Three-phase circuit series fault arc detection method and device under vibration condition

Abstract

The embodiment of the invention provides a three-phase circuit series fault arc detection method and device under a vibration condition, and the method comprises the steps: collecting a current signal of a three-phase loop under the vibration condition during the load operation period of a three-phase motor and a frequency converter; inputting the current signal into a pre-trained fault arc identification model, and outputting a prediction result of the current signal; and determining the phase selection positioning of the series fault arc according to the prediction result of the current signal. In this way, the fault arc occurring in the load circuit of the three-phase motor and the frequency converter can be timely and accurately detected, the fault phase can be positioned, the operation reliability of the circuit of the three-phase motor and the frequency converter can be improved, and the electrical fire accident caused by the series fault arc can be prevented.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of fault detection related to electric motors, and in particular to a method and device for detecting series fault arcs in a three-phase circuit under vibration conditions. Background Art

[0002] With the continuous advancement of industrial automation, the use of three-phase motors and variable-frequency drive loads has become increasingly common, but this has also brought new electrical safety hazards. Due to factors such as poor contact and aging of electrical equipment during the operation of three-phase motor and variable-frequency drive load circuits, the problem of series fault arcs has emerged.

[0003] Currently, the mainstream methods for identifying series fault arcs at home and abroad mainly include the following three: First, detection and research are carried out according to physical characteristics. Fault arcs are accompanied by various physical phenomena, including strong arc light, arc sound, electromagnetic radiation, and temperature changes, etc. Observing and analyzing physical characteristics is an effective means of identifying faults. This method is suitable for identifying fault arcs occurring at specific positions and cannot identify fault arcs for the entire line. Second, detection and research are carried out according to time-domain, frequency-domain, and time-frequency-domain characteristics. Fault arc signals usually exhibit non-stationarity, and their time-domain and frequency-domain characteristics change dynamically with time. Time-frequency-domain analysis can simultaneously capture the dynamic change characteristics of fault arc signals in time and frequency, thus more comprehensively reflecting their non-stationarity and transient characteristics. However, most of the features extracted by this method are highly subjective and have certain limitations, which in turn leads to poor generalization ability of the model and affects the detection effect and performance. Third, detection and research are carried out according to artificial intelligence algorithms. With the rapid development of artificial intelligence technology, machine learning and deep learning methods have gradually attracted wide attention in the academic community. It can automatically extract and learn the non-linear and high-dimensional characteristics of fault arc signals without relying on the complex manual feature design in traditional methods, showing broad application prospects in the field of fault arc detection.

[0004] Currently, there have been many studies on the field of fault arcs at home and abroad, which have greatly improved the detection speed and accuracy of fault arcs. However, most of the studies mainly focus on the detection of fault arcs in household loads, and there are few studies on the detection of three-phase motors and variable-frequency drive loads, and the detection research on series fault arcs under vibration conditions is also relatively scarce. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a detection scheme for series fault arcs in a three-phase circuit under vibration conditions, which can timely and accurately detect the fault arcs occurring in the three-phase motor and variable-frequency drive load circuit and locate the faulty phase.

[0006] In the first aspect of the present application, a method for detecting series fault arcs in a three-phase circuit under vibration conditions is provided, including: During the load operation of a three-phase motor and a frequency converter, collect the current signals of the three-phase circuit under vibration conditions; Input the current signal into a pre-trained fault arc recognition model to output the prediction result of the current signal; Determine the phase selection and positioning of the series fault arc according to the prediction result of the current signal.

[0007] In some embodiments, the fault arc recognition model is trained in the following manner: Collect historical current signal samples and corresponding arc voltage signal samples of the series circuits of a preset number of three-phase motor and frequency converter load lines; According to the arc voltage signal samples, divide the corresponding current signal samples into normal class samples and fault class samples, assign labels to each class of sample data, and construct a data set containing series fault arcs; Divide it into a training set and a test set according to a preset ratio, and construct input matrices for the data samples in the training set and the test set; Use the training set and the test set constructed as input matrices to train and test a pre-built fault arc recognition model of a residual neural network and a hybrid attention mechanism; Obtain a trained arc fault recognition model.

[0008] In some embodiments, the fault arc recognition model includes an initial convolutional layer, a residual layer, a batch normalization layer, and a fully connected layer; The initial convolutional layer is used to capture the features of the local area, and while extracting features through convolutional operations, it can also reduce the size of the output data; The residual layer enables the input signal to be directly transmitted to the subsequent layers through skip connections, thereby alleviating the problem of gradient disappearance in the training of deep neural networks; The batch normalization layer is used to accelerate training, improve the stability and performance of the model; The fully connected layer is used for feature calculation.

[0009] In some embodiments, the input channel of the initial convolutional layer is 1, the output channel is 64, the convolutional kernel size is 7, and the stride is 2. The calculation formula is: ; Among them, C_in represents the number of input channels, K represents the convolutional kernel size, C_out is the number of output channels, and 1 is the bias term.

[0010] In some embodiments, there are 4 residual layers, each residual layer contains 2 residual blocks, a total of 8 residual blocks, and each residual block contains two 3×3 convolutional layers.

[0011] In some embodiments, the prediction result is a vector with a shape of 4×1. If the largest index value in the vector is 0, it indicates that the data sample is a normal class sample; if the largest index value in the vector is 1, it indicates that the data sample is an A-phase fault class sample; if the largest index value in the vector is 2, it indicates that the data sample is a B-phase fault class sample; if the largest index value in the vector is 3, it indicates that the data sample is a C-phase fault class sample.

[0012] In some embodiments, it further includes: Converting the collected current signal into a voltage signal; The collecting of historical current signal samples of a series circuit of a preset number of three-phase motors and inverter load lines includes: Converting the historical current signals of a series circuit of a preset number of three-phase motors and inverter load lines collected into voltage signals as samples.

[0013] In a second aspect of the present application, there is provided a three-phase circuit series fault arc detection device under vibration conditions, including: A signal acquisition module, configured to collect current signals of a three-phase circuit under vibration conditions during the operation of a three-phase motor and an inverter load; A current signal prediction module, configured to input the current signal into a pre-trained fault arc recognition model and output a prediction result of the current signal; A phase prediction module, configured to determine the phase selection and positioning of a series fault arc according to the prediction result of the current signal.

[0014] In a third aspect of the present application, there is provided an electronic device. The electronic device includes: a memory and a processor. A computer program is stored on the memory, and when the processor executes the program, the method as described above is implemented.

[0015] In a fourth aspect of the present application, there is provided a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method according to the first aspect of the present application is implemented.

[0016] The three-phase circuit series fault arc detection method provided by the embodiment of the present application collects the current signals of the three-phase circuit under vibration conditions during the operation of the three-phase motor and the frequency converter load, and inputs the current signals into a pre-trained fault arc recognition model to output the prediction result of the current signals. According to the prediction result of the current signals, the phase selection and positioning of the series fault arc are determined. Thus, it can timely and accurately detect the fault arc occurring in the three-phase motor and the frequency converter load line and perform fault phase positioning, improving the operation reliability of the three-phase motor and the frequency converter circuit and preventing electrical fire accidents caused by series fault arcs.

[0017] It should be understood that the content described in the summary of the invention section is not intended to limit the key or important features of the embodiments of the present application, nor is it used to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Combined with the drawings and referring to the following detailed description, the above and other features, advantages, and aspects of the embodiments of the present application will become more obvious. In the drawings, the same or similar reference numerals represent the same or similar elements, where: Figure 1 It is a system architecture diagram related to the method provided by the embodiment of the present application.

[0019] Figure 2 It is a flowchart of the three-phase circuit series fault arc detection method under vibration conditions according to the embodiment of the present application; Figure 3 It is a block diagram of the three-phase circuit series fault arc detection device under vibration conditions according to the embodiment of the present application; Figure 4 It is a schematic structural diagram of a terminal device or a server suitable for implementing the embodiment of the present application; Figure 5 It is a schematic diagram of the improved residual network structure in the embodiment of the present application; Figure 6 It is a structural diagram of the series fault arc recognition model in the embodiment of the present application; Figure 7 It is a spatial attention visualization diagram in the embodiment of the present application; Figure 8 It is a confusion matrix of the recognition model in the embodiment of the present application; Figure 9 It is a t-SNE dimensionality reduction analysis result diagram in the embodiment of the present application; Figure 10 It is a schematic structural diagram of a device for providing series fault arc vibration conditions; Figure 11This is the schematic diagram of the phase selection and positioning experiment platform for the three-phase motor and frequency converter load circuit in the embodiments of this application. Specific embodiments

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present disclosure clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present disclosure with reference to the accompanying drawings in the embodiments of the present disclosure. Apparently, the described embodiments are some, but not all, of the embodiments of the present disclosure. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present disclosure without creative efforts shall fall within the protection scope of the present disclosure.

[0021] In addition, the term "and / or" in this article is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0022] Figure 1 The schematic diagram of an exemplary operating environment 100 in which the embodiments of the present disclosure can be implemented is shown. The operating environment 100 includes a client 110, a communication module 120, a server 130, and a device 140.

[0023] Among them, the client 110 can be the current user terminal for reporting instructions to the server 130. The server 130 can be a cloud server for maintaining the information of the devices currently connected in the cloud, such as the online / offline status and device locking status of the devices. When the device goes online or offline, update the online / offline status of the device and maintain the device locking status information, etc. Furthermore, the server 130 further includes: A device locking module for issuing locking / unlocking tasks to the device 140 to ensure the consistency of the status of the device 140 with that in the server 130. When the device 140 is offline, synchronize the locking status of the device 140 to the server 130 and create a locking task simultaneously. When the device 140 goes online, issue the locking task. The locking task is completed until the device 140 returns a locking success status.

[0024] The device 140 can be the target user terminal for receiving the instructions issued by the server 130, performing corresponding operations according to the instructions, and reporting the operation result information to the server 130.

[0025] The communication module 120 is used to maintain the communication link between the device 140 and the server 130 and report callback information to the server 130 when the device 140 goes online or offline. That is, the server 130 can be connected to the device 140 through the communication module 120.

[0026] Figure 2 The figure shows a flowchart of a method for detecting series fault arcs in a three-phase circuit under vibration conditions according to an embodiment of the present disclosure. The method can be executed by the server 130 in Figure 1 and includes the following steps: S201: During the operation of the three-phase motor and the frequency converter load, collect the current signals of the three-phase circuit under vibration conditions.

[0027] Specifically, a current transformer can be used to collect the loop current signals during the operation of the three-phase motor and the frequency converter load line, and convert the loop current signals into voltage signals.

[0028] S202: Input the current signal into a pre-trained fault arc recognition model to output the prediction result of the current signal.

[0029] Specifically, input the voltage signal obtained by converting the loop current signal into a pre-trained fault arc recognition model, so as to output the prediction result of the loop current signal.

[0030] Among them, the fault arc recognition model is trained in the following way: collect historical current signal samples and corresponding arc voltage signal samples of the series circuit of a preset number of three-phase motors and frequency converter load lines; divide the corresponding current signal samples into normal class samples and fault class samples according to the arc voltage signal samples, assign labels to each class of sample data, and construct a data set containing series fault arcs; divide it into a training set and a test set according to a preset ratio, and construct an input matrix for the data samples in the training set and the test set; use the training set and the test set constructed as the input matrix to train and test a pre-built fault arc recognition model of a residual neural network and a hybrid attention mechanism; obtain a trained arc fault recognition model.

[0031] The training of the fault arc recognition model in this embodiment is based on a device for providing vibration conditions for series fault arcs and a phase selection and positioning experimental platform for three-phase motors and frequency converter load lines. As Figure 10 shown, it is a structural schematic diagram of the device for providing vibration conditions for series fault arcs.

[0032] The series fault arc vibration condition providing device in this embodiment includes: a base, a spring shock absorber, a vibration plate, fastening screws, an electromagnetic chamber wall vibrator, a cable, and a copper nose. Among them, the base is used to support the entire experimental platform, provide a stable foundation, and avoid excessive displacement or shaking during the experiment; the spring shock absorber is used to absorb and buffer vibrations, reduce unnecessary mechanical shocks, and ensure the safety and controllability of the experimental equipment; the vibration plate is used to carry the entire vibration system, transmit vibrations to the experimental device, and make the experimental environment in a controllable vibration state; the fastening screws are used to fix key parts such as the electromagnetic chamber wall vibrator, ensure the stability of the experimental platform, and prevent vibrations from affecting the normal operation of the equipment; the electromagnetic chamber wall vibrator is used to generate mechanical vibrations, simulate the vibration environment under actual working conditions, and consider the influence of vibrations on fault arcs; the cable is used to transmit current and signals, connect different components of the experimental equipment, and ensure the normal operation of the circuit; the copper nose is an electrode made of brass material and is the main device for generating series fault arcs under vibration conditions. An arc is generated only when the vibration amplitude is sufficient to separate the electrodes through the contact of two loose electrodes.

[0033] Figure 11 It is the schematic diagram of the phase selection and positioning experimental platform for the three-phase motor and frequency converter load line. The colored dotted lines in the figure indicate that the fault arc generator is connected in series to one of the phases, while the other two phases remain in normal working conditions. Taking phase A as an example, when a fault occurs, first disconnect the connection between P1 and P2, and connect the vibration fault arc generator in series between these two points to simulate the generation of series fault arcs under vibration conditions. The same operation is performed when a fault occurs in other phases. The current sensor only collects the current of phase A and transmits it to the upper computer through the data acquisition card in the signal acquisition unit 1. This time-sequential current data is used for the detection and phase selection of series fault arcs under vibration conditions in the three-phase motor and frequency converter load line.

[0034] During the training process of the fault arc recognition model, the signal acquisition unit collects the arc voltage signal and the loop current signal of any one phase during the operation of the three-phase motor and the frequency converter load line, and synchronously transmits the signal data to the data processing unit; the data processing unit divides the loop current signal into normal and fault data samples according to the fault arc voltage signal, and adds labels to the data samples with faults occurring in different phases respectively to obtain a series fault arc data set. The data set is randomly shuffled and divided into a model training set and a test set according to a certain ratio; the model construction unit constructs an input matrix from the data samples in the model training set and test set, and builds a series fault arc recognition model that improves the residual network and combines the hybrid attention mechanism (CBAM-ResNet-18). Use the model training set and test set constructed as the input matrix to train and test the series fault arc recognition model to obtain a trained series fault arc recognition model; the fault arc detection and phase selection unit is deployed on the Raspberry Pi and is used to output the final recognition result of the recognition model according to the prediction result of the series fault arc recognition model. In the embodiment of the present application, the signal acquisition unit includes a voltage transformer, a current transformer, a data acquisition card, and a DC power supply; among them, The voltage transformer is used to convert the arc voltage signal into a voltage signal during the operation of the three-phase motor and the frequency converter load line. The current transformer is used to convert the loop current signal into a voltage signal during the operation of the three-phase motor and the frequency converter load line. In one embodiment, both the voltage transformer and the current transformer are of the Hall type, and they convert the primary side signal into a voltage signal based on the Hall closed-loop zero-flux principle; The data acquisition card is used to perform analog-to-digital conversion processing on the voltage signal. Specifically, the voltage signal is converted into a corresponding digital quantity through ADC sampling; The DC power supply is used to provide power for the voltage transformer, current transformer, and data acquisition card.

[0035] In the embodiment of the present invention, the data processing unit is implemented using Python 3.9.

[0036] During the division of the series fault arc data set, taking the three-phase motor and the frequency converter load as an example, the loop current sample data set A is obtained when the line is operating normally, and the loop current sample data set B is obtained when a series fault arc appears in the line. Combine the sample data sets A and B and perform a random shuffling operation to obtain the series fault arc data set C. Divide the data set C according to a certain ratio to obtain the model training set and test set. In the embodiment of the present application, the data set C is divided into the model training set and test set according to the ratio of 75% and 25%.

[0037] For the schematic diagram of the improved residual network structure, see Figure 5By applying the improved residual structure to the residual network, deep feature fitting can be effectively achieved with a relatively low parameter scale, while suppressing the problem of network performance degradation. The proposed improved ResNet-18 residual neural network based on the hybrid attention mechanism has a network structure as shown in Table 1.

[0038] Table 1 Improved Structure of ResNet-18

[0039] The model construction unit is implemented using Python 3.9 and the PyTorch deep learning framework. The model construction unit includes a batch reading module, an identification model module, and a model training and testing module; among them, The batch reading module is used to batch read data samples constructed as input matrices into the series fault arc identification model. In the embodiments of the present application, during each iteration of model training, the model training set is randomly shuffled, and every 16 data samples are used as a batch and input into the series fault arc identification model; The identification model module processes the time-series current data input by the host computer using the improved residual neural network model proposed in the present invention. See Figure 6 , the improved residual neural network greatly reduces the total number of parameters of the model, and the calculation steps of the number of parameters for each layer are as follows: Step 1: The initial convolutional layer is used to capture features in the local area, and while extracting features through convolutional operations, it can also reduce the size of the output data. In the embodiments of the present application, the input channel of the initial convolutional layer is 1, the output channel is 64, the convolutional kernel size is 7, and the stride is 2. Among them, C_in represents the number of input channels, K represents the convolutional kernel size, C_out is the number of output channels, and 1 is the bias term. Its calculation formula is as follows:

[0040] For the initial convolutional layer:

[0041] Therefore, the initial convolutional layer has 512 parameters.

[0042] Step 2: The residual layer contains 2 residual blocks, which are used to solve the problems of gradient disappearance and network degradation. By introducing "skip connections", the input signal can be directly transmitted to the subsequent layers, thereby alleviating the problem of gradient disappearance in the training of deep networks. In the embodiments of the present application, there are a total of 4 residual layers, each residual layer contains 2 residual blocks, and there are a total of 8 residual blocks. And each residual block contains two 3×3 convolutional layers, so there are a total of 16 convolutional layers.

[0043] For one residual layer:

[0044] Therefore, there are a total of 98,816 parameters in the 4 residual layers.

[0045] Step 3: The batch normalization layer is used to accelerate training, improve model stability and performance. For the batch normalization layer after each convolutional layer, its calculation formula is as follows:

[0046] Since the output channels of each convolution are 64, for each batch normalization layer:

[0047] Therefore, there are a total of 2,048 parameters in the 16 batch normalization layers.

[0048] Step 4: The final fully connected layer receives 512 inputs and outputs 4 categories:

[0049] Therefore, there are a total of 2,049 parameters in the fully connected layer.

[0050] The total number of parameters of the network is obtained by summing up the number of parameters of each layer, and its total number of parameters is 103,425. Compared with the classical residual network, after calculating through the same steps above, the total number of parameters is 154,884, and the difference between the two is 1.5 times. In the embodiment of the present application, an improved residual neural network recognition model is built using the PyTorch deep learning framework, and a hybrid attention mechanism is introduced into the recognition model. The recognition model includes a feature extraction module and a feature calculation module. The feature extraction module consists of 1 initial convolutional layer C1, 4 residual layers (16 convolutional layers C2 - C17, 16 batch normalization layers B1 - B16, 8 hybrid attention modules H1 - H8). The feature calculation module consists of 1 unfolding layer Z1 and 1 fully connected layer F1, see Figure 6 .

[0051] The model training and testing module uses the model training set processed by the batch reading module to train the series fault arc recognition model according to certain training settings. In the embodiment of the present application, the training iteration times of the recognition model are set to 60, the optimizer uses the Adam optimization algorithm, and the categorical cross-entropy function is used to measure the loss between the predicted value and the true value of the series fault arc recognition model. Its calculation formula is as follows:

[0052] In the formula, represents the true value, represents the predicted value of the recognition model, is the data sample, represents the number of data samples.

[0053] During each training iteration, the Adam optimization algorithm is used to update the weight parameters of the recognition model, and the loss function values and accuracies of the recognition model on the model training set and the test set are examined. The recognition model is trained according to the number of training iterations. When the number of training iterations reaches 30, if the loss function value of the recognition model approaches 0 and is basically stable, the training iteration process of the recognition model is terminated. Otherwise, the number of training iterations of the recognition model is appropriately extended.

[0054] In the embodiment of the present application, the spatial attention in the improved ResNet-18 detection model is visually analyzed, and a periodic fault arc current signal is multiplied by the spatial attention in the hybrid attention mechanism. See Figure 7 . The spatial attention mechanism enables the model to focus on the key regions where distortions occur in the current signal. In these regions, the product features are more similar to the original current waveform and obtain higher attention weights. This mechanism improves the sensitivity and recognition ability of the model to the time-domain features of fault arcs.

[0055] In the embodiment of the present application, the loss value of the trained series fault arc detection and phase selection recognition model on the model training set and the test set is less than 0.7, and the accuracy is higher than 92.5%, indicating that the trained series fault arc detection and phase selection recognition model has high reliability and can meet the actual detection needs.

[0056] In the embodiment of the present application, the fault arc detection and phase selection unit outputs the final recognition result of the recognition model according to the prediction value of the series fault arc detection and phase selection recognition model trained by the model training and test module of the model construction unit. Specifically, a data sample processed by the batch reading module of the model construction unit is input into the series fault arc recognition model, and the recognition model will output a vector with a shape of 4×1. If the maximum index value in this vector is 0, it indicates that this data sample is a normal class sample. If the maximum index value in this vector is 1, it indicates that this data sample is an A-phase fault class sample. If the maximum index value in this vector is 2, it indicates that this data sample is a B-phase fault class sample. If the maximum index value in this vector is 3, it indicates that this data sample is a C-phase fault class sample.

[0057] In the embodiment of the present application, the experimental current is set to 5A, and the vibration frequencies of the electromagnetic bin wall vibrator are 40Hz, 45Hz, and 50Hz. The sampling frequency of the data acquisition card is 10kHz. During the experiment, the fault arc generator is connected in series to the A-phase, B-phase, and C-phase respectively, and the A-phase current is collected. For the experimental scheme of series fault arcs in a three-phase circuit under vibration conditions, see Table 2. The collected current samples are input into the series fault arc recognition model of the improved residual network combined with the convolutional block attention module (CBAM-ResNet-18) to obtain the detection model confusion matrix, see Figure 8 . Under vibration conditions, the fault arc recognition accuracy of the proposed model can reach 92.19%, showing a high recognition ability for the fault arc data set.

[0058] Table 2 Experimental scheme of series fault arcs In the embodiment of the present application, the t-SNE method is used to reduce the dimension and classify the visualization results of the output of the model, see Figure 9 . Most of the fault arc samples are concentrated in the feature space, and the model has good recognition ability. This method can accurately identify faults in different phases when only detecting the current of one phase, and has good generalization and classification performance.

[0059] S103: Determine the phase selection and location of the series fault arc according to the prediction result of the current signal.

[0060] Referring to the above rules, determine the phase selection and location of the series fault arc according to the prediction result of the current signal.

[0061] The method for detecting series fault arcs in a three-phase circuit under vibration conditions in the embodiment of the present application can timely and accurately detect the fault arcs occurring in the three-phase motor and the load line of the frequency converter and perform fault phase location, improving the operation reliability of the three-phase motor and the frequency converter circuit and preventing electrical fire accidents caused by series fault arcs.

[0062] In addition, an electromagnetic bin wall vibrator and a voltage regulating vibration feeding controller are used to generate a specific vibration environment to construct a series fault arc under vibration conditions. This fills the gap in the lack of research on series fault arcs under vibration conditions in existing detection methods, enabling the effective capture and identification of fault characteristics under vibration interference. Moreover, by analyzing only one-phase current signal, the series fault arc occurring in any one phase of a three-phase motor can be accurately detected. The number of current sensors required is small, facilitating engineering implementation. There is no need to artificially preprocess the loop current signal, thus eliminating the subjectivity in signal analysis and fault feature extraction, making this method have high recognition accuracy, good versatility, and anti-interference ability. By improving the residual network fault arc recognition model, the total number of parameters of the model can be significantly reduced, simplifying the computational complexity of the model and reducing the calculation time of the recognition model. Introducing a hybrid attention mechanism into the proposed improved residual network fault arc recognition model can focus on the important features in the input data and further optimize the weights of the important features.

[0063] It should be noted that for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0064] The above is the introduction of the method embodiments. The following further illustrates the solution of this application through device embodiments.

[0065] Figure 3 The block diagram of a three-phase circuit series fault arc detection device under vibration conditions according to an embodiment of the present application is shown. The device includes: A signal acquisition module 301, configured to acquire the current signals of a three-phase circuit under vibration conditions during the operation of a three-phase motor and a frequency converter load; A current signal prediction module 302, configured to input the current signals into a pre-trained fault arc recognition model and output the prediction result of the current signals; A phase prediction module 303, configured to determine the phase selection and positioning of the series fault arc according to the prediction result of the current signals.

[0066] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the described modules can refer to the corresponding processes in the foregoing method embodiments and will not be repeated here.

[0067] Figure 4The figure shows a schematic structural diagram of a terminal device or a server suitable for implementing the embodiments of the present application.

[0068] As Figure 4 shown, the terminal device or the server includes a central processing unit (CPU) 401, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 402 or the program loaded from the storage section 408 into the random access memory (RAM) 403. In the RAM 403, various programs and data required for the operation of the terminal device or the server are also stored. The CPU 401, the ROM 402, and the RAM 403 are connected to each other via a bus 404. The input / output (I / O) interface 405 is also connected to the bus 404.

[0069] The following components are connected to the I / O interface 405: an input section 406 including a keyboard, a mouse, etc.; an output section 407 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc. and a speaker, etc.; a storage section 408 including a hard disk, etc.; and a communication section 409 including a network interface card such as a LAN card, a modem, etc. The communication section 409 performs communication processing via a network such as the Internet. A drive 410 is also connected to the I / O interface 405 as required. A removable medium 411, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 410 as required, so that the computer program read from it can be installed into the storage section 408 as required.

[0070] Specifically, according to the embodiments of the present application, the above method flow steps can be implemented as a computer software program. For example, the embodiments of the present application include a computer program product, which includes a computer program carried on a machine-readable medium, and the computer program includes program codes for performing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication section 409, and / or installed from the removable medium 411. When the computer program is executed by the central processing unit (CPU) 401, the above functions defined in the system of the present application are executed.

[0071] It should be noted that the computer-readable medium shown in this application can be a computer-readable signal medium, a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of a computer-readable storage medium can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. And in this application, a computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0072] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram can represent a module, a program segment, or a part of code that contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks can occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of dedicated hardware and computer instructions.

[0073] The units or modules involved in the embodiments described in this application can be implemented in software or in hardware. The described units or modules can also be provided in a processor. Among them, the names of these units or modules do not, in some cases, constitute a limitation on the units or modules themselves.

[0074] As another aspect, this application also provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiments; or may exist separately without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the foregoing programs are executed by one or more processors, they implement the methods described in this application.

[0075] The above description is only a preferred embodiment of this application and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of the application involved in this application is not limited to the technical solutions formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the foregoing application concept. For example, the technical solutions formed by mutually replacing the above features with technical features having similar functions (but not limited to) described in this application.

Claims

1. A method for detecting arc faults in a three-phase circuit under vibration conditions, characterized in that: include: During the operation of the three-phase motor and inverter load, the current signal of the three-phase circuit under vibration conditions is collected; Inputting the current signal into a pre-trained fault arc identification model, and outputting a prediction result of the current signal; The phase selection and positioning of the series fault arc is determined according to the prediction result of the current signal.

2. The method according to claim 1, characterized in that The arc fault identification model is trained in the following way: Collecting a preset number of historical current signal samples and corresponding arc voltage signal samples of the series circuit of the three-phase motor and the inverter load line; According to the arc voltage signal samples, the corresponding current signal samples are divided into normal samples and fault samples, labels are assigned to each type of sample data, and a data set containing series fault arcs is constructed; Divide the data into training set and test set according to the preset ratio, and construct the input matrix with the data samples in the training set and test set; The pre-built residual neural network and hybrid attention mechanism fault arc recognition model is trained and tested using the training set and test set constructed as input matrices; The trained arc fault recognition model is obtained.

3. The method according to claim 2, characterized in that The fault arc recognition model includes an initial convolutional layer, a residual layer, a batch normalization layer, and a fully connected layer; The initial convolution layer is used to capture the features of the local area, and can also reduce the size of the output data while extracting the features through the convolution operation; The residual layer uses skip connections to allow the input signal to be directly transmitted to the subsequent layers, thereby alleviating the gradient vanishing problem in deep network training; The batch normalization layer is used to accelerate training and improve model stability and performance; The fully connected layer is used for feature calculation.

4. The method according to claim 3, characterized in that The initial convolution layer has an input channel of 1, an output channel of 64, a convolution kernel size of 7, a step size of 2, and a calculation formula of: ； Among them, C_in represents the number of input channels, K represents the convolution kernel size, C_out is the number of output channels, and 1 is the bias term.

5. The method according to claim 3, characterized in that: There are 4 residual layers, each of which contains 2 residual blocks, with a total of 8 residual blocks, and each residual block contains two 3×3 convolutional layers.

6. The method according to claim 1, characterized in that The prediction result is a vector with a shape of 4×1. If the largest index value in the vector is 0, it indicates that the data sample is a normal class sample; if the largest index value in the vector is 1, it indicates that the data sample is a phase A fault class sample; if the largest index value in the vector is 2, it indicates that the data sample is a phase B fault class sample; if the largest index value in the vector is 3, it indicates that the data sample is a phase C fault class sample.

7. The method according to claim 2, characterized in that Also includes: Convert the collected current signal into a voltage signal; The collecting of a preset number of historical current signal samples of the series circuit of the three-phase motor and the inverter load line comprises: The collected historical current signals of a preset number of series circuits of three-phase motors and inverter load lines are converted into voltage signals as samples.

8. A three-phase circuit series fault arc detection device under vibration conditions, characterized in that: include: A signal acquisition module is used to collect current signals of the three-phase circuit under vibration conditions during the operation of the three-phase motor and the inverter load; A current signal prediction module, used for inputting the current signal into a pre-trained fault arc recognition model and outputting a prediction result of the current signal; The phase prediction module is used to determine the phase selection and positioning of the series fault arc according to the prediction result of the current signal.

9. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images