good mechanical connection, so as to improve the quality of signal acquisition.

[0037] In the actual collection phase of vibration data, it is necessary to ensure the stability and accuracy of the collection system. Therefore, it is particularly important to build a reliable hardware platform. This platform not only includes sensors, but also covers signal conditioning circuits, data acquisition cards, and necessary power supply devices and other components. The signal conditioning circuit is used to amplify and filter the original signal to remove noise interference and enhance useful signal components; the data acquisition card is responsible for converting analog signals into digital signals for subsequent processing. In order to ensure the authenticity and integrity of the data, the characteristics of the circuit breaker vibration signal should be fully considered when setting the sampling frequency. Generally speaking, the sampling frequency should be at least twice the highest vibration frequency to meet the requirements of the Nyquist sampling theorem and avoid spectral aliasing. At the same time, adjust the sampling resolution according to actual needs to obtain sufficient accuracy to reflect the subtle changes in the vibration signal.

[0038] In addition to hardware configuration, software support is also indispensable. By writing a special data collection program, automation control and real-time monitoring of the collection process can be achieved. Such programs are usually built based on professional data collection software development kits (SDK), allowing users to flexibly set sampling parameters, trigger conditions, and data storage formats, etc. When writing the program, pay attention to optimizing the algorithm to reduce the calculation delay and ensure that the device state changes can be responded in time. In addition, considering the need for long-term monitoring, designing a reasonable data management strategy is also very important. This means not only properly saving the original data file, but also establishing a corresponding database system to facilitate future queries and analysis. For large-scale data sets, using compression techniques can save storage space and improve transmission efficiency without affecting data quality.

[0039] In actual operation, due to the influence of external environmental factors, vibration signals often contain various noise components, which poses additional challenges to data collection. Therefore, it is necessary to take effective noise reduction measures. On the one hand, physical isolation methods such as using shielded cables, sealed cases, etc. can be used to minimize electromagnetic interference and other external noise sources; on the other hand, with the help of digital signal processing technology, adaptive filters or wavelet transform algorithms can be applied during data collection to dynamically adjust filter parameters to adapt to changing noise characteristics, thereby effectively improving the signal-to-noise ratio. It is worth noting that although noise reduction processing helps to improve signal quality, excessive filtering may result in the loss of useful information, so a balance needs to be found.

[0040] For different working conditions, it is also a wise choice to implement diversified data collection schemes according to the differences in vibration characteristics. For example, under normal operating conditions, periodic monitoring can be performed with a lower sampling rate and longer time interval; while at special moments such as device startup or shutdown, the sampling frequency needs to be increased to capture the vibration information in the transient process. In addition, combined with specific application scenarios, multi-channel synchronous acquisition technology can be introduced to record the vibration of the circuit breaker from multiple angles at the same time, providing more comprehensive data support. In this way, not only can the device health status be more accurately evaluated, but also more clues can be provided for in-depth analysis of fault causes.

[0041] In the above fault monitoring method of the vacuum circuit breaker, the step S2 of acquiring the device operating environment data of the vacuum circuit breaker includes a time queue of environmental temperature and a time queue of environmental humidity. It should be understood that the operating environment of the device has an important influence on its vibration characteristics. Specifically, temperature changes will cause thermal expansion and contraction of each component of the vacuum circuit breaker. When the temperature rises, the components expand, which may cause the originally tightly connected parts to loosen and produce additional vibrations during operation; when the temperature decreases, the components contract, which may cause some components to bear additional stress and also affect the vibration characteristics. For example, the connecting bolts may loosen under the action of thermal expansion and contraction, resulting in an increase in vibration amplitude. Therefore, when setting the fault detection threshold, the influence of thermal expansion and contraction caused by temperature on vibration data should be considered. Similarly, in a high humidity environment, water may condense on the surface of the insulation components of the vacuum circuit breaker, causing a decrease in insulation performance. When the insulation performance decreases, partial discharge and other phenomena are more likely to occur, and the impact force generated by partial discharge will cause additional vibrations, making the vibration data abnormal. At this time, the fault detection threshold needs to consider the influence of humidity on insulation performance, and the threshold should be appropriately lowered to detect potential faults more timely. Therefore, by acquiring the device operating environment data of the vacuum circuit breaker, it is helpful to fully understand the external conditions of device operation, and to provide environmental factor reference for subsequent analysis of fault detection threshold, so as to avoid misjudgment or omission caused by environmental factors.

[0042] Specifically, for data collection on ambient temperature, a high-precision temperature sensor is essential. These sensors are typically based on thermocouples, thermistors, or semiconductor technology, offering fast response, high sensitivity, and good stability. The choice of sensor type depends on the specific application scenario and requirements. For example, in environments where temperature fluctuations are extremely sensitive, a thermistor with higher resolution may be a better choice; while in cases where remote signal transmission is required, a thermocouple is more suitable. The choice of installation location is also important, ensuring that the sensor can truly reflect the changes in the ambient temperature around the circuit breaker, avoiding measurement deviations caused by local heat sources or cold sources. Generally, installing the temperature sensor near the circuit breaker but not directly affected by the internal heating components can achieve more ideal results.

[0043] For the measurement of ambient humidity, capacitive humidity sensors are the first choice due to their high precision, long service life, and strong anti-pollution ability. This type of sensor determines the relative humidity in the air by detecting the change in capacitance of a humidity-sensitive material as humidity changes. When installing, the sensor should be exposed to the air while preventing direct contact with water droplets or other liquids to avoid damaging the components or causing reading errors. Like temperature sensors, humidity sensors also need to be connected to a complete acquisition system, which includes signal conditioning circuits and data recording devices. Since humidity changes are often related to temperature, in some cases, it may be considered to use a combined temperature and humidity sensor, which not only simplifies the installation process but also improves the measurement accuracy, because the interaction between the two parameters helps to obtain more accurate results.

[0044] It is worth noting that in actual operation, environmental factors can change at any time. For example, temperature fluctuations caused by seasonal changes, or sudden changes in humidity caused by weather conditions, can affect the normal operation of vacuum circuit breakers. Therefore, in addition to conventional data collection methods, some additional measures need to be taken to deal with unexpected situations. For example, in extreme weather conditions, temporary sensor nodes can be added to enhance monitoring in key areas, or adaptive algorithms can be introduced to automatically adjust the sampling frequency and threshold settings to ensure that the system is always in the best working state. In addition, considering that power facilities are usually located outdoors or semi-open spaces, protective measures are essential. Providing sensors with waterproof and dustproof covers, regularly checking line connections, and replacing aging or damaged components in a timely manner are important steps to ensure data accuracy.

[0045] In the fault monitoring method of the vacuum circuit breaker described above, the step S3 of extracting the currently set fault detection threshold value, wherein the fault detection threshold value includes a vibration frequency threshold value and a vibration amplitude threshold value. Here, the currently set fault detection threshold value is set based on past experience and fixed standards, which is a reference value for judging whether the vacuum circuit breaker has a fault.

[0046] Specifically, for the setting of the vibration frequency threshold value, it is necessary to comprehensively analyze the vibration spectrum characteristics of the vacuum circuit breaker in normal operation. The vibration frequency directly reflects the working state of the internal mechanical parts of the circuit breaker. For example, during the contact operation process, the normal switching action will produce a specific vibration frequency, and this frequency range is usually stable. By collecting a large amount of vibration data of the vacuum circuit breaker in normal working state, a reference vibration spectrum diagram can be constructed. This diagram shows the amplitude distribution at different frequencies, and some frequency bands may correspond to key mechanical movements or structural resonance points. According to this information, combined with engineering experience, the upper and lower limits of the vibration frequency within a safe range are determined as the initial threshold value.

[0047] When it comes to the setting of the vibration amplitude threshold value, it also needs to be determined based on a large amount of experimental and field data. The change of vibration amplitude is often closely related to the state of internal parts of the circuit breaker, such as poor contact, spring loosening or wear, etc. which can cause abnormal increase of vibration amplitude. Therefore, by monitoring a plurality of vacuum circuit breakers in different health states (from good to different degrees of damage), the change rule of vibration amplitude is recorded. Using statistical methods such as calculating the mean and standard deviation, the typical range of vibration amplitude under normal working conditions can be found. On this basis, considering a certain proportion of safety margin, the upper limit threshold value of vibration amplitude is set. In addition, it is also necessary to distinguish between transient vibration (such as short-time high-amplitude phenomenon during start-up or stop) and persistent abnormal vibration, so as to avoid unnecessary maintenance or shutdown caused by misjudgment.

[0048] In the specific implementation process, in order to ensure that the set fault detection threshold value has high reliability and adaptability, it is very important to adopt a multi-level data verification strategy. On the one hand, information about design parameters and recommended operating conditions can be obtained from the technical documents provided by the manufacturer as a preliminary reference. On the other hand, combined with the actual operation data in the field, especially the test results after experiencing multiple repairs or replacement of parts, the initially set threshold value is corrected.

[0049] In the fault monitoring method of the vacuum circuit breaker, the step S4 is based on the equipment operating environment data of the vacuum circuit breaker to analyze and adjust the fault detection sensitivity of the current set fault detection threshold to obtain an adjusted fault detection threshold. Specifically, due to the complex and changeable operating environment of the vacuum circuit breaker, the normal vibration characteristics of the equipment will change under different environmental temperature and humidity conditions, and the original fixed threshold may no longer be applicable. Therefore, the application further analyzes the equipment operating environment data and the current set fault detection threshold, identifies the current fault detection sensitivity, and dynamically adjusts the fault detection threshold according to the identification result. For example, in a high-temperature environment, due to the thermal expansion and contraction effect, the equipment vibration amplitude may increase, and the current set vibration amplitude threshold may be too low, which may easily lead to misjudgment. At this time, the vibration amplitude threshold can be appropriately increased to adapt to the vibration characteristics in the high-temperature environment and reduce the occurrence of misjudgment. Similarly, in an environment with high humidity, the decrease in insulation performance may cause an increase in partial discharge and abnormal vibration data, and the current set vibration frequency threshold may be too high, which may not be sensitive enough to the small vibration abnormality. At this time, the vibration frequency threshold can be appropriately reduced to improve the sensitivity to potential faults. In this way, the fault detection threshold can always match the actual operating environment of the equipment, thereby improving the accuracy and reliability of fault detection. Wherein, Figure 2 Flow chart of sub-step S4 of the fault monitoring method of the vacuum circuit breaker according to the embodiment of the application. Figure 3 Data flow diagram of sub-step S4 of the fault monitoring method of the vacuum circuit breaker according to the embodiment of the application. As shown in Figure 2 and Figure 3 The step S4 includes the steps of: S41, performing temperature-humidity time sequence interaction correlation analysis on the equipment operating environment data to obtain an equipment operating environment parameter time sequence correlation coding feature map; S42, performing low-dimensional embedding coding on the current set fault detection threshold to obtain a fault detection threshold low-dimensional embedding coding vector; S43, performing fault detection sensitivity analysis based on cross-domain joint coding on the fault detection threshold low-dimensional embedding coding vector and the equipment operating environment parameter time sequence correlation coding feature map to obtain a fault detection sensitivity hidden coding feature vector; and S44, adjusting the fault detection threshold based on the fault detection sensitivity hidden coding feature vector to obtain the adjusted fault detection threshold.

[0050] Specifically, the step S41, the device operating environment data is analyzed to obtain the device operating environment parameter time sequence correlation coding feature map. It should be understood that, since the environmental temperature and humidity do not affect the operation of the vacuum circuit breaker, there is a certain correlation and synergy between the two, and single analysis of temperature or humidity data cannot fully capture the influence of environmental factors on the device operating state. Therefore, the application captures the time sequence of environmental temperature and humidity by analyzing the time sequence of the environmental temperature and the time sequence of the environmental humidity. In specific implementation, the device operating environment data of the vacuum circuit breaker can be structured and sorted according to time sequence and parameter attributes to form a two-dimensional matrix, and the rows of the matrix represent time sequences and the columns represent different environmental parameters (such as temperature and humidity). Then, the two-dimensional matrix is analyzed by using a convolutional neural network model to extract features and analyze time sequences, so as to use the powerful feature extraction capability of the convolutional neural network model to mine the potential correlation between the environmental temperature and humidity and the change rule over time, obtain the device operating environment parameter time sequence correlation coding feature map, and realize comprehensive understanding and representation of the device operating environment.

[0051] Specifically, the step S42, the current set fault detection threshold is low-dimensional embedding coding to obtain the fault detection threshold low-dimensional embedding coding vector. Specifically, in order to realize the correlation analysis between the current set fault detection threshold and the device operating environment data, the current set fault detection threshold needs to be further vectorized. Here, the application adopts embedding coding technology, which converts text description into low-dimensional continuous vector representation by word embedding and vector space mapping of attribute description (such as vibration frequency threshold and vibration amplitude threshold) of the current set fault detection threshold, and combines with the corresponding threshold value to form a comprehensive feature representation containing threshold attribute and numerical information, that is, the fault detection threshold low-dimensional embedding coding vector. In this way, not only the original information of the fault detection threshold is retained, but also the subsequent feature correlation analysis and calculation with the device operating environment data is facilitated, which provides a basis for the dynamic adjustment of the subsequent fault detection threshold.

[0052] Specifically, the step S43, the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map are analyzed based on cross-domain joint coding to obtain a fault detection sensitivity hidden coding feature vector. It should be understood that, since the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map respectively describe the fault detection threshold and the equipment operation environment data of the vacuum circuit breaker, and belong to different data domains, have different data structures and feature representations. Therefore, in order to realize the effective association analysis of the two, the application proposes a fault detection sensitivity analysis based on cross-domain joint coding, which mines key features and feature interactions of the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map, constructs an explicit association structure between the two, and uses it as a prior information constraint condition to guide the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map to perform cross-modal fine-grained interaction based on the attention mechanism, automatically focus on the key association points between the fault detection threshold and the equipment operation environment data, and realize the deep understanding and coding representation of the fault detection sensitivity, to obtain a fault detection sensitivity hidden coding feature vector. Wherein, Figure 4 The flow chart of the sub-step S43 of the fault monitoring method of the vacuum circuit breaker according to the embodiment of the application. As shown in Figure 4 The step S43 includes the steps of: S431, extracting the key feature interaction information between the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map to obtain a fault detection threshold-equipment operation environment key feature interaction coding matrix; S432, based on the fault detection threshold-equipment operation environment key feature interaction coding matrix, guiding the fault detection threshold low-dimensional embedding coding vector and the equipment operation environment parameter time sequence association coding feature map to perform cross-modal interaction feature learning based on the attention mechanism to obtain the fault detection sensitivity hidden coding feature vector.

[0053] Figure 5 The flow chart of the sub-step S431 of the fault monitoring method of the vacuum circuit breaker according to the embodiment of the application. As shown in Figure 5As shown, the step S431 includes steps of: S4311, performing point convolution coding based key feature extraction on the fault detection threshold low-dimensional embedding coding vector to obtain a fault detection threshold key feature coding vector; S4312, extracting a device running environment parameter time sequence key feature coding vector from the device running environment parameter time sequence association coding feature map; and S4313, performing association coding on the fault detection threshold key feature coding vector and the device running environment parameter time sequence key feature coding vector to obtain the fault detection threshold-device running environment key feature interaction coding matrix.

[0054] In one specific example of the present application, the step S4311 is expressed by a formula as follows:

[0055] v c1 = Leaky ReLU{Conv 1×1 (v1) + b1}

[0056] wherein, v1 represents the fault detection threshold low-dimensional embedding coding vector, Conv 1×1 (·) represents a point convolution operation, b1 represents a bias term, Leaky ReLU(·) represents an activation function, and vc1 represents the fault detection threshold key feature coding vector.

[0057] That is, in order to highlight important features in the fault detection threshold low-dimensional embedding coding vector and reduce redundant information, the point convolution coding method is adopted in the present application to extract core feature clues from the fault detection threshold low-dimensional embedding coding vector to obtain the fault detection threshold key feature coding vector. It can be understood that the point convolution coding can focus on the key information in the vector, so that the accuracy and efficiency of fault detection can be improved in subsequent processing.

[0058] In one specific example of the present application, the step S4312 includes: first, performing feature partitioning on the device running environment parameter time sequence association coding feature map to obtain a set of device running environment parameter local time sequence association feature maps, which is expressed by a formula as follows:

[0059]

[0060] wherein, Partition(·) represents a partition function, F2 represents the device running environment parameter time sequence association coding feature map, and respectively represent the first, second, i-th and m-th device running environment parameter local time sequence association feature maps in the set of device running environment parameter local time sequence association feature maps, and m represents the length of the set of device running environment parameter local time sequence association feature maps, i.e., the number of device running environment parameter local time sequence association feature maps.

[0061] That is, by feature partitioning the device running environment parameter time sequence correlation coding feature map, the feature map is divided into a plurality of device running environment parameter local time sequence correlation feature maps, so as to capture the feature differences of different local regions for subsequent key feature extraction.

[0062] Secondly, each device running environment parameter local time sequence correlation feature map in the set of device running environment parameter local time sequence correlation feature maps is subjected to convolutional coding and global pooling processing to obtain a set of device running environment parameter local time sequence correlation feature vectors, which is expressed by the formula:

[0063]

[0064] Wherein, Conv 3×3 (·) represents a 3x3 convolutional coding operation, sigmoid(·) represents a sigmoid activation function, represents the device running environment parameter local time sequence correlation coding feature map after convolutional coding, Gloabl pool(·) represents global pooling processing, represents the corresponding device running environment parameter local time sequence correlation feature vector.

[0065] That is, by convolutional coding, the spatial dimension of each device running environment parameter local time sequence correlation feature map is compressed, and then by global pooling processing, the device running environment parameter local time sequence correlation feature map is converted into a fixed-length device running environment parameter local time sequence correlation feature vector, obtaining a set of device running environment parameter local time sequence correlation feature vectors. In this way, while preserving the important information of local features, the data dimension is reduced, facilitating subsequent calculation and analysis.

[0066] Then, based on the feature distribution of each device running environment parameter local time sequence correlation feature vector in the set of device running environment parameter local time sequence correlation feature vectors, a key feature correlation factor is calculated to obtain a set of environment parameter local time sequence correlation feature correlation factors, which is expressed by the formula:

[0067]

[0068] Wherein, max(·) and min(·) represent taking the maximum value and taking the minimum value, respectively, σ(·) represents the standard deviation of the vector, σ(·) 2 represents the variance of the vector, α and β represent different weight parameters, respectively, γ represents a regularization term, D i represents A corresponding environmental parameter local time sequence correlation feature correlation factor.

[0069] That is, by analyzing the feature distribution of each device operating environment parameter local time sequence correlation feature vector, the key feature correlation factor is calculated to measure the core degree and importance of the temperature and humidity time sequence interaction information contained in each device operating environment parameter local time sequence correlation feature vector, so as to filter out the time sequence core interaction mode of temperature and humidity as the key feature for subsequent fault detection.

[0070] Finally, based on the set of environmental parameter local time sequence correlation feature correlation factors, the set of device operating environment parameter local time sequence correlation feature vectors is aggregated to obtain the device operating environment parameter time sequence key feature encoding vector, which is expressed by the formula:

[0071]

[0072] Where exp(·) represents the exponential function operation with e as the base, v c2 represents the device operating environment parameter time sequence key feature encoding vector.

[0073] That is, by normalizing the set of environmental parameter local time sequence correlation feature correlation factors to convert them into a weight distribution, and then based on the weight distribution, the set of device operating environment parameter local time sequence correlation feature vectors is weighted and summed to aggregate them into a device operating environment parameter time sequence key feature encoding vector that can represent the device operating environment parameter time sequence key feature. In this way, by integrating local feature information and highlighting key features, the representativeness and effectiveness of the features are improved.

[0074] In one specific example of the present application, the step S4313 is expressed by the formula:

[0075]

[0076] Where, and φ(·) are both feature mapping functions for mapping heterogeneous features to a unified hidden space, such as a linear mapping function, (·) T represents the transpose of the vector, M tem represents the fault detection threshold-device operating environment key feature interaction encoding matrix.

[0077] That is, by mapping the fault detection threshold key feature encoding vector and the device operating environment parameter time sequence key feature encoding vector into the same feature space, correlation coding is performed to construct a fault detection threshold-device operating environment key feature interaction encoding matrix that can reflect the key feature interaction relationship between the fault detection threshold and the device operating environment parameter, thereby providing prior information guidance for subsequent cross-modal interaction coding, and thus guiding focusing on the key association points between the fault detection threshold and the device operating environment data.

[0078] Figure 6 The flowchart of sub-step S432 of the fault monitoring method of the vacuum circuit breaker according to the embodiment of the present application. As shown in the figure, the step S432 includes the steps of: S4321, performing feature decoupling along the channel dimension on the device operating environment parameter time sequence correlation coding feature map to obtain a set of device operating environment parameter local time sequence correlation feature matrices; S4322, taking the fault detection threshold key feature encoding vector as a query vector, taking each device operating environment parameter local time sequence correlation feature matrix in the set of device operating environment parameter local time sequence correlation feature matrices as a key matrix, and taking the fault detection threshold-device operating environment key feature interaction encoding matrix as a prior knowledge constraint matrix, performing cross-modal style constraint coding on the query vector, the set of key matrices, and the prior knowledge constraint matrix to obtain a set of fault detection threshold-device operating environment fine-grained implicit interaction encoding vectors; S4323, calculating the position-wise mean vector of the set of fault detection threshold-device operating environment fine-grained implicit interaction encoding vectors to obtain the fault detection sensitivity implicit coding feature vector. Figure 6

[0079] In one specific example of the present application, the step S4321 is expressed by the formula:

[0080] Decompose(F2)=(M1,M2,...,M n}

[0081] wherein Decompose(·) represents a decoupling function, M1, M2, and M n represent the first, second, and n-th device operating environment parameter local time sequence correlation feature matrices in the set of device operating environment parameter local time sequence correlation feature matrices.

[0082] ​That is, the application decouples the feature map of the time sequence correlation of the device operating environment parameters along the channel dimension, decomposes it to obtain a set of local time sequence correlation feature matrices of the device operating environment parameters, so as to more carefully analyze the features of different channels. Through feature decoupling, the features of the device operating environment parameters can be more comprehensively analyzed, the differences and connections between different channels can be captured, and more rich feature information can be provided for subsequent cross-modal interaction.

[0083] In particular, in one preferred example of the application, the step S4322 comprises: first, preliminarily aligning the fault detection threshold key feature encoding vector and the device operating environment parameter time sequence key feature encoding vector to obtain a fault detection threshold-device operating environment key feature alignment encoding vector and a device operating environment-fault detection threshold key feature alignment encoding vector, which is expressed by the formula:

[0084]

[0085] wherein, denotes the subtraction of the position points, v m denotes the fault detection threshold-device operating environment key feature alignment encoding vector, v m denotes the device operating environment-fault detection threshold key feature alignment encoding vector.

[0086] That is, for the fault detection threshold key feature encoding vector and the device operating environment parameter time sequence key feature encoding vector used for the key feature representation of the dissimilar modal features, a preliminary structured semantic alignment of the general explicit correlation in the shared representation space is performed through the residual mechanism based on the dissimilar modal mapping to obtain the fault detection threshold-device operating environment key feature alignment encoding vector and the device operating environment-fault detection threshold key feature alignment encoding vector, which provides a better basis for subsequent interaction operation, so that the information between the two can be more effectively fused and interacted, and the effect of cross-modal interaction is improved.

[0087] Secondly, the fault detection threshold-device operating environment key feature alignment encoding vector and the device operating environment-fault detection threshold key feature alignment encoding vector are adaptively compensated and integrated based on the residual architecture to obtain a fault detection threshold-device operating environment key feature difference compensation encoding matrix, which is expressed by the formula:

[0088]

[0089] wherein ||·||2 denotes the two-norm of the vector, e denotes the natural constant, denotes matrix multiplication, M cor denotes the fault detection threshold-device operating environment key feature difference compensation encoding matrix.

[0090] Then, based on the fault detection threshold-device operating environment key feature difference compensation coding matrix, the fault detection threshold-device operating environment key feature interaction coding matrix is subjected to feature parameter tuning optimization to obtain an optimized priori knowledge constraint matrix, which is expressed by a formula as follows:

[0091]

[0092] Wherein, M te m' represents the optimized priori knowledge constraint matrix.

[0093] That is, based on the explicit modeling graph structure global semantic spatialization representation under the mapping preference, the residual mechanism is further utilized to capture the "missing" information in the alignment process under the different mapping modes, and the explicit association is performed for adaptive compensation, so as to further capture the inter-modal fine-grained saliency difference information in the preliminary alignment process. In this way, the contradiction between the alignment complexity of the cross-modal different internal driving mapping and the explicit association intuitiveness of the graph structure can be optimized, and the inter-modal global semantic constraint illusion under the condition of fine-grained relationship misalignment between local salient regions in the fault detection threshold-device operating environment key feature interaction coding matrix is reduced.

[0094] Finally, the query vector, the set of key matrices and the optimized priori knowledge constraint matrix are input into the cross-modal template constraint encoder to obtain a set of fault detection threshold-device operating environment fine-grained implicit interaction coding vectors, which is expressed by a formula as follows:

[0095]

[0096] Wherein, L is the feature dimension value of M i , and softmax(·) represents a normalized exponential function, v ti represents the i-th fault detection threshold-device operating environment fine-grained implicit interaction coding vector in the set of fault detection threshold-device operating environment fine-grained implicit interaction coding vectors.

[0097] That is, the cross-modal template constraint encoder is adopted in the present application to perform cross-modal encoding, and the powerful capability of the transformer architecture is utilized to efficiently perform cross-modal encoding on the input query vector, the set of key matrices and the optimized priori knowledge constraint matrix, so as to obtain the set of fault detection threshold-device operating environment fine-grained implicit interaction coding vectors which can reflect the fine-grained implicit interaction relationship between the fault detection threshold and the device operating environment parameters, thereby providing an important intermediate result for subsequent calculation of the fault detection sensitivity implicit coding feature vector.

[0098] In one specific example of the present application, the step S4323 is expressed by a formula as follows:

[0099]

[0100] wherein n represents the length of the set of fault detection threshold - device operating environment fine-grained implicit interaction encoding vectors, and vf represents the fault detection sensitivity implicit encoding feature vector.

[0101] That is, by calculating the position-wise mean vector, the information of the fault detection threshold - device operating environment fine-grained implicit interaction encoding vectors is integrated to obtain the fault detection sensitivity implicit encoding feature vector, which can be used to more accurately evaluate the sensitivity of fault detection and provide a more reliable basis for fault detection.

[0102] Specifically, in one specific example of the present application, the step S44 comprises: first, inputting the fault detection sensitivity implicit encoding feature vector into a decoder-based fault detection sensitivity explicit modeling unit to obtain a decoding value of the fault detection sensitivity. It should be understood that the fault detection sensitivity implicit encoding feature vector comprehensively reflects the sensitivity relationship between the device operating environment data and the fault detection threshold, and reveals the adaptability of the fault detection threshold under different environmental conditions. The decoder adopts a deep learning network structure, and by decoding the fault detection sensitivity implicit encoding feature vector, it is mapped back to the explicit expression space of the fault detection sensitivity, thereby obtaining a specific decoding value of the fault detection sensitivity. The decoding value directly reflects the detection ability of the fault detection threshold for potential faults under the current device operating environment conditions, i.e., the sensitivity level, which provides a direct basis for the subsequent dynamic adjustment of the fault detection threshold.

[0103] Then, based on the decoding value of the fault detection sensitivity, the fault detection threshold is adjusted to obtain the adjusted fault detection threshold. Specifically, if the decoding value shows that the fault detection sensitivity is low, it means that the currently set fault detection threshold may not be able to effectively capture potential faults under the current environment, and the threshold needs to be appropriately lowered to improve the sensitivity; on the contrary, if the decoding value shows that the fault detection sensitivity is high, it may lead to misjudgment, and the threshold needs to be appropriately increased.

[0104] In particular, in the technical solution of the present application, the fault detection threshold low-dimensional embedding encoding vector and the equipment operating environment parameter time sequence association encoding feature map respectively represent the low-dimensional embedding encoding features of the fault detection threshold and the time sequence high-dimensional implicit association features of the equipment operating environment data. When performing the fault detection sensitivity analysis based on the neural network model, the feature dimension and feature encoding mode between the fault detection threshold low-dimensional embedding encoding vector and the equipment operating environment parameter time sequence association encoding feature map are essentially different, which will cause the target domain decoding discrete of the fault detection sensitivity implicit encoding feature vector relative to the decoder-based fault detection sensitivity explicit modeling unit to affect the accuracy of the decoding value of the fault detection sensitivity.

[0105] Based on this, in the technical solution of the present application, before the fault detection sensitivity implicit encoding feature vector is input into the decoder-based fault detection sensitivity explicit modeling unit to obtain the decoding value of the fault detection sensitivity, the fault detection sensitivity implicit encoding feature vector is first adjusted to obtain an optimized fault detection sensitivity implicit encoding feature vector, and the specific process includes the following steps:

[0106] The feature values of each position in the fault detection sensitivity implicit encoding feature vector are subjected to first scale coherence scaling to obtain a first fault detection sensitivity implicit encoding coherence scaling matrix, denoted as:

[0107]

[0108] wherein v i and v j respectively represent the feature values of the i-th position and the j-th position in the fault detection sensitivity implicit encoding feature vector, and M1(i,j) represents the feature value of the (i,j) position in the first fault detection sensitivity implicit encoding coherence scaling matrix.

[0109] The feature values of each position in the fault detection sensitivity implicit encoding feature vector are subjected to second scale coherence scaling to obtain a second fault detection sensitivity implicit encoding coherence scaling matrix, denoted as:

[0110]

[0111] wherein v i and v j respectively represent the feature values of the i-th position and the j-th position in the fault detection sensitivity implicit encoding feature vector, and M2(i,j) represents the feature value of the (i,j) position in the second fault detection sensitivity implicit encoding coherence scaling matrix.

[0112] The first fault detection sensitivity implicit coding coherent scaling matrix and the second fault detection sensitivity implicit coding coherent scaling matrix are scale-coupled to obtain a fault detection sensitivity implicit coding multi-scale coherent scaling matrix, which is expressed as:

[0113] M x =Sigmoid(W x (αM1+βM2)+B x )

[0114] Where M1 represents the first fault detection sensitivity implicit coding coherent scaling matrix, M2 represents the second fault detection sensitivity implicit coding coherent scaling matrix, α represents the first weight hyperparameter, β represents the second weight hyperparameter, W x represents the first pre-training weight matrix, B x Represents the first pre-training bias matrix, Sigmoid represents the Sigmoid activation function, M x Represents the implicitly encoded multi-scale coherent scaling matrix of fault detection sensitivity.

[0115] Extract the fault detection sensitivity decoding weight matrix M from the decoder-based fault detection sensitivity explicit modeling unit d ;

[0116] Based on the fault detection sensitivity implicitly coded multi-scale coherent scaling matrix, the fault detection sensitivity decoding weight matrix is ​​dynamically coherently modulated to obtain an optimized decoding weight matrix, which is expressed as:

[0117] M t =ReLU(W t (εM x +ωM d )+B t )

[0118] Among them, M x represents the multi-scale coherent scaling matrix of the implicit coding of fault detection sensitivity, M d represents the fault detection sensitivity decoding weight matrix, ε represents the third weight hyperparameter, ω represents the fourth weight hyperparameter, W t Represents the second pre-training weight matrix, Bx represents the second pre-training bias matrix, ReLU represents the ReLU activation function, M t Represents the optimized decoding weight matrix.

[0119] Based on the optimized decoding weight matrix, the fault detection sensitivity implicit coding feature vector is reconstructed by performing a self-similar decoding trajectory to obtain the optimized fault detection sensitivity implicit coding feature vector, which is expressed as:

[0120]

[0121] wherein V represents the fault detection sensitivity latent encoding feature vector, M t represents the optimized decoding weight matrix, represents the matrix multiplication, V f represents the optimized fault detection sensitivity latent encoding feature vector.

[0122] Correspondingly, in this specific example, the fault detection sensitivity latent encoding feature vector is multiscale coherent scaled and scale-coupled to obtain a vortex coherence scale of the system. Further, based on the vortex coherence scale, the prior weight matrix of the decoder is optimized by an analog dynamics, and the fault detection sensitivity latent encoding feature vector is reconstructed by a self-similar decoding trajectory with the optimized decoding weight matrix, so as to optimize the phase transition threshold response performance of the fault detection sensitivity latent encoding feature vector with respect to different clusters (especially with respect to the target decoding domain). In this way, the accuracy of the decoding value of the fault detection sensitivity obtained by the decoder-based fault detection sensitivity explicit modeling unit is improved.

[0123] In the above-mentioned fault monitoring method of the vacuum circuit breaker, the step S5, based on the adjusted fault detection threshold, performs fault detection on the vibration data to determine whether the vacuum circuit breaker has a fault. Here, the collected vibration data is compared and analyzed with the adjusted fault detection threshold. For example, a sliding window method can be used to select vibration data within a certain time period (such as 10 seconds) each time, calculate the statistical characteristics such as mean value of the frequency and amplitude, and compare with the adjusted threshold. If the vibration data exceeds the adjusted threshold range, it can be judged that the vacuum circuit breaker has a potential fault; otherwise, it is considered that the device is in normal operation. Through this process, potential faults of the vacuum circuit breaker can be found in time and accurately, providing a strong basis for maintenance and repair of the device.

[0124] In summary, the fault monitoring method of the vacuum circuit breaker based on the embodiments of the present application is illustrated, which realizes fault monitoring of the vacuum circuit breaker by analyzing the frequency and amplitude of the vibration data of the vacuum circuit breaker during device operation. At the same time, by monitoring the running environment data of the vacuum circuit breaker in real time, and using deep learning-based data processing technology to interactively analyze the device running environment data and the currently set fault detection threshold, the current fault detection sensitivity is identified, and the fault detection threshold is dynamically adjusted to improve the accuracy and timeliness of fault detection. In this way, the probability of misjudgment and missed judgment can be effectively reduced, the intelligent monitoring of the vacuum circuit breaker fault is realized, and the stability and safety of the power system are improved.

[0125] Further, a fault monitoring system of a vacuum circuit breaker is also provided.

[0126] Figure 7 FIG. 1 is a block diagram of a fault monitoring system for a vacuum circuit breaker according to an embodiment of the present application. Figure 7 As shown, according to an embodiment of the present application, the fault monitoring system 100 for a vacuum circuit breaker includes: a contact resistance power consumption proportional coefficient acquisition module 110, which is used to obtain a time queue of the power consumption proportional coefficient of the contact resistance of the contact in the vacuum circuit breaker; an operating structure power consumption proportional coefficient acquisition module 120, which is used to obtain a time queue of the power consumption proportional coefficient of the operating structure in the vacuum circuit breaker; a control circuit power consumption proportional coefficient acquisition module 130, which is used to obtain a time queue of the power consumption proportional coefficient of the control circuit in the vacuum circuit breaker; a power consumption abnormality judgment module 140, which is used to judge the power consumption proportional coefficient of the control circuit based on the time queue of the power consumption proportional coefficient of the contact resistance, the time queue of the power consumption proportional coefficient of the operating structure and the time queue of the control circuit. A time queue of the power consumption proportional coefficient of the control circuit is used to determine whether the power consumption of the vacuum circuit breaker is abnormal, wherein determining whether the power consumption of the vacuum circuit breaker is abnormal includes: performing multi-scale temporal correlation aggregation analysis based on the power consumption proportional coefficient on the contact resistance, the operating structure and the control circuit and different temporal scale core feature anchoring interaction to obtain a judgment result; a power consumption abnormality type determination module 150 is used to determine the power consumption abnormality type of the vacuum circuit breaker in response to the judgment result that the power consumption of the vacuum circuit breaker is abnormal to obtain a power consumption abnormality type set; a power consumption abnormality feedback module 160 is used to feed back the power consumption abnormality type set to a remote monitoring center of the vacuum circuit breaker.

[0127] Here, those skilled in the art will appreciate that the specific operations of each module in the above-mentioned vacuum circuit breaker fault monitoring system have been described in detail above. Figures 1 to 6 The fault monitoring method of the vacuum circuit breaker has been described in detail, and therefore, its repeated description will be omitted.

[0128] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, and effects mentioned in the present invention are merely illustrative and non-limiting, and should not be construed as necessarily possessed by each embodiment of the present invention. Furthermore, the specific details of the above embodiments are provided for illustrative purposes and to facilitate understanding, and are not intended to be limiting. These details do not necessarily limit the present invention to being implemented using these specific details.

[0129] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in a certain embodiment can be referred to the relevant description of other embodiments. In the several embodiments provided by the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiments described above are merely schematic, for example, the unit division is only a logical function division, and other division manners can be used in actual implementation. The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e. they can be located in one place or distributed on a plurality of network units. In actual implementation, some or all of the units can be selected according to the actual needs to achieve the purposes of the embodiments.

[0130] It is apparent that the present application is not limited to the details of the foregoing exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or essential characteristics of the application. Accordingly, no matter from which point of view the embodiments are to be considered, they are to be regarded only as exemplary, and not as restrictive, the scope of the application being defined by the appended claims rather than the description given above and it is intended to embrace all alternatives falling within the meaning and range of equivalents of the claims. Any reference signs in the claims should not be construed as limiting the claims to the figures in which the reference signs are used.

[0131] Further, it is evident that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. A plurality of units can be implemented by one unit by means of software or hardware.

[0132] Finally, it should be noted that the above description is given for illustrative and descriptive purposes only. Furthermore, the above embodiments are merely used to illustrate the technical solutions of the present application but not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application.

Claims

1. A method for monitoring a vacuum circuit breaker fault, characterized in that: include: Collect vibration data of vacuum circuit breaker equipment during operation; Acquiring equipment operating environment data of the vacuum circuit breaker; Extracting a currently set fault detection threshold, wherein the fault detection threshold includes a vibration frequency threshold and a vibration amplitude threshold; Based on the equipment operating environment data of the vacuum circuit breaker, performing a fault detection sensitivity analysis and adjusting the currently set fault detection threshold to obtain an adjusted fault detection threshold; Based on the adjusted fault detection threshold, fault detection is performed on the vibration data to determine whether the vacuum circuit breaker has a fault.

2. The fault monitoring method for a vacuum circuit breaker according to claim 1, characterized in that: The equipment operating environment data includes a time queue of the ambient temperature and a time queue of the ambient humidity.

3. The fault monitoring method for a vacuum circuit breaker according to claim 2, characterized in that: Based on the equipment operating environment data of the vacuum circuit breaker, performing a fault detection sensitivity analysis and adjusting the currently set fault detection threshold to obtain an adjusted fault detection threshold, including: Performing a temperature-humidity time series interactive correlation analysis on the device operating environment data to obtain a device operating environment parameter time series correlation coding feature map; Performing low-dimensional embedded coding on the currently set fault detection threshold to obtain a low-dimensional embedded coding vector of the fault detection threshold; Performing a fault detection sensitivity analysis based on cross-domain joint coding on the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter time series correlation coding feature map to obtain a fault detection sensitivity implicit coding feature vector; Based on the fault detection sensitivity implicitly encoded feature vector, the fault detection threshold is adjusted to obtain the adjusted fault detection threshold.

4. The fault monitoring method for a vacuum circuit breaker according to claim 3, characterized in that: Performing a fault detection sensitivity analysis based on cross-domain joint coding on the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter time series correlation coding feature map to obtain a fault detection sensitivity implicit coding feature vector, including: Extracting key feature interaction information between the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter temporal correlation coding feature map to obtain a fault detection threshold-device operating environment key feature interaction coding matrix; Based on the fault detection threshold-device operating environment key feature interaction coding matrix, the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter temporal association coding feature map are guided to perform cross-modal interaction feature learning based on the attention mechanism to obtain the fault detection sensitivity implicit coding feature vector.

5. The fault monitoring method for a vacuum circuit breaker according to claim 4, characterized in that: Extracting key feature interaction information between the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter temporal correlation coding feature map to obtain a fault detection threshold-device operating environment key feature interaction coding matrix, including: Performing point convolution coding-based key feature extraction on the fault detection threshold low-dimensional embedded coding vector to obtain a fault detection threshold key feature coding vector; Extracting a device operating environment parameter timing key feature coding vector from the device operating environment parameter timing association coding feature map; The fault detection threshold key feature coding vector and the device operating environment parameter timing key feature coding vector are associated and coded to obtain the fault detection threshold-device operating environment key feature interaction coding matrix.

6. The fault monitoring method for a vacuum circuit breaker according to claim 5, characterized in that: Extracting a device operating environment parameter timing key feature coding vector from the device operating environment parameter timing association coding feature map includes: Performing feature partitioning on the device operating environment parameter time series correlation coding feature map to obtain a set of device operating environment parameter local time series correlation feature maps; Performing convolution coding and global pooling processing on each local time series correlation feature map of the device operating environment parameters in the set of local time series correlation feature maps of the device operating environment parameters to obtain a set of local time series correlation feature vectors of the device operating environment parameters; Calculating a key feature correlation factor of each device operating environment parameter local time series correlation feature vector in the set of device operating environment parameter local time series correlation feature vectors based on a feature distribution thereof to obtain a set of environmental parameter local time series correlation feature correlation factors; Based on the set of correlation factors of the local temporal correlation features of the environmental parameters, feature aggregation is performed on the set of local temporal correlation feature vectors of the device operating environment parameters to obtain the device operating environment parameter temporal key feature coding vector.

7. The fault monitoring method for a vacuum circuit breaker according to claim 6, characterized in that: Based on the fault detection threshold-device operating environment key feature interaction coding matrix, guiding the fault detection threshold low-dimensional embedded coding vector and the device operating environment parameter temporal correlation coding feature map to perform cross-modal interactive feature learning based on an attention mechanism to obtain the fault detection sensitivity implicit coding feature vector, including: Performing feature decoupling along the channel dimension on the device operating environment parameter temporal correlation coding feature map to obtain a set of local temporal correlation feature matrices of the device operating environment parameters; Using the fault detection threshold key feature encoding vector as a query vector, using each device operating environment parameter local temporal correlation feature matrix in the set of device operating environment parameter local temporal correlation feature matrices as a key matrix, and using the fault detection threshold-device operating environment key feature interaction encoding matrix as a priori knowledge constraint matrix, cross-modal style constraint encoding is performed on the query vector, the set of key matrices, and the priori knowledge constraint matrix to obtain a set of fault detection threshold-device operating environment fine-grained implicit interaction encoding vectors; The position-wise mean vector of the set of the fault detection threshold-device operating environment fine-grained implicit interaction coding vectors is calculated to obtain the fault detection sensitivity implicit coding feature vector.

8. The fault monitoring method for a vacuum circuit breaker according to claim 7, characterized in that: Cross-modal style constraint encoding is performed on the query vector, the set of key matrices, and the prior knowledge constraint matrix to obtain a set of fault detection threshold-device operating environment fine-grained implicit interaction encoding vectors, including: Preliminarily aligning the fault detection threshold key feature coding vector and the device operating environment parameter timing key feature coding vector to obtain a fault detection threshold-device operating environment key feature alignment coding vector and a device operating environment-fault detection threshold key feature alignment coding vector; Performing adaptive compensation integration based on a residual architecture on the fault detection threshold-device operating environment key feature alignment coding vector and the device operating environment-fault detection threshold key feature alignment coding vector to obtain a fault detection threshold-device operating environment key feature difference compensation coding matrix; Based on the fault detection threshold-equipment operating environment key feature difference compensation coding matrix, characteristic parameter tuning optimization is performed on the fault detection threshold-equipment operating environment key feature interaction coding matrix to obtain an optimized prior knowledge constraint matrix; The query vector, the set of key matrices and the optimized prior knowledge constraint matrix are input into a cross-modal template constraint encoder to obtain a set of fault detection threshold-device operating environment fine-grained implicit interaction encoding vectors.

9. The fault monitoring method for a vacuum circuit breaker according to claim 8, characterized in that: Adjusting the fault detection threshold based on the fault detection sensitivity implicitly encoded feature vector to obtain the adjusted fault detection threshold includes: Inputting the fault detection sensitivity implicitly encoded feature vector into a fault detection sensitivity explicit modeling unit based on a decoder to obtain a decoded value of the fault detection sensitivity; The fault detection threshold is adjusted based on the decoded value of the fault detection sensitivity to obtain the adjusted fault detection threshold.

10. A fault monitoring system for a vacuum circuit breaker, characterized in that: include: A vibration data collection module is used to collect vibration data of the vacuum circuit breaker equipment during operation; An operating environment data acquisition module, used to acquire equipment operating environment data of the vacuum circuit breaker; A fault detection threshold extraction module is used to extract the currently set fault detection threshold, wherein the fault detection threshold includes a vibration frequency threshold and a vibration amplitude threshold; a fault detection threshold adjustment module, configured to perform a fault detection sensitivity analysis and adjust the currently set fault detection threshold based on the equipment operating environment data of the vacuum circuit breaker to obtain an adjusted fault detection threshold; The vacuum circuit breaker fault judgment module is configured to perform fault detection on the vibration data based on the adjusted fault detection threshold value to judge whether the vacuum circuit breaker has a fault.

Images