Fault on-line detection and alarm system and method for low-voltage distribution box

Abstract

The invention relates to the field of online fault detection and alarm, and particularly discloses an online fault detection and alarm system and method for a low-voltage distribution box. The method comprises the following steps: firstly, acquiring incoming line voltages of the low-voltage distribution box at a plurality of preset time points acquired by a sensor, outgoing line voltages of the low-voltage distribution box at a plurality of preset time points acquired by the sensor, and an arc light signal oscillogram of the low-voltage distribution box acquired by an arc light sensor; and finally, a classification result is obtained through the classifier to judge whether the low-voltage distribution box has an electrical fault or not, so that accurate diagnosis of the electrical fault is realized, the maintenance cost is reduced, the service life of equipment is prolonged, and the reliability and safety of the equipment are improved.

Description

Technical Field

[0001] This application relates to the field of online fault detection and alarm, and more specifically, to an online fault detection and alarm system and method for a low-voltage distribution box. Background Art

[0002] A low-voltage distribution box is an electrical device used to distribute and control electric energy in a low-voltage power grid. It contains circuit breakers, fuses, and other protection devices to prevent overload, short circuit, and ground faults. Low-voltage distribution boxes are usually installed in residential, commercial, and industrial buildings to supply power to lighting, sockets, and other electrical devices, and are one of the important power distribution devices in the power system.

[0003] Currently, it is usually relied on manual judgment to determine whether there are electrical faults in the low-voltage distribution box. However, manual judgment is easily affected by external factors such as environmental conditions and personal emotions, and it is easy to make judgment errors, resulting in inaccurate judgment results.

[0004] Therefore, an online fault detection and alarm system and method for a low-voltage distribution box are desired. Summary of the Invention

[0005] To solve the above technical problems, this application is proposed. Embodiments of this application provide an online fault detection and alarm system and method for a low-voltage distribution box. First, it obtains the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by a sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by a sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by an arc light sensor. Then, using deep learning technology, feature extraction and correlation analysis are performed on the three. Finally, a classification result is obtained through a classifier to determine whether there are electrical faults in the low-voltage distribution box, so as to achieve precise diagnosis of electrical faults, thereby reducing maintenance costs, extending the service life of the equipment, and further improving the reliability and safety of the equipment.

[0006] According to one aspect of this application, an online fault detection and alarm system for a low-voltage distribution box is provided, which includes:

[0007] A low-voltage distribution box data acquisition module, configured to obtain the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by a sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by a sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by an arc light sensor;

[0008] A low-voltage distribution box data extraction module, configured to extract an arc voltage correlation feature vector and a low-voltage distribution box arc light signal feature vector from the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor;

[0009] The low-voltage distribution box fault judgment module is used to judge whether there is an electrical fault in the low-voltage distribution box based on the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector.

[0010] According to another aspect of the present application, there is provided a method for online detection and alarm of faults in a low-voltage distribution box, which includes:

[0011] Obtain the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor;

[0012] Extract the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector from the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor;

[0013] Based on the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector, judge whether there is an electrical fault in the low-voltage distribution box.

[0014] Compared with the prior art, the online detection and alarm system and method for faults in a low-voltage distribution box provided by the present application first obtain the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor, then use deep learning technology to perform feature extraction and correlation analysis on the three, and finally obtain a classification result through a classifier to judge whether there is an electrical fault in the low-voltage distribution box, so as to achieve accurate diagnosis of electrical faults, thereby reducing maintenance costs, extending the service life of the equipment, and further improving the reliability and safety of the equipment. Description of the Drawings

[0015] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of an online detection and alarm system for faults in a low-voltage distribution box according to an embodiment of the present application.

[0017] Figure 2It is a block diagram of a low-voltage distribution box data extraction module in a fault online detection and alarm system for a low-voltage distribution box according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a low-voltage distribution box fault judgment module in a fault online detection and alarm system for a low-voltage distribution box according to an embodiment of the present application.

[0019] Figure 4 It is a flowchart of a fault online detection and alarm method for a low-voltage distribution box according to an embodiment of the present application.

[0020] Figure 5 It is a block diagram of an electronic device according to an embodiment of the present application. Detailed implementation manners

[0021] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described herein.

[0022] Exemplary system

[0023] Figure 1 It is a block diagram of a fault online detection and alarm system for a low-voltage distribution box according to an embodiment of the present application. As Figure 1 shown, a fault online detection and alarm system 100 for a low-voltage distribution box according to an embodiment of the present application includes: a low-voltage distribution box data acquisition module 110, configured to acquire the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by a sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by a sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by an arc light sensor; a low-voltage distribution box data extraction module 120, configured to extract an arc voltage correlation feature vector and a low-voltage distribution box arc light signal feature vector from the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor; a low-voltage distribution box fault judgment module 130, configured to judge whether there is an electrical fault in the low-voltage distribution box based on the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector.

[0024] In the above-mentioned online fault detection and alarm system 100 for low-voltage distribution boxes, the low-voltage distribution box data acquisition module 110 is used to acquire the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, and the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors. It should be understood that a low-voltage distribution box is a device used for power distribution and control, usually installed in buildings or industrial sites. It receives electrical energy from the power supply system, distributes it through internal switches, protection devices, and circuits, and transmits the electrical energy to different circuits and devices. The low-voltage distribution box has the functions of protecting the circuit and controlling power distribution. At the same time, it can monitor parameters such as current and voltage to ensure the safe operation of the electrical system. Through the low-voltage distribution box, effective management and control of the power system can be achieved, ensuring the normal operation and safe use of electrical equipment. In current technical operations, it is usually determined whether there are electrical faults in the low-voltage distribution box through manual judgment. However, this method is easily affected by external factors, such as environmental conditions and personal emotions, which may lead to inaccuracies in subjective judgment, thus affecting the accurate identification of electrical faults. Moreover, considering that relying solely on voltage data cannot comprehensively capture information, while the arc signal waveform diagram can provide detailed information about the arc activity in the electrical system, helping to identify potential types of electrical faults, such as arc faults, short circuits, etc. The technical solution of this application combines the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, and the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors, and uses deep learning technology to judge whether there are electrical faults in the low-voltage distribution box. To sum up, acquiring the data of the incoming line voltage, the outgoing line voltage, and the arc signal waveform diagram is for the system to comprehensively monitor the electrical state of the distribution box, timely detect potential faults and provide early warnings to ensure the safe operation of electrical equipment and reduce the damage caused by faults to equipment and personnel.

[0025] In the above-mentioned online fault detection and alarm system 100 for low-voltage distribution boxes, the low-voltage distribution box data extraction module 120 is used to extract the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector from the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor. It should be understood that the incoming line voltage and outgoing line voltage data can help the system monitor the voltage conditions of the distribution box, including voltage stability, fluctuation conditions, etc. By monitoring the changes in the incoming line and outgoing line voltages, the system can timely detect abnormal voltages, such as too high, too low or large fluctuations, so as to take timely measures to avoid potential faults. An arc is an electrical fault phenomenon, usually accompanied by arc discharge, sparks and energy release. By extracting the arc voltage correlation feature vector, the system can capture the characteristics of arc discharge, such as voltage waveform, frequency, amplitude change, etc. These characteristics can help the system identify arc phenomena, so as to timely warn of possible electrical faults, such as short circuits, ground faults, etc. The arc signal waveform diagram provides information about possible arc phenomena in the distribution box. By extracting the feature vector of the arc signal, the system can analyze the characteristics of the arc, such as intensity, frequency, duration, etc., and then judge whether there are potential fault conditions. Therefore, extracting the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector enables the system to comprehensively analyze the electrical state in the distribution box, identify potential fault phenomena, give early warnings and take necessary maintenance measures to ensure the safe operation of the equipment and reduce the harm caused by faults to the equipment and personnel.

[0026] Figure 2 It is a block diagram of the low-voltage distribution box data extraction module in the online fault detection and alarm system for low-voltage distribution boxes according to the embodiments of the present application. As Figure 2As shown, in a specific embodiment of the present application, the low-voltage distribution box data extraction module 120 includes: an incoming line voltage feature extraction unit 121, configured to extract features from the incoming line voltages of the low-voltage distribution box at multiple predetermined time points collected by the sensor to obtain an incoming line voltage feature vector of the low-voltage distribution box; an outgoing line voltage feature extraction unit 122, configured to extract features from the outgoing line voltages of the low-voltage distribution box at multiple predetermined time points collected by the sensor to obtain an outgoing line voltage feature vector of the low-voltage distribution box; an arc voltage feature correlation unit 123, configured to perform feature correlation on the incoming line voltage feature vector and the outgoing line voltage feature vector of the low-voltage distribution box to obtain an arc voltage correlation feature vector; and an arc light signal waveform feature extraction unit 124, configured to extract features from the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor to obtain an arc light signal feature vector of the low-voltage distribution box. It should be understood that by extracting features from the incoming line voltage data, the stability, fluctuation condition, and change rule of the voltage can be analyzed. This helps the system understand the normal range and fluctuation of the voltage, and timely detect voltage abnormalities, such as too high, too low, or large fluctuations, so as to prevent potential electrical faults. Feature extraction can help the system capture key features in the voltage data, such as frequency, amplitude, waveform, etc., so as to identify voltage abnormalities or abnormal patterns. Through the obtained incoming line voltage feature vector of the low-voltage distribution box, the system can monitor the voltage status in real time, issue an alarm in time, and take necessary maintenance measures to ensure the safe operation of the equipment and reduce the damage caused by faults to the equipment and personnel.

[0027] Furthermore, by extracting features from the outgoing line voltage data, the system can analyze the stability, fluctuation condition, and change rule of the voltage. This helps to monitor the stability of the voltage output, and timely detect voltage abnormalities, such as too high, too low, or large fluctuations, so as to prevent potential equipment failures. Feature extraction can help the system capture key features in the voltage data, such as frequency, amplitude, waveform, etc. By analyzing these features, the system can identify abnormal patterns in the voltage output, such as abnormal voltage fluctuation frequency, voltage distortion, etc., so as to take measures in time to prevent the occurrence of faults. By establishing an outgoing line voltage feature vector, the system can monitor the long-term voltage output situation of the equipment, analyze the change trend of the voltage data, and thus predict the health status and performance degradation of the equipment.

[0028] Furthermore, arc fault is a dangerous electrical fault that may cause fires and equipment damage. By comprehensively analyzing the incoming line voltage feature vector, outgoing line voltage feature vector, and arc voltage data, it is possible to better identify the signs and characteristics of arc faults, thereby detecting them in a timely manner and taking measures to prevent the occurrence of arc faults. This helps the system understand the voltage transfer and variation laws during the operation of electrical equipment, providing more information and basis for the detection of arc faults. Moreover, the occurrence of arc faults is usually accompanied by abnormal voltage changes. Considering the characteristics of the incoming line voltage, outgoing line voltage, and arc voltage comprehensively can more comprehensively describe the characteristics of arc faults. Therefore, correlating these features to form an arc voltage correlation feature vector can improve the detection and prevention capabilities of arc faults.

[0029] In particular, an arc light sensor can capture the arc light signal waveforms generated by the arc phenomena occurring in electrical equipment. By extracting features from these waveforms, it is possible to analyze the features such as the spectrum, amplitude, and pulse width of the arc light signal, thereby detecting and identifying the presence and type of arc faults. The arc light signal waveforms contain rich information, and through feature extraction, this information can be converted into specific numerical features, which helps to more accurately detect and diagnose arc faults. Thus, by monitoring the changes in the arc light signals in electrical equipment in real time, the signs of arc faults can be detected in a timely manner, and early warnings can be issued for possible faults. This helps to reduce equipment damage and improve safety.

[0030] In a specific embodiment of the present application, the incoming line voltage feature extraction unit 121 includes: performing multi-resolution wavelet transform on the incoming line voltages of the low-voltage distribution box at multiple predetermined time points collected by the sensor to obtain multiple incoming line voltage feature values of the low-voltage distribution box; and passing the multiple incoming line voltage feature values of the low-voltage distribution box through the incoming line voltage time series encoder of the low-voltage distribution box to obtain the incoming line voltage feature vector of the low-voltage distribution box. It should be understood that multi-resolution wavelet transform can decompose and extract the features of a signal at different time scales, obtaining spectral information at different scales. By performing multi-scale wavelet transform on the incoming line voltage signal, it is possible to obtain feature values in different frequency ranges, thereby more comprehensively describing the characteristics of the voltage signal. Wavelet transform combines time-domain and frequency-domain analysis methods, and can simultaneously capture the changes in the signal in terms of time and frequency. Performing multi-resolution wavelet transform on the incoming line voltage signal can reveal the spectral characteristics of the voltage signal at different time scales, which helps to more comprehensively understand the variation law of the voltage signal.

[0031] Furthermore, the low-voltage distribution box incoming line voltage time series encoder can integrate and encode the voltage characteristic values at different time points in chronological order to form a feature vector containing time series information. This can better capture the changing trends and patterns of voltage signals over time. By integrating multiple characteristic values into a single feature vector through the time series encoder, the dimensionality of the original data can be reduced, avoiding the curse of dimensionality caused by excessive features while retaining important time series information, which helps simplify the data processing and analysis process. Based on the obtained low-voltage distribution box incoming line voltage feature vector, pattern recognition, anomaly detection, and predictive analysis can be performed to help monitor the operating status of electrical equipment, timely detect potential problems, and take corresponding measures to improve the reliability and safety of the equipment. Specifically, arrange the multiple low-voltage distribution box incoming line voltage characteristic values in a time dimension to form an incoming line input vector; perform normalization processing based on the maximum value on the incoming line input vector to obtain a normalized incoming line input vector; use the fully connected layer of the low-voltage distribution box incoming line voltage time series encoder to perform fully connected encoding on the normalized incoming line input vector to extract the high-dimensional hidden features of the characteristic values at each position in the normalized incoming line input vector; and use the one-dimensional convolutional layer of the low-voltage distribution box incoming line voltage time series encoder to perform one-dimensional encoding on the normalized incoming line input vector to extract the high-dimensional hidden correlation features of the correlations between the characteristic values at each position in the normalized incoming line input vector.

[0032] In a specific embodiment of the present application, the outgoing line voltage feature extraction unit 122 includes: constructing the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor into a low-voltage distribution box outgoing line voltage input vector; passing the low-voltage distribution box outgoing line voltage input vector through a low-voltage distribution box outgoing line voltage multi-scale neighborhood feature extractor to obtain the low-voltage distribution box outgoing line voltage feature vector. It should be understood that constructing the outgoing line voltage data at multiple predetermined time points into an input vector can make the data representation form more concise and unified, making it more convenient for feature extraction and analysis. By constructing the data in vector form, the efficiency of data processing can be improved, the complexity in the data processing process can be reduced, making it easier to process and manage, which is beneficial to improving the calculation efficiency and reducing the calculation cost.

[0033] Furthermore, the multi-scale neighborhood feature extractor can extract features from voltage data at different scales, capturing details and overall features at different scales. This helps to comprehensively understand the characteristics of voltage signals from multiple perspectives, improving the richness and diversity of features. Through multi-scale neighborhood feature extraction, local and global features of voltage data can be better captured, thereby enhancing the feature representation ability, reducing information loss, and improving the accuracy and robustness of data representation. Specifically, the multi-scale neighborhood feature extractor for the outgoing line voltage of the low-voltage distribution box includes: a first convolutional layer, a second convolutional layer parallel to the first convolutional layer, and a concatenation layer connected to the first convolutional layer and the second convolutional layer, where the first convolutional layer uses a one-dimensional convolutional kernel with a first scale, and the second convolutional layer uses a one-dimensional convolutional kernel with a second scale. More specifically, the first convolutional layer of the multi-scale neighborhood feature extractor for the outgoing line voltage of the low-voltage distribution box performs one-dimensional convolutional encoding on the input vector of the outgoing line voltage of the low-voltage distribution box to obtain a first-scale feature vector; the second convolutional layer of the multi-scale neighborhood feature extractor for the outgoing line voltage of the low-voltage distribution box performs convolutional encoding on the input vector of the outgoing line voltage of the low-voltage distribution box to obtain a second-scale feature vector; the first-scale feature vector and the second-scale feature vector are concatenated to obtain the feature vector of the outgoing line voltage of the low-voltage distribution box.

[0034] In a specific embodiment of the present application, the arc voltage feature correlation unit 123 includes: correlating the feature vector of the incoming line voltage of the low-voltage distribution box and the feature vector of the outgoing line voltage of the low-voltage distribution box to obtain an arc voltage feature matrix; passing the arc voltage feature matrix through an arc voltage feature encoder based on a convolutional neural network to obtain the arc voltage correlation feature vector. It should be understood that the correlation features between them, such as the transmission law of voltage change and power loss situation, can be extracted by combining the feature vectors of the incoming line voltage and the outgoing line voltage. This helps to more deeply understand the formation mechanism and characteristics of arc voltage in the circuit system. In this way, the input and output features of the circuit system can be comprehensively considered, so as to more comprehensively analyze the characteristics of arc voltage, that is, to reveal the influence and change law of arc voltage in the entire circuit system, providing a more comprehensive basis for the evaluation and monitoring of arc voltage.

[0035] Furthermore, the convolutional neural network has strong feature extraction and learning capabilities in processing image and sequence data. By inputting the arc voltage feature matrix into the convolutional neural network, the network can learn the abstract features in the arc voltage data and capture the spatial correlation and the association between features in the arc voltage feature matrix, which helps to extract richer and more meaningful arc voltage association features, thereby better describing the overall characteristics of the arc voltage data. Through the convolutional neural network encoder, the arc voltage feature matrix can be efficiently dimensionally reduced and represented, thereby reducing the dimension and complexity of the data while retaining important feature information. This helps to improve the representation ability and processing efficiency of the arc voltage data. Specifically, each layer of the arc voltage feature encoder based on the convolutional neural network performs convolutional processing, mean pooling processing based on the local feature matrix, and non-linear activation processing on the input data respectively during the forward pass of the layer to output the arc voltage association feature vector by the last layer of the arc voltage feature encoder based on the convolutional neural network, where the input of the arc voltage feature encoder based on the convolutional neural network is the arc voltage feature matrix.

[0036] In a specific embodiment of the present application, the arc light signal waveform feature extraction unit 124 includes: intercepting a plurality of sampling window data from the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor to obtain a plurality of arc light signal sampling window data of the low-voltage distribution box; constructing the plurality of arc light signal sampling window data of the low-voltage distribution box according to the channel dimension to obtain a three-dimensional input tensor of the arc light signal of the low-voltage distribution box; and passing the three-dimensional input tensor of the arc light signal of the low-voltage distribution box through a low-voltage distribution box arc light signal feature extractor based on a three-dimensional convolutional neural network to obtain the arc light signal feature vector of the low-voltage distribution box. It should be understood that a plurality of sampling window data provides more data samples. By intercepting a plurality of sampling window data, the entire signal waveform can be segmented into multiple local windows, so as to analyze the arc light signal features in each window in more detail, thereby increasing the richness and diversity of the data. This helps to obtain more comprehensive information during the analysis and processing, and improves the accuracy and generalization ability of the model. Moreover, through a plurality of sampling window data, the timing information of the signal, that is, the signal change situation in different time periods, can be considered. This helps to better understand the dynamic characteristics and change trends of the signal. By comparing the differences between different sampling window data, abnormal detection and fault diagnosis can be carried out more effectively.

[0037] Furthermore, the arc light signal sampling window data of the low-voltage distribution box may contain information of multiple channels, and each channel represents different signal characteristics or sensors. Constructing it into a three-dimensional tensor can preserve the relationships between these channels, enabling the model to consider the information of different channels simultaneously. After constructing the data into a three-dimensional tensor, a convolutional neural network can be directly applied for processing. Convolutional neural networks perform excellently in processing image and sequence data. By constructing it into a three-dimensional tensor, the structure and characteristics of the data can be better represented, which helps to improve the data representation ability. Considering that the input data accepted by many deep learning models (such as convolutional neural networks, recurrent convolutional neural networks, etc.) is usually a multi-dimensional tensor, constructing the data into a three-dimensional tensor meets the input requirements of these models and is beneficial to the training and application of the model.

[0038] Furthermore, a three-dimensional convolutional neural network can effectively extract spatial features from three-dimensional data. For three-dimensional data such as the arc light signal of the low-voltage distribution box, through the stacking of convolutional layers and pooling layers, the convolutional neural network can learn local and global features in the data, thereby better capturing the feature information of the signal. The convolutional neural network uses the mechanism of parameter sharing, which can reduce the number of parameters of the model and improve the generalization ability of the model at the same time. This is very beneficial for processing complex three-dimensional data because features can be learned by sharing weights, thus better processing the data. By stacking multiple convolutional layers and pooling layers, the convolutional neural network can gradually learn the hierarchical feature representation of the data. For the arc light signal of the low-voltage distribution box, these hierarchical features can capture the features at different abstraction levels of the signal, thereby better representing the essence of the signal. Specifically, each layer of the low-voltage distribution box arc light signal feature extractor based on the three-dimensional convolutional neural network performs the following operations on the input data respectively during the forward pass of the layer: performing convolutional processing on the input data based on the convolutional kernel to generate a convolutional feature map; performing global average pooling processing on the convolutional feature map based on the feature matrix to generate a pooling feature map; and performing non-linear activation on the feature values at each position in the pooling feature map to generate an activation feature map; wherein, the output of the last layer of the low-voltage distribution box arc light signal feature extractor based on the three-dimensional convolutional neural network is the arc light signal feature vector of the low-voltage distribution box, the input of the second layer to the last layer of the low-voltage distribution box arc light signal feature extractor based on the three-dimensional convolutional neural network is the output of the previous layer, and the input of the low-voltage distribution box arc light signal feature extractor based on the three-dimensional convolutional neural network is the three-dimensional input tensor of the arc light signal of the low-voltage distribution box.

[0039] In the above-mentioned on-line fault detection and alarm system 100 of the low-voltage distribution box, the low-voltage distribution box fault judgment module 130 is used to judge whether there is an electrical fault in the low-voltage distribution box based on the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector. It should be understood that the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector respectively capture the feature information of the arc voltage and the distribution box signal, representing different types of data features. By integrating these two feature vectors, more comprehensive and rich information can be obtained, realizing the fusion of multi-dimensional information, which helps to improve the accuracy of fault diagnosis.

[0040] Figure 3 It is a block diagram of the low-voltage distribution box fault judgment module in the on-line fault detection and alarm system of the low-voltage distribution box according to the embodiment of the present application. As Figure 3 shown, in a specific embodiment of the present application, the low-voltage distribution box fault judgment module 130 includes: an arc data feature difference projection unit 131, which is used to perform intrinsic decomposition entropy balance optimization on the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector to obtain a low-voltage distribution box electrical fault classification feature vector; an electrical fault judgment generation unit 132, which is used to pass the low-voltage distribution box electrical fault classification feature vector through a classifier to obtain a classification result, and the classification result is used to judge whether there is an electrical fault in the low-voltage distribution box. It should be understood that fusing multiple features can help reduce the misjudgment rate. Since the arc voltage correlation feature and the distribution box arc light signal feature may cause interference or noise in different situations, fusing them together can reduce the possibility of misjudgment caused by a single feature and improve the accuracy of classification. Generally speaking, fusing the arc voltage correlation feature vector and the low-voltage distribution box arc light signal feature vector can provide a more comprehensive and accurate feature description for the electrical fault classification of the low-voltage distribution box, which helps to improve the accuracy and performance of fault classification.

[0041] In a specific embodiment of the present application, the arc data feature difference projection unit 131 is used to: calculate the autocorrelation matrix of the arc voltage correlation feature vector, and perform core component extraction on the autocorrelation matrix of the arc voltage correlation feature vector to obtain a set of arc voltage correlation feature core component coding vectors, which is expressed by the formula:

[0042]

[0043] where V represents the arc voltage correlation feature vector, T represents the transpose of the vector, M z represents the autocorrelation matrix, U represents the set of arc voltage correlation feature core component coding vectors, v1, v2, v mThey represent the first, second, and mth arc voltage correlation feature core component encoding vectors, Λ represents the arc voltage correlation feature diagonal matrix after core component extraction, λ1, λ m They represent the first and mth eigenvalues on the diagonal of the arc voltage correlation characteristic diagonal matrix respectively.

[0044] That is, by constructing an autocorrelation matrix, it is possible to quantify the mutual dependence of arc voltage-related feature vectors in the time series and reveal the linear coupling between different feature dimensions. This coupling may contain key information about the dynamic characteristics of the arc under fault conditions, but the noise and repeated information mixed in the original high-dimensional feature space will interfere with the effective discrimination of subsequent classifiers. Therefore, the present application maps the arc voltage-related feature vectors to a new low-dimensional space through core component extraction, and extracts the principal component with the largest variance in the data, that is, the core feature with the most concentrated information. This process not only eliminates redundancy and compresses data dimensions to reduce computational complexity, but also ensures that the most discriminative fault mode in the arc voltage-related features is effectively extracted by retaining the principal component with the largest variance.

[0045] In a specific embodiment of the present application, the arc data feature difference projection unit 131 is used to input the set of arc voltage related feature core component encoding vectors into a sequence encoder based on a forward LSTM model to obtain a set of arc voltage related feature core component context-related encoding vectors, which is expressed as follows:

[0046] F=LSTM([v1,v2,…,v m ])=[s1,s2,…,s m ]

[0047] Among them, LSTM represents the forward LSTM model, F represents the set of context-related encoding vectors of the core components of the arc voltage correlation feature, s1, s2, s m Represents the first, second, and mth arc voltage correlation feature core component context association encoding vector.

[0048] Specifically, a forward LSTM sequence encoder is introduced, and its gated recurrent structure is used to model the long-range dependencies between the core components of the arc voltage correlation features, specifically to mine their implicit structural information. The LSTM model transforms the discrete core components into feature expressions with temporal contextual associations according to the implicit logical order of their sorting, generating contextual encoding vectors for the core components of the arc voltage correlation features, thereby capturing cross-scale correlation patterns in the dynamic evolution of arc voltage.

[0049] In a specific embodiment of the present application, the arc data feature difference projection unit 131 is configured to: calculate the displacement dynamic entropy between each corresponding arc voltage correlation feature core component context correlation coding vector and arc voltage correlation feature core component coding vector in the set of arc voltage correlation feature core component context correlation coding vectors and the set of arc voltage correlation feature core component coding vectors to obtain a set of displacement dynamic entropy, which is expressed by the formula:

[0050]

[0051]

[0052] where w represents a bit-by-bit comparison function, v i represents the i-th arc voltage correlation feature core component coding vector, represents the feature value at the j-th position of the i-th arc voltage correlation feature core component coding vector, s i represents the i-th arc voltage correlation feature core component context correlation coding vector, represents the feature value at the j-th position of the i-th arc voltage correlation feature core component context correlation coding vector, ε represents a predetermined threshold, r i represents the i-th bit-by-bit displacement matching feature vector, represents the feature value at the j-th position of the i-th bit-by-bit displacement matching feature vector, L represents the length of the i-th bit-by-bit displacement matching feature vector, e i represents the i-th displacement dynamic entropy.

[0053] That is, by introducing displacement dynamic entropy, the difference pattern between the arc voltage correlation feature core component context correlation coding vector and the arc voltage correlation feature core component coding vector at the binary bit level is quantified, and the information reconstruction effect of context correlation modeling on the original features is captured. This information entropy-based measurement method can reveal the increase or decrease of feature uncertainty in the sense of information theory, distinguish information gain (such as effective information concentration after redundancy elimination) from information loss (such as loss of key feature bits), thereby providing a sensitive differential signal for subsequent optimization.

[0054] In a specific embodiment of the present application, the arc data feature difference projection unit 131 is configured to: perform a weighting process on the set of displacement dynamic entropy based on the normalized exponential function to obtain a set of displacement dynamic entropy weight coefficients, which is expressed by the formula:

[0055] a i = softmax(e i )

[0056] where softmax represents the normalized exponential function, ai represents the i-th bit dynamic entropy weight coefficient.

[0057] That is, through the competitive allocation characteristic of Softmax, the weights are forced to concentrate on high response values, thereby amplifying the contribution of key information bits. At the same time, through smoothing processing, the instability of the model caused by extreme weight allocation is avoided. Specifically, the exponential weight calculation strengthens the relative gap between entropy values, enabling small entropy value differences to be converted into significant weight differentiations, and then guiding the subsequent optimization process to focus on the feature dimensions sensitive to fault diagnosis. The generated set of bit dynamic entropy weight coefficients realizes the adaptive calibration of feature importance through probabilistic weights, ultimately improving the diagnostic accuracy and generalization ability of the low-voltage distribution box fault classifier.

[0058] In a specific embodiment of the present application, the arc data feature differential projection unit 131 is used to: based on the set of bit dynamic entropy weight coefficients, fuse the set of core component coding vectors of the arc voltage correlation features to obtain an optimized arc voltage correlation feature vector, which is expressed by the formula:

[0059]

[0060] where, v f represents the optimized arc voltage correlation feature vector.

[0061] That is, by element-wise weighted fusion of the bit dynamic entropy weight coefficient and the core component coding vector of the arc voltage correlation feature, both the basic information structure of the core component coding vector of the arc voltage correlation feature is retained, and the differential information components sensitive to fault diagnosis are specifically strengthened, thereby enhancing the sensitivity of the subsequent classifier to weak fault patterns and the generalization ability to complex working conditions.

[0062] In a specific embodiment of the present application, the arc data feature differential projection unit 131 is used to: perform weighted fusion on the optimized arc voltage correlation feature vector and the arc light signal feature vector of the low-voltage distribution box to obtain the electrical fault classification feature vector of the low-voltage distribution box. That is, by assigning differential weights to the optimized arc voltage correlation feature vector and the arc light signal feature vector of the low-voltage distribution box, both the key information of each can be retained, and the feature components sensitive to specific fault patterns can be strengthened through weight adjustment. This fusion strategy can eliminate the limitations of a single feature space, construct a more comprehensive and robust electrical fault classification feature vector for the low-voltage distribution box, and reduce the misjudgment risk caused by environmental noise or sensor errors.

[0063] Furthermore, the classifier can quickly process a large amount of data and generate classification results in a short time. This high efficiency enables the timely detection of electrical faults and the adoption of corresponding measures to reduce the losses caused by the faults to the system. Using the classifier for judgment can implement a standardized fault diagnosis process, improving the consistency and comparability of diagnosis. This helps establish a unified fault diagnosis standard and enhance the efficiency of electrical system maintenance and management.

[0064] In summary, in the embodiment of the present application, first, the incoming line voltages at multiple predetermined time points of the low-voltage distribution box collected by the sensor, the outgoing line voltages at multiple predetermined time points of the low-voltage distribution box collected by the sensor, and the arc light signal waveform diagram of the low-voltage distribution box collected by the arc light sensor are obtained. Then, using deep learning technology, feature extraction and correlation analysis are performed on the three, and finally, a classification result is obtained through the classifier to determine whether there is an electrical fault in the low-voltage distribution box, so as to achieve accurate diagnosis of electrical faults, thereby reducing maintenance costs, extending the service life of the equipment, and further improving the reliability and safety of the equipment.

[0065] As described above, the on-line fault detection and alarm system 100 of the low-voltage distribution box according to the embodiment of the present application can be implemented in various terminal devices, such as a server deployed with the on-line fault detection and alarm algorithm of the low-voltage distribution box. In one example, the on-line fault detection and alarm system 100 of the low-voltage distribution box can be integrated into the terminal device as a software module and / or a hardware module. For example, the on-line fault detection and alarm system 100 of the low-voltage distribution box can be a software module in the operating system of the terminal device, or can be an application program developed for the terminal device; of course, the on-line fault detection and alarm system 100 of the low-voltage distribution box can also be one of the many hardware modules of the terminal device.

[0066] Alternatively, in another example, the on-line fault detection and alarm system 100 of the low-voltage distribution box and the terminal device can also be separate devices, and the on-line fault detection and alarm system 100 of the low-voltage distribution box can be connected to the terminal device through a wired and / or wireless network and transmit interaction information according to a predefined data format.

[0067] Exemplary method

[0068] Figure 4 is a flowchart of the on-line fault detection and alarm method of the low-voltage distribution box according to the embodiment of the present application. As Figure 4As shown, the on-line fault detection and alarm method for a low-voltage distribution box according to an embodiment of the present application includes: S110, obtaining the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by a sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by a sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by an arc sensor; S120, extracting an arc voltage correlation feature vector and a low-voltage distribution box arc signal feature vector from the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor; S130, based on the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector, determining whether there is an electrical fault in the low-voltage distribution box.

[0069] Here, those skilled in the art can understand that the specific operations of each step in the above on-line fault detection and alarm method for a low-voltage distribution box have been described in detail in the description of the on-line fault detection and alarm system for a low-voltage distribution box referred to above Figures 1 to 3 and therefore, the repeated description thereof will be omitted.

[0070] Exemplary electronic device

[0071] Next, reference will be made to Figure 5 to describe an electronic device according to an embodiment of the present application.

[0072] As Figure 5 shown, the electronic device 10 includes an input device 11, an input interface 12, a central processing unit 13, a memory 14, an output interface 15, an output device 16, and a bus 17. Among them, the input interface 12, the central processing unit 13, the memory 14, and the output interface 15 are connected to each other through the bus 17, and the input device 11 and the output device 16 are respectively connected to the bus 17 through the input interface 12 and the output interface 15, and then connected to other components of the electronic device 10.

[0073] Specifically, the input device 11 receives input information from the outside and transmits the input information to the central processing unit 13 through the input interface 12; the central processing unit 13 processes the input information based on computer-executable instructions stored in the memory 14 to generate output information, temporarily or permanently stores the output information in the memory 14, and then transmits the output information to the output device 16 through the output interface 15; the output device 16 outputs the output information to the outside of the electronic device 10 for use by the user.

[0074] In one embodiment, Figure 5The electronic device 10 shown can be implemented as a network device, which may include: a memory configured to store programs; and a processor configured to run the programs stored in the memory to execute any one of the fault online detection and alarm methods for the low-voltage distribution box described in the above embodiments.

[0075] According to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program tangibly embodied on a machine-readable medium, and the computer program includes program codes for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network, and / or installed from a removable storage medium.

[0076] Those of ordinary skill in the art can understand that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, and their appropriate combinations. In the hardware implementation, the division between the functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component can have multiple functions, or a function or step can be executed by several physical components in cooperation. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or a non-transitory medium) and a communication medium (or a transitory medium). As is well known to those of ordinary skill in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. The computer storage medium includes, but is not limited to, RAM, ROM, EEPROM, flash memory, or other memory technologies, CD-ROM, digital versatile disk (DVD), or other optical disk storage, magnetic cassette, tape, magnetic disk storage, or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, as is well known to those of ordinary skill in the art, the communication medium generally includes computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and can include any information delivery medium.

[0077] It is understandable that the above embodiments are merely exemplary embodiments adopted to illustrate the principle of the present application. However, the present application is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present application, and these modifications and improvements are also regarded as the protection scope of the present application.

Claims

1. An on-line fault detection and alarm system for a low-voltage distribution box, characterized in that, Including: A low-voltage distribution box data acquisition module, configured to acquire the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, and the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors; A low-voltage distribution box data extraction module, configured to extract an arc voltage correlation feature vector and a low-voltage distribution box arc signal feature vector from the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors, and the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors; A low-voltage distribution box fault judgment module, configured to judge whether there is an electrical fault in the low-voltage distribution box based on the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector.

2. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 1, wherein, The low-voltage distribution box data extraction module includes: An incoming line voltage feature extraction unit, configured to perform feature extraction on the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors to obtain an incoming line voltage feature vector of the low-voltage distribution box; An outgoing line voltage feature extraction unit, configured to perform feature extraction on the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors to obtain an outgoing line voltage feature vector of the low-voltage distribution box; An arc voltage feature correlation unit, configured to perform feature correlation on the incoming line voltage feature vector and the outgoing line voltage feature vector of the low-voltage distribution box to obtain the arc voltage correlation feature vector; An arc signal waveform feature extraction unit, configured to perform feature extraction on the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors to obtain the low-voltage distribution box arc signal feature vector.

3. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 2, characterized in that, The incoming line voltage feature extraction unit includes: Performing multi-resolution wavelet transform on the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors to obtain multiple incoming line voltage feature values of the low-voltage distribution box; Passing the multiple incoming line voltage feature values of the low-voltage distribution box through an incoming line voltage time series encoder of the low-voltage distribution box to obtain the incoming line voltage feature vector of the low-voltage distribution box.

4. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 3, characterized in that, The outgoing line voltage feature extraction unit includes: Constructing the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by sensors into an outgoing line voltage input vector of the low-voltage distribution box; Passing the outgoing line voltage input vector of the low-voltage distribution box through an outgoing line voltage multi-scale neighborhood feature extractor of the low-voltage distribution box to obtain the outgoing line voltage feature vector of the low-voltage distribution box.

5. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 4, characterized in that, The arc voltage feature correlation unit includes: Performing correlation on the incoming line voltage feature vector and the outgoing line voltage feature vector of the low-voltage distribution box to obtain an arc voltage feature matrix; Passing the arc voltage feature matrix through an arc voltage feature encoder based on a convolutional neural network to obtain the arc voltage correlation feature vector.

6. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 5, characterized in that, The arc signal waveform feature extraction unit includes: Intercepting multiple sampling window data from the arc signal waveform diagram of the low-voltage distribution box collected by arc sensors to obtain multiple arc signal sampling window data of the low-voltage distribution box; Construct the arc signal sampling window data of the multiple low-voltage distribution boxes according to the channel dimension to obtain a three-dimensional input tensor of the low-voltage distribution box arc signal; Pass the three-dimensional input tensor of the low-voltage distribution box arc signal through a low-voltage distribution box arc signal feature extractor based on a three-dimensional convolutional neural network to obtain the low-voltage distribution box arc signal feature vector.

7. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 6, characterized in that, The low-voltage distribution box fault judgment module includes: An arc data feature difference projection unit, configured to perform eigen-decomposition entropy balance optimization on the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector to obtain a low-voltage distribution box electrical fault classification feature vector; An electrical fault judgment generation unit, configured to pass the low-voltage distribution box electrical fault classification feature vector through a classifier to obtain a classification result, and the classification result is used to judge whether there is an electrical fault in the low-voltage distribution box.

8. The on-line fault detection and alarm system for the low-voltage distribution box according to claim 7, characterized in that, The arc data feature difference projection unit is configured to: Calculate the autocorrelation matrix of the arc voltage correlation feature vector, and perform core component extraction on the autocorrelation matrix of the arc voltage correlation feature vector to obtain a set of arc voltage correlation feature core component coding vectors; Input the set of arc voltage correlation feature core component coding vectors into a sequence encoder based on a forward LSTM model to obtain a set of arc voltage correlation feature core component context correlation coding vectors; Calculate the displacement dynamic entropy between each corresponding arc voltage correlation feature core component context correlation coding vector and arc voltage correlation feature core component coding vector in the set of arc voltage correlation feature core component context correlation coding vectors and the set of arc voltage correlation feature core component coding vectors to obtain a set of displacement dynamic entropies; Perform weighted processing based on the normalized exponential function on the set of displacement dynamic entropies to obtain a set of displacement dynamic entropy weight coefficients; Based on the set of displacement dynamic entropy weight coefficients, fuse the set of arc voltage correlation feature core component coding vectors to obtain an optimized arc voltage correlation feature vector; Perform weighted fusion on the optimized arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector to obtain the low-voltage distribution box electrical fault classification feature vector.

9. An on-line fault detection and alarm method for a low-voltage distribution box, characterized in that, It includes: Obtain the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor; Extract the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector from the incoming line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, the outgoing line voltage of the low-voltage distribution box at multiple predetermined time points collected by the sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor; Based on the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector, judge whether there is an electrical fault in the low-voltage distribution box.

10. The on-line fault detection and alarm method for the low-voltage distribution box according to claim 9, characterized in that, Extract the arc voltage correlation feature vector and the low-voltage distribution box arc signal feature vector from the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor, and the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor, including: Perform feature extraction on the incoming line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor to obtain the incoming line voltage feature vector of the low-voltage distribution box; Perform feature extraction on the outgoing line voltage at multiple predetermined time points of the low-voltage distribution box collected by the sensor to obtain the outgoing line voltage feature vector of the low-voltage distribution box; Perform feature correlation on the incoming line voltage feature vector of the low-voltage distribution box and the outgoing line voltage feature vector of the low-voltage distribution box to obtain the arc voltage correlation feature vector; Perform feature extraction on the arc signal waveform diagram of the low-voltage distribution box collected by the arc sensor to obtain the arc signal feature vector of the low-voltage distribution box.

Images