Arc detection method, device, electronic equipment and storage medium

Abstract

The present invention discloses an arc detection method, device, electronic device and storage medium, and relates to the technical field of power system safety monitoring, wherein the method comprises: real-time collection of electric quantity data, preprocessing and data fusion of the electric quantity data, obtaining input data, obtaining an arc data set and constructing an arc detection model based on a hole convolution and a residual network, training the arc detection model according to the arc data set to obtain a trained arc detection model, performing arc detection on the input data through the arc detection model to obtain a detection result, and performing early warning processing and model optimization according to the detection result. The present invention can timely discover and handle potential problems in the power grid through real-time monitoring and early warning processing, reduce the occurrence rate of faults and power outage time, improve the operation efficiency and reliability of the power grid, and solve the problem that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient.

Description

Technical Field

[0001] The present invention relates to the technical field of power system safety monitoring, and in particular to an arc detection method, device, electronic equipment and storage medium. Background Art

[0002] In the power systems of modern industry and daily life, the generation of arcs is a safety hazard that cannot be ignored. Arcing is a gas discharge phenomenon that usually occurs when current passes through the air gap between two conductive objects. Since arcs can generate extremely high temperatures, up to thousands of degrees Celsius, it may not only quickly damage electrical equipment, but also cause fires, resulting in casualties and huge economic losses. Therefore, arc detection plays a vital role in the safety management of power systems.

[0003] However, the existing technology has obvious defects: first, the frequency band coverage is incomplete, resulting in insufficient sensitivity to certain types of arc detection; second, it is easily affected by environmental noise and electromagnetic interference, which limits the detection accuracy; third, the algorithm is highly complex and requires a lot of calculations, which affects real-time performance; fourth, in a complex noise environment, it is difficult to effectively distinguish between noise and arc characteristics, which is prone to false alarms or missed alarms. These defects limit the widespread application and effectiveness of arc detection technology.

[0004] Therefore, there is an urgent need for an arc detection method that can accurately identify arc characteristics and be widely applicable to practical application scenarios, thereby improving efficiency and safety. Summary of the invention

[0005] The embodiments of the present invention provide an arc detection method to solve the problem that the prior art cannot cope with complex practical application scenarios and has insufficient arc detection range, detection accuracy and response capability. The technical solution is as follows:

[0006] According to one aspect of the present invention, an arc detection method comprises: real-time acquisition of electrical quantity data; the electrical quantity data comprises instantaneous current value, instantaneous voltage value, frequency value, current and voltage waveform in time domain, and current and voltage waveform in frequency domain; preprocessing and data fusion are performed on the electrical quantity data to obtain input data; the preprocessing comprises filtering, denoising and normalization; an arc data set is acquired and an arc detection model based on a hole convolution and a residual network is constructed, and the arc detection model is trained according to the arc data set to obtain a trained arc detection model; arc detection is performed on the input data by the arc detection model to obtain a detection result, and early warning processing and model optimization are performed according to the detection result.

[0007] In one of the embodiments, the electric quantity data is preprocessed and data fused to obtain input data by the following steps: the electric quantity data is preprocessed by median filtering and standard normalization to obtain preprocessed data; the preprocessed data is fused and integrated into input data.

[0008] In one of the embodiments, the structure of the arc detection model includes a main body, branches and a tail, the main body includes a dilated convolution module and a residual network module, the dilated convolution module includes a dilated convolution, BN and an activation function, the residual network module includes a residual structure, a dilated convolution module, a feature fusion module and a pooling layer, the branch includes a dilated convolution module, upsampling, downsampling and a bidirectional weighted feature pyramid, and the tail includes a dilated convolution module and a dilated convolution.

[0009] In one of the embodiments, the arc detection model is trained according to the arc data set to obtain a trained arc detection model by the following steps: feature extraction, feature learning and feature fusion are performed on the arc data set by the arc detection model to obtain an output result; whether the set conditions are met is determined according to the error between the output result and the true result and the number of model iterations, and if the set conditions are met, a trained arc detection model is obtained; if the set conditions are not met, a loss value is calculated according to the error between the output result and the true result by a loss function, and the learning rate, parameters, thresholds and weights are adjusted according to the loss value before continuing the training.

[0010] In one of the embodiments, performing arc detection on the input data through the arc detection model to obtain a detection result is achieved through the following steps: identifying the time period in which the input data is located through the time period identification mechanism in the arc detection model; the time period includes peak period, trough period and peace period; the time period is used to indicate the intensity of electricity consumption; and selecting the corresponding network branch for arc detection according to the time period in which the input data is located to obtain a detection result.

[0011] In one of the embodiments, early warning processing and model optimization based on the detection results are achieved through the following steps: using an improved Kalman filter to process the detection results, processing and judging the detection results, and if an arc occurs, an alarm is issued, otherwise no alarm is issued; using the detection results to optimize the detection strategy and model parameters of the arc detection model.

[0012] In one embodiment, the detection result is processed using an improved Kalman filter, and the processing and judgment of the detection result are achieved by the following steps: The calculation formula for processing the detection result using the improved Kalman filter is as follows:

[0013] , , ;

[0014] in, represents the output of the model at time t, represents the result of the model output at time t-1, is a nonlinear state transfer function, Indicates compression operation. represents the Kalman gain, represents the observation vector, Indicates the result after processing; if If the value is not less than 1, it means that an arc is generated, otherwise it means that no arc is generated.

[0015] According to one aspect of the present invention, an arc detection device includes: a data acquisition module for collecting electrical quantity data in real time; a data processing module for preprocessing and data fusion of the electrical quantity data to obtain input data; a model training module for acquiring an arc data set and constructing an arc detection model based on a hole convolution and a residual network, and training the arc detection model according to the arc data set to obtain a trained arc detection model; an arc detection module for performing arc detection on the input data through the arc detection model to obtain a detection result, and performing early warning processing and model optimization according to the detection result.

[0016] According to one aspect of the present invention, an electronic device includes at least one processor and at least one memory, wherein the memory stores computer-readable instructions; the computer-readable instructions are executed by one or more of the processors, so that the electronic device implements the arc detection method as described above.

[0017] According to one aspect of the present invention, a storage medium stores computer-readable instructions thereon, wherein the computer-readable instructions are executed by one or more processors to implement the arc detection method as described above.

[0018] The beneficial effects brought by the technical solution provided by the present invention are:

[0019] In the above technical scheme, the present invention first collects electric quantity data in real time, pre-processes and fuses the electric quantity data to obtain input data, obtains an arc data set and constructs an arc detection model based on a void convolution and a residual network, trains the arc detection model according to the arc data set to obtain a trained arc detection model, performs arc detection on the input data through the arc detection model to obtain a detection result, performs early warning processing and model optimization according to the detection result, and can timely discover and deal with potential problems in the power grid through real-time monitoring and early warning processing, reduce the occurrence rate of faults and power outage time, and improve the operation efficiency and reliability of the power grid. The automated monitoring and early warning system can reduce the frequency and intensity of manual inspections and reduce operation and maintenance costs. Timely early warning and processing can reduce the inconvenience and losses caused to users by power outages and other reasons, and improve user satisfaction and trust, thereby effectively solving the problems that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in describing the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention, and those skilled in the art can obtain other drawings based on these drawings without creative work.

[0021] Figure 1 is a flow chart of an arc detection method according to an exemplary embodiment;

[0022] Figure 2 It is a flowchart of an arc detection method in an application scenario;

[0023] Figure 3 yes Figure 2 Schematic diagram of data acquisition and preprocessing in corresponding application scenarios;

[0024] Figure 4 yes Figure 2 Schematic diagram of the process of network model training in the corresponding application scenario;

[0025] Figure 5 yes Figure 2 A schematic diagram of the structure of the network model in the corresponding application scenario;

[0026] Figure 6 is a block diagram of an arc detection device according to an exemplary embodiment;

[0027] Figure 7 is a hardware structure diagram of an electronic device according to an exemplary embodiment;

[0028] Figure 8It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0029] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0030] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present disclosure refers to the presence of the features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be intermediate elements. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0031] The present invention provides an arc detection method, which realizes effective monitoring and early warning processing of arc faults in power grids through real-time collection of electric quantity data, data preprocessing and fusion, construction of advanced arc detection models, timely early warning processing and model optimization, etc., solves the problem that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient. The arc detection method is suitable for arc detection devices, which can be electronic devices. The arc detection method in the embodiment of the present invention can be applied to a variety of scenarios, such as arc detection.

[0032] See also Figure 1 , an embodiment of the present invention provides an arc detection method, which is applicable to electronic equipment.

[0033] In the following method embodiments, for ease of description, the execution subject of each step of the method is taken as an electronic device as an example for illustration, but this does not constitute a specific limitation.

[0034] like Figure 1 As shown, the method may include the following steps:

[0035] Step 110: collect power data in real time.

[0036] The electrical quantity data includes instantaneous current value, instantaneous voltage value, frequency value, current and voltage waveform in the time domain, and current and voltage waveform in the frequency domain.

[0037] Step 130 , preprocessing and data fusion are performed on the electric quantity data to obtain input data.

[0038] Among them, preprocessing includes filtering, denoising and normalization.

[0039] In a possible implementation, the power data is preprocessed by using median filtering and standard normalization to obtain preprocessed data, and the preprocessed data is fused and integrated into input data.

[0040] Step 150, obtaining an arc data set and constructing an arc detection model based on a dilated convolution and a residual network, and training the arc detection model according to the arc data set to obtain a trained arc detection model.

[0041] In one possible implementation, the structure of the arc detection model includes a main body, branches and a tail. The main body includes a dilated convolution module and a residual network module. The dilated convolution module includes a dilated convolution, BN and an activation function. The residual network module includes a residual structure, a dilated convolution module, a feature fusion module and a pooling layer. The branch includes a dilated convolution module, upsampling, downsampling and a bidirectional weighted feature pyramid. The tail includes a dilated convolution module and a dilated convolution.

[0042] In one possible implementation, the arc detection model is used to perform feature extraction, feature learning and feature fusion on the arc data set to obtain an output result. The error between the output result and the true result and the number of iterations of the model are used to determine whether the set conditions are met. If the set conditions are met, a trained arc detection model is obtained; if the set conditions are not met, the loss value is calculated according to the error between the output result and the true result through the loss function, and the learning rate, parameters, thresholds and weights are adjusted according to the loss value before continuing the training.

[0043] Step 170, arc detection is performed on the input data through the arc detection model to obtain a detection result, and early warning processing and model optimization are performed according to the detection result.

[0044] In a possible implementation, the time period of the input data is identified through a time period identification mechanism in the arc detection model, and a corresponding network branch is selected to perform arc detection according to the time period of the input data to obtain a detection result.

[0045] The time periods include peak period, off-peak period and peace time, and the time periods are used to indicate the intensity of electricity consumption.

[0046] In one possible implementation, an improved Kalman filter is used to process the detection results. The detection results are processed and judged. If an arc occurs, an alarm is issued, otherwise no alarm is issued. The detection results are used to optimize the detection strategy and model parameters of the arc detection model.

[0047] In a possible implementation, the calculation formula for processing the detection result using the improved Kalman filter is as follows:

[0048] , , ;

[0049] in, represents the output of the model at time t, represents the result of the model output at time t-1, is a nonlinear state transfer function, Indicates compression operation. represents the Kalman gain, represents the observation vector, Indicates the result after processing;

[0050] like If the value is not less than 1, it means that an arc is generated, otherwise it means that no arc is generated.

[0051] Through the above process, the present invention first collects electric quantity data in real time, pre-processes and fuses the electric quantity data to obtain input data, obtains an arc data set and constructs an arc detection model based on a void convolution and a residual network, trains the arc detection model according to the arc data set to obtain a trained arc detection model, performs arc detection on the input data through the arc detection model to obtain a detection result, performs early warning processing and model optimization according to the detection result, and can timely discover and handle potential problems in the power grid through real-time monitoring and early warning processing, reduce the occurrence rate of faults and power outage time, and improve the operation efficiency and reliability of the power grid. The automated monitoring and early warning system can reduce the frequency and intensity of manual inspections and reduce operation and maintenance costs. Timely early warning and processing can reduce the inconvenience and loss of users caused by power outages and other reasons, and improve user satisfaction and trust, thereby effectively solving the problem that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient.

[0052] In one application scenario, Figure 2 A flow chart showing the application of an arc detection method in a smart meter and an arc detection system is presented.

[0053] like Figure 2As shown, the arc detection system includes an information acquisition unit, a data processing unit, a model training unit, an arc detection unit, and an arc warning unit.

[0054] The information collection unit collects electricity data in real time through smart meters. The collected data include instantaneous current value, instantaneous voltage value and other electricity data. Other electricity data include frequency, current and voltage in the time domain and specific waveforms of current and voltage in the frequency domain.

[0055] like Figure 3 As shown, the data processing unit collects real-time data from the acquisition unit, and then normalizes the original data using filtering, denoising and normalization methods, and finally performs feature fusion on the processed data and inputs it into the network model.

[0056] Among them, filtering and denoising use median filtering, and normalization uses standard normalization method.

[0057] Specifically, data preprocessing is implemented as follows:

[0058] ;

[0059] in, Indicates the data collected by the information collection unit, Represents the centralized data, =[221.2253, 10.4764, 50.3592, 220.9756, 15.2585, 220.3989, 21.4983].

[0060] Furthermore, the covariance matrix is ​​calculated:

[0061] ;

[0062] in, represents the covariance matrix after the data is centralized, and n=7 represents the sample data of the information collection unit.

[0063] Furthermore, the eigenvalues ​​of the covariance matrix are calculated and the eigenvector , where the eigenvalue and the eigenvector The following relations are satisfied:

[0064] ;

[0065] in, .

[0066] Further, the data preprocessing matrix W and the preprocessed data are calculated , the formula is as follows:

[0067] , ;

[0068] in, represents the eigenvector matrix, represents a diagonal matrix, Represents the data preprocessing matrix, which is applied to the centralized data , get the preprocessed data .

[0069] like Figure 4 As shown in the figure, the training process of the network model, i.e., the arc detection model, in the model training unit is demonstrated. First, the network structure is constructed and the network parameters are initialized. Then, the sample data set of arc occurrence and its features are input into the network model so that the network model can learn the state and features when the arc occurs. The network model extracts the features of the sample data set, performs feature learning and feature fusion based on the extracted features, and then outputs the results. Then, the error between the output result of the network model and the true result of the data set is calculated, and the number of iterations and the error are judged. If the set conditions are met, the training ends. If the set conditions are not met, the loss value is calculated according to the error through the loss function, and then the learning rate and parameters are dynamically adjusted according to the loss value, the threshold is adjusted and the weight is updated to enter the next round of training. The set conditions are that the error is less than the threshold or the set number of iterations is reached.

[0070] like Figure 5 As shown in the figure, the network model structure is shown. The model is constructed using dilated convolution and residual network. The main part of the model is composed of dilated convolution, BN (Batch Normalization), loss function and residual network. The branch part is composed of upsampling, downsampling and feature pyramid, and finally a series of dilated convolutions are used.

[0071] Specifically, the main part of the model consists of c=1 dilated convolution modules and m=5 residual network modules. Each dilated convolution module contains dilated convolution, BN and activation function. The residual network module contains residual structure and dilated convolution module, feature fusion module and pooling layer. The dilated convolution module consists of a 32×3×3, r=2 dilated convolution, BN layer and LeakyRelu activation function. The residual network module consists of a pooling layer, a dilated convolution module and a certain number of residual units. The residual unit consists of 2 dilated convolution modules and add. Add is used to add the feature maps after the dilated convolution module.

[0072] As shown in Table 1, the network model has 5 residual network modules, corresponding to the number of residual units of 1, 2, 8, 8, and 4 respectively.

[0073] Table 1 Parameter settings of the main part of the network model

[0074]

[0075] Specifically, the branch part of the model consists of L=17 atrous convolution modules, K=1 upsampling, J=1 downsampling and H=1 bidirectional weighted feature pyramid. The bidirectional weighted feature pyramid is used to allow the network model features to communicate with each other, and then the Concat operation is used to fuse the features and increase the dimension of the features.

[0076] As shown in Table 2, the parameter settings of the branch part of the network model are displayed.

[0077] Table 2 Parameter settings of the branch part of the network model

[0078]

[0079] Specifically, the tail part of the network model consists of P=3 dilated convolution modules and Q=3 dilated convolutions, and the structure of the network model consists of a main body, branches and a tail.

[0080] Furthermore, the arc detection data set processed by the information acquisition unit and the data processing unit is sent to the network model in the model training unit for training and learning to obtain a trained network model.

[0081] Specifically, the dataset is randomly divided into three parts, with the training dataset accounting for 80% of the total dataset, the test dataset accounting for 10% of the total dataset, and the validation dataset accounting for 10% of the total dataset. The default parameters are set at the beginning of model training, with Batch Size set to S=32, the learning rate set to D=0.0001, the number of iterations set to F=5000, and the Adam optimizer selected.

[0082] Furthermore, the arc detection unit inputs the real-time data processed by the data processing unit into the trained network model for real-time monitoring.

[0083] Specifically, Figure 5 As shown in the figure, the time period recognition mechanism built into the network model intelligently identifies the time period of the real-time data according to the historical power consumption pattern, and performs detection in N=3 time periods when detecting arcs. During the peak power consumption period (7:00-11:00, 19:00-23:00), the valley period (23:00-7:00 the next day) and the peacetime period (11:00-19:00), the power intensity used in each time period is different. In order to adapt to different power consumption intensities and patterns, different network branches are selected under different power consumption conditions. The arc detection unit can achieve accurate and efficient detection, minimize false alarms and missed alarms, and the network model has adaptive learning capabilities, and can continuously optimize the detection strategy and model parameters based on new detection data and feedback.

[0084] Furthermore, the arc warning unit is used to process and judge the result output by the arc detection unit, and to give an alarm if an arc occurs, otherwise, no alarm is given.

[0085] Specifically, the output result of the arc detection unit is processed by an improved Kalman filter and a self-learning feedback mechanism. When the system detects that an arc has occurred, an alarm is issued. At the same time, the data collected by the system data unit at this moment is recorded and fed back to the model training unit for model training.

[0086] The process of using improved Kalman filtering to process the output results of the arc detection unit to make the output smoother is as follows:

[0087] , , ;

[0088] in, represents the output of the model at time t, represents the result of the model output at time t-1, is a nonlinear state transfer function , represents the compression operation, Kalman gain , represents the observation vector, Indicates the result after processing. represents the error covariance, represents the process noise covariance, represents the state transfer Jacobian matrix, represents the observation Jacobian matrix.

[0089] Further, The value of is used to judge and update the error covariance matrix ,like If the value is not less than 1, it means that an arc is generated, otherwise it means that no arc is generated.

[0090] Through the above steps, the embodiment of the present invention can obtain the user's electricity data in real time through the real-time monitoring function of smart meters and other equipment, improve data quality and enhance the generalization ability of the model through data preprocessing, and the hole convolution can capture richer context information. The residual network solves the gradient vanishing and network degradation problems in the deep neural network by introducing residual connections, improves the training efficiency and detection accuracy of the model, and performs real-time detection of input data through the trained arc detection model, which can timely detect arc faults and immediately issue early warning signals to avoid fault expansion and loss aggravation, thereby improving user satisfaction and trust, thereby effectively solving the problems that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient.

[0091] The following is an embodiment of the device of the present invention, which can be used to perform the arc detection method involved in the present invention. For details not disclosed in the embodiment of the device of the present invention, please refer to the method embodiment of the arc detection method involved in the present invention.

[0092] See also Figure 6 , an arc detection device 800 is provided in an embodiment of the present invention.

[0093] The arc detection device 800 includes but is not limited to: a data acquisition module 810 , a data processing module 830 , a model training module 850 and an arc detection module 870 .

[0094] The data collection module 810 is used to collect power data in real time.

[0095] The data processing module 830 is used to pre-process and fuse the electric quantity data to obtain input data.

[0096] The model training module 850 is used to obtain an arc data set and construct an arc detection model based on a dilated convolution and a residual network, and train the arc detection model according to the arc data set to obtain a trained arc detection model.

[0097] The arc detection module 870 is used to perform arc detection on input data through an arc detection model to obtain a detection result, and perform early warning processing and model optimization according to the detection result.

[0098] It should be noted that, in the arc detection provided in the above embodiments, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the arc detection device will be divided into different functional modules to complete all or part of the functions described above.

[0099] In addition, the arc detection device and arc detection method provided in the above embodiments belong to the same concept, and the specific manner in which each module performs the operation has been described in detail in the method embodiment, which will not be repeated here.

[0100] Figure 7 The present invention schematically illustrates the structure of an electronic device according to an exemplary embodiment.

[0101] It should be noted that the electronic device is only an example adapted to the present invention and cannot be considered to provide any limitation on the scope of use of the present invention. The electronic device cannot be interpreted as requiring dependence on or having Figure 7 One or more components of exemplary electronic device 2000 are shown.

[0102] The hardware structure of the electronic device 2000 may vary greatly due to differences in configuration or performance, such as Figure 7 As shown, the electronic device 2000 includes: a power supply 210 , an interface 230 , at least one memory 250 , and at least one central processing unit (CPU) 270 .

[0103] Specifically, the power supply 210 is used to provide operating voltage for each hardware device on the electronic device 2000 .

[0104] The interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. Of course, in other examples adapted by the present invention, the interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input-output interface 235, and at least one USB interface 237, etc. Figure 7 As shown, this is not intended to be a specific limitation.

[0105] The memory 250 is a carrier for storing resources, which may be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon include an operating system 251, an application 253 and data 255, etc. The storage method may be temporary storage or permanent storage.

[0106] Among them, the operating system 251 is used to manage and control various hardware devices and application programs 253 on the electronic device 2000 to enable the central processor 270 to calculate and process the massive data 255 in the memory 250. It can be WindowsServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, etc.

[0107] The application program 253 is a computer-readable instruction that performs at least one specific task based on the operating system 251, and may include at least one module ( Figure 7 (not shown), each module may respectively contain computer-readable instructions for the electronic device 2000. For example, the arc detection device may be regarded as an application 253 deployed in the electronic device 2000.

[0108] The data 255 may be signal information, etc., stored in the memory 250 .

[0109] The central processor 270 may include one or more processors and is configured to communicate with the memory 250 through at least one communication bus to read the computer-readable instructions stored in the memory 250, thereby realizing the operation and processing of the mass data 255 in the memory 250. For example, the arc detection method is completed by the central processor 270 reading a series of computer-readable instructions stored in the memory 250.

[0110] In addition, the present invention can also be implemented by hardware circuits or hardware circuits combined with software. Therefore, the implementation of the present invention is not limited to any specific hardware circuits, software, or a combination of the two.

[0111] See also Figure 8 In an embodiment of the present invention, an electronic device 4000 is provided. The electronic device 4000 may include: a desktop computer, a laptop computer, a server, etc. with sensor recognition capability.

[0112] exist Figure 8 In the embodiment, the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0113] The data interaction between the processor 4001 and the memory 4003 can be realized through at least one communication bus 4002. The communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 8 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0114] Optionally, the electronic device 4000 may further include a transceiver 4004, which may be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception, etc. It should be noted that in actual applications, the transceiver 4004 is not limited to one, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0115] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of the present invention. Processor 4001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0116] The memory 4003 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program instructions or codes in the form of instructions or data structures and can be accessed by the electronic device 4000, but is not limited to this.

[0117] Computer-readable instructions are stored in the memory 4003 , and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002 .

[0118] The computer-readable instructions are executed by one or more processors 4001 to implement the arc detection method in the above-mentioned embodiments.

[0119] In addition, an embodiment of the present invention provides a storage medium on which computer-readable instructions are stored. The computer-readable instructions are executed by one or more processors to implement the arc detection method as described above.

[0120] A computer program product is provided in an embodiment of the present invention. The computer program product includes computer-readable instructions. The computer-readable instructions are stored in a storage medium. One or more processors of an electronic device read the computer-readable instructions from the storage medium, load and execute the computer-readable instructions, so that the electronic device implements the arc detection method as described above.

[0121] Compared with the related art, the present invention has the following beneficial effects:

[0122] 1. The present invention first collects electric quantity data in real time, pre-processes and fuses the electric quantity data to obtain input data, obtains an arc data set and constructs an arc detection model based on a void convolution and a residual network, trains the arc detection model according to the arc data set to obtain a trained arc detection model, performs arc detection on the input data through the arc detection model to obtain a detection result, performs early warning processing and model optimization according to the detection result, and can timely discover and handle potential problems in the power grid through real-time monitoring and early warning processing, reduce the occurrence rate of faults and power outage time, and improve the operation efficiency and reliability of the power grid. The automated monitoring and early warning system can reduce the frequency and intensity of manual inspections and reduce operation and maintenance costs. Timely early warning and processing can reduce the inconvenience and loss of users caused by power outages and the like, and improve user satisfaction and trust, thereby effectively solving the problem that the existing technology cannot cope with complex practical application scenarios, and the arc detection range, detection accuracy and response capability are insufficient.

[0123] 2. The present invention can obtain the user's electricity data in real time, including key parameters such as electricity consumption, voltage, current, etc., ensuring the freshness and accuracy of the data, and providing a solid foundation for subsequent preprocessing and arc detection; through data preprocessing, it can eliminate outliers, noise values ​​and redundant values, improve data quality, and enhance the generalization ability of the model; by fusing data from different sources and different types, it can more comprehensively reflect the operating status of the power system, help to discover potential problems and anomalies, and improve the accuracy of arc detection.

[0124] 3. The hole convolution used in the present invention can increase the receptive field and capture richer contextual information without increasing the number of parameters and the amount of calculation. The residual network solves the problems of gradient vanishing and network degradation in deep neural networks by introducing residual connections, thereby improving the training efficiency and detection accuracy of the model. The model is trained using arc data sets, so that the model can learn the characteristics and laws of arc faults. Through a large amount of data training, the model can have higher recognition accuracy and generalization ability.

[0125] 4. The present invention can detect the input data in real time through the trained arc detection model, and can detect arc faults in time. Once an arc fault is detected, the system can immediately send out an early warning signal to remind relevant personnel to handle it, so as to avoid the expansion of the fault and the increase of losses; according to the detection results and actual conditions, the model is continuously optimized and improved. By introducing new data, adjusting model parameters or adopting more advanced algorithms, the detection accuracy and adaptability of the model are continuously improved.

[0126] 5. Through real-time monitoring and early warning processing, potential problems in the power grid can be discovered and dealt with in a timely manner, reducing the occurrence of faults and power outage time, and improving the operating efficiency and reliability of the power grid; the automated monitoring and early warning system can reduce the frequency and intensity of manual inspections and reduce operation and maintenance costs; improve user experience: timely early warning and processing can reduce the inconvenience and losses caused to users by power outages and other reasons, and improve user satisfaction and trust.

[0127] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0128] The above descriptions are only some embodiments of the present invention. It should be pointed out that, for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An arc detection method, characterized in that: The method comprises: Collecting electric quantity data in real time; the electric quantity data includes instantaneous current value, instantaneous voltage value, frequency value, current and voltage waveform in time domain, and current and voltage waveform in frequency domain; Preprocessing and data fusion are performed on the electric quantity data to obtain input data; the preprocessing includes filtering, denoising and normalization; An arc data set is obtained and an arc detection model based on a hole convolution and a residual network is constructed. The arc detection model is trained according to the arc data set to obtain a trained arc detection model; the arc detection model includes a main part, a first branch, a second branch, and a third branch. The first branch, the second branch, and the third branch are all connected to the corresponding tail. The first branch performs arc detection based on the data of power period 1, the second branch performs arc detection based on the data of power period 2, and the third branch performs arc detection based on the data of power period 3. The first branch, the second branch, and the third branch are all composed of 5 hole convolution modules DBL, and each tail includes 1 hole convolution module DBL and 1 hole convolution Dconv connected in sequence. The main part is composed of a hole convolution module DBL, a residual block res1, a residual block res2, a first residual block res8, a second residual block res8, and a residual block res4 connected in sequence. The first residual block res8 is sequentially connected through a bidirectional weighted feature pyramid network BiFPN, a first hole convolution module DBL, and a downsampling module subsa mpling and a first concatenation operation concat module to connect the third branch, the second residual block res8 is connected to the second branch through the second concatenation operation concat module, the residual block res4 is connected to the first branch, the output of the first branch is sequentially input into the second branch through the second hole convolution module DBL and the upsampling module upsampling and the second concatenation operation concat module, the output of the second branch is connected to the first hole convolution module DBL and the downsampling module subsampling, and the first concatenation operation concat module to input the third branch, the hole convolution module DBL includes a hole convolution Dconv, a batch normalization layer BN and an activation function LeakyRelu, the residual block res1, the residual block res2, the first residual block res8, the second residual block res8, and the residual block res4 all include a pooling layer pooling, a hole convolution module DBL and a certain number of residual units Resunit, and the residual unit Resunit includes 2 hole convolution modules DBL and an addition operation add; The time period of the input data is identified by a time period identification mechanism in the arc detection model; the time period includes a peak period, a valley period, and a peace period; the time period is used to indicate the power consumption intensity; According to the time period of the input data, a corresponding network branch is selected to perform arc detection, a detection result is obtained, and early warning processing and model optimization are performed according to the detection result.

2. The arc detection method according to claim 1, characterized in that: The preprocessing and data fusion of the electric quantity data to obtain input data includes: Preprocessing the electric quantity data by using median filtering and standard normalization to obtain preprocessed data; The preprocessed data are fused and integrated into input data.

3. The arc detection method according to claim 1, characterized in that: The step of training the arc detection model according to the arc data set to obtain a trained arc detection model includes: Performing feature extraction, feature learning and feature fusion on the arc data set through the arc detection model to obtain an output result; Judging whether a set condition is met according to the error between the output result and the true result and the number of iterations of the model, and obtaining a trained arc detection model if the set condition is met; If the set conditions are not met, the loss value is calculated according to the error between the output result and the true result through the loss function, and the learning rate, parameters, thresholds and weights are adjusted according to the loss value to continue training.

4. An arc detection device, characterized in that: The device is applied to the arc detection method according to claim 1, and the device comprises: Data acquisition module, used to collect power data in real time; A data processing module, used for preprocessing and data fusion of the electric quantity data to obtain input data; a model training module, used for acquiring an arc data set and constructing an arc detection model based on a hole convolution and a residual network, and training the arc detection model according to the arc data set to obtain a trained arc detection model; The arc detection module is used to perform arc detection on the input data through the arc detection model to obtain a detection result, and perform early warning processing and model optimization according to the detection result.

5. An electronic device, characterized in that: include: at least one processor and at least one memory, wherein: The memory has computer-readable instructions stored thereon; The computer-readable instructions are executed by one or more of the processors so that the electronic device implements the arc detection method according to any one of claims 1 to 3.

6. A storage medium having computer-readable instructions stored thereon, characterized in that: The computer readable instructions are executed by one or more processors to implement the arc detection method according to any one of claims 1 to 3.

Images