Quasi-periodic radiation phenomenon detection method and system of Zhangheng-1 satellite based on DETR

Abstract

The present invention discloses a method and system for detecting quasi-periodic radiation phenomena of Zhangheng-1 satellite based on DETR, the method comprising: converting an input image into high-level semantic features through a convolutional neural network to obtain a feature map; multiplying channel weights with corresponding elements of the feature map to obtain a feature map for final output; mapping the feature map for final output into a sequence form, then inputting the encoder, and adding position coding; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; mapping features to a prediction space through multiple fully connected layers to predict the category and position of the target. Through the processing scheme disclosed in the present invention, the accuracy and detection speed are improved, and it can be deployed in satellite equipment.

Description

Technical Field

[0001] The present invention relates to the field of detection technology, and in particular to a method and system for detecting quasi-periodic radiation phenomena of a Zhangheng-1 satellite based on DETR. Background Art

[0002] Ionospheric quasi-periodic radiation is a typical extremely low frequency / very low frequency electromagnetic fluctuation phenomenon that propagates in the form of whistler waves. It has important scientific research value and has become a research hotspot in the academic community in the past three decades. At present, relevant research has gained a preliminary understanding of ionospheric quasi-periodic radiation: the source region may be located near the geomagnetic equatorial plane of the magnetosphere, the frequency range is approximately in the range of 500Hz-4kHz, the intensity of the fluctuation has significant periodic changes, the duration is 10-80, it mainly appears during the day, usually in a period of relatively calm geomagnetic fields, and is most easily observed at locations with L shell values ​​greater than 3 in terms of spatial distribution, and its period is approximately consistent with the Pc3-Pc5 geomagnetic pulsation.

[0003] Electromagnetic field observation is an important means to fully understand the quasi-periodic radiation phenomenon of the ionosphere in near-Earth space. We already have a certain understanding of quasi-periodic radiation, but due to the limitations of processing large amounts of electromagnetic field data from satellites, none of them can provide a deep understanding of this phenomenon. Before overcoming these problems, a comprehensive and in-depth study of a large amount of ionospheric quasi-periodic emissions is needed. At present, there are ionospheric electromagnetic field observation data, and CSES has accumulated five years of observation data. However, in the face of such a large amount of electromagnetic field data, the traditional means of selecting electromagnetic wave phenomena relies on manual identification methods, which have the problems of low detection efficiency and easy missed detection. With the accumulation of satellite observation data, a large number of space phenomena hidden in the observation data are in urgent need of fast, accurate and intelligent extraction methods.

[0004] Since its launch on February 2, 2018, CSES has been operating stably in orbit for more than five years, comprehensively monitoring the global space electromagnetic field, ionospheric plasma environment, etc., and obtaining a large amount of electromagnetic field waveform data. These massive electromagnetic information carries information on electromagnetic disturbances from the sun to the lithosphere, and has important scientific research and practical application value. However, in the face of these large amounts of electromagnetic field data, traditional data processing and research methods are inefficient and have very limited information mining capabilities, so they have not been fully studied so far. Therefore, we cross-integrate advanced artificial intelligence technologies and apply them to this field to achieve the ability to quickly and accurately extract target information from electromagnetic big data. Therefore, it is necessary to explore artificial intelligence methods for quasi-periodic radiation phenomena to solve the problem of automatic detection of a large number of phenomena in the database, as well as the problem of near real-time detection in subsequent orbital satellite observations.

[0005] DETR (Detection Transformer) is an end-to-end object detector based on the Transformer architecture. Compared with traditional object detection methods, DETR adopts a completely different approach to solve the object detection problem, and can directly predict the location and category of the object from the input image without using traditional object detection components such as manually designed anchor boxes and non-maximum suppression.

[0006] The following is the general process of DETR to achieve target detection:

[0007] Encoder: The input image is passed through a pre-trained convolutional neural network (usually ResNet or other CNN models) as an encoder to extract image features. The output of the encoder is a series of feature vectors that contain the semantic information of each position in the image.

[0008] Decoder: The decoder is a module based on the Transformer structure, which contains a self-attention layer and several multi-head attention layers. The input of the decoder is the output feature vector of the encoder, which is processed to generate the result of target detection.

[0009] Object Queries: The decoder initializes some special object query vectors to guide the decoder to find objects at specific locations. These object query vectors usually contain predefined location and category information.

[0010] Self-Attention: The decoder’s self-attention mechanism allows global associations between decoders, allowing them to see the entire input feature map when generating results. The self-attention mechanism maps the input features to a new representation, allowing semantic relationships between different positions to be encoded.

[0011] Object classification and location prediction: The decoder aligns the location in the feature map with the object query vector through a self-attention mechanism to predict the category and location of the object. This can be achieved by performing classification and regression tasks on the feature map.

[0012] Matching process (Bipartite Matching): The predicted target is matched with the real target through methods such as the Hungarian algorithm, and the loss function is calculated during the training process to supervise the prediction accuracy of the model.

[0013] Directly output results: Finally, DETR directly outputs information such as predicted target boxes and categories without the need for post-processing steps (such as non-maximum suppression) to further filter and optimize the results.

[0014] However, the density of quasi-periodic radiation phenomena varies greatly, the background is complex, and there are targets of different sizes. When facing these problems, DETR's accuracy and speed cannot meet the requirements, and a lightweight model is required to be deployed on satellite equipment.

[0015] It can be seen that the above existing detection methods still have inconveniences and defects in use, and are in urgent need of further improvement. How to create a new detection method has become a goal that the industry urgently needs to improve. Summary of the invention

[0016] In view of this, an embodiment of the present disclosure provides a method for detecting quasi-periodic radiation phenomena of the Zhang Heng-1 satellite based on DETR, which at least partially solves the problems existing in the prior art.

[0017] In a first aspect, an embodiment of the present disclosure provides a method for detecting quasi-periodic radiation phenomena of the Zhangheng-1 satellite based on DETR, the method comprising the following steps:

[0018] The input image is converted into high-level semantic features through a convolutional neural network to obtain a feature map;

[0019] Based on the efficient channel attention module, the input feature map is globally averaged pooled, and then a convolution operation is performed, and the channel weight is obtained through an activation function, and the channel weight is multiplied by the corresponding element of the feature map to obtain the final output feature map;

[0020] The feature map of the final output is mapped into a sequence form, then input into the encoder, and position encoding is added; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain the representation information of the feature, and the feature is mapped to the prediction space through multiple fully connected layers to predict the category and position of the target.

[0021] According to a specific implementation of the embodiment of the present disclosure, the method further includes: adding physical constraints on the proton cyclotron frequency during prediction.

[0022] According to a specific implementation of the embodiment of the present disclosure, the physical constraint of the proton cyclotron frequency is added during prediction, including:

[0023] The proton cyclotron frequency of the predicted orbit is calculated from the observed magnetic field strength based on the following formula:

[0024]

[0025] Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass;

[0026] Interpolate the latitude value and frequency value to the number of pixels to obtain the position of the image where the pixel is located; the horizontal axis is the latitude and the vertical axis is the frequency;

[0027] When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

[0028] According to a specific implementation of an embodiment of the present disclosure, the convolutional neural network is an EfficiectNetV2 network.

[0029] According to a specific implementation of the embodiment of the present disclosure, the step of converting the input image into high-level semantic features through a convolutional neural network includes:

[0030] Acquiring global features through deformable convolution in an encoder; wherein the encoder is also provided with a multi-layer deformable sparse attention fusion mechanism;

[0031] Distinct individual features are parsed through the decoder.

[0032] According to a specific implementation of the embodiment of the present disclosure, the obtaining of global features through deformable convolution in the encoder includes:

[0033] K offsets are calculated and added to the reference point to obtain the coordinates of the K focus points.

[0034] In a second aspect, an embodiment of the present disclosure provides a quasi-periodic radiation phenomenon detection system for Zhang Heng-1 satellite based on DETR, the system comprising:

[0035] A feature extraction module configured to convert an input image into high-level semantic features through a convolutional neural network to obtain a feature map; and,

[0036] Based on the efficient channel attention module, the input feature map is globally averaged pooled, and then a convolution operation is performed, and the channel weight is obtained through an activation function, and the channel weight is multiplied by the corresponding element of the feature map to obtain the final output feature map;

[0037] A prediction module is configured to map the feature map of the final output into a sequence form, and then input it into an encoder, and add position encoding; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain feature representation information, and the feature is mapped to a prediction space through multiple fully connected layers to predict the category and position of the target;

[0038] According to a specific implementation of the embodiment of the present disclosure, the system further includes:

[0039] The constraint module is configured to add physical constraints on the proton cyclotron frequency when predicting, including:

[0040] The proton cyclotron frequency of the predicted orbit is calculated from the observed magnetic field strength based on the following formula:

[0041]

[0042] Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass;

[0043] Interpolate the latitude value and frequency value to the number of pixels to obtain the position of the image where the pixel is located; the horizontal axis is the latitude and the vertical axis is the frequency;

[0044] When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

[0045] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0046] at least one processor; and,

[0047] a memory communicatively connected to the at least one processor; wherein,

[0048] The memory stores instructions that can be executed by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor implements the Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method based on DETR as described in any one of the first aspect or any one of the implementations of the first aspect.

[0049] In a fourth aspect, an embodiment of the present disclosure further provides a non-transitory computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by at least one processor, the at least one processor executes the Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method based on DETR in the aforementioned first aspect or any implementation of the first aspect.

[0050] In the fifth aspect, the embodiments of the present disclosure also provide a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method based on DETR in the aforementioned first aspect or any implementation of the first aspect.

[0051] The DETR-based Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method in the disclosed embodiment can improve the detection accuracy and speed of DETR when the density of quasi-periodic radiation phenomena varies greatly, the background is complex, and there are targets of different sizes, and can be deployed in satellite equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The above is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0053] Figure 1 A schematic flow chart of a method for detecting quasi-periodic radiation phenomena of Zhang Heng-1 satellite based on DETR provided in an embodiment of the present disclosure;

[0054] Figure 2 A flowchart of a method for detecting quasi-periodic radiation phenomena of Zhang Heng-1 satellite based on DETR provided in an embodiment of the present disclosure;

[0055] Figure 3 A schematic diagram of a process of obtaining global features through an encoder in a Transformer by a DETR provided in an embodiment of the present disclosure;

[0056] Figure 4 A schematic diagram of a flow chart of a feature extraction process performed by an encoder through a multi-layer attention fusion mechanism provided in an embodiment of the present disclosure;

[0057] Figure 5 A schematic diagram of a proton cyclotron frequency provided in an embodiment of the present disclosure;

[0058] Figure 6 A schematic diagram of a result of detection on a test set provided by an embodiment of the present disclosure;

[0059] Figure 7 A schematic diagram of a detection result on a new data set provided by an embodiment of the present disclosure;

[0060] Figure 8 A schematic diagram of the structure of a quasi-periodic radiation phenomenon detection system for Zhang Heng-1 satellite based on DETR provided in an embodiment of the present disclosure; and

[0061] Fig. 9 A schematic diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0062] The embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0063] The following describes the embodiments of the present disclosure through specific examples, and those skilled in the art can easily understand other advantages and effects of the present disclosure from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present disclosure, rather than all of the embodiments. The present disclosure can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present disclosure. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present disclosure.

[0064] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein may be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present disclosure, it should be understood by those skilled in the art that an aspect described herein may be implemented independently of any other aspect, and two or more of these aspects may be combined in various ways. For example, any number of aspects described herein may be used to implement a device and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein may be used to implement this device and / or practice this method.

[0065] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the aspects described may be practiced without these specific details.

[0066] The embodiment of the present invention provides a method for detecting quasi-periodic radiation phenomena of Zhang Heng-1 satellite based on DETR, comprising the following steps: collecting electric field waveform data of the ELF band of Zhang Heng satellite EFD data, and transforming to obtain a time-frequency diagram; selecting significant quasi-periodic radiation phenomena in the time-frequency diagram, and annotating data to form a data set; dividing the data set into a training set and a test set; building a neural network, inputting the annotated data of the training set into the network for training until the test requirements of the test set are met, and obtaining a quasi-periodic radiation intelligent detection model; using the quasi-periodic radiation intelligent detection model to detect the time-frequency diagram to be detected, and obtaining a quasi-periodic radiation detection result.

[0067] The quasi-periodic radiation intelligent detection model is used to detect the time-frequency diagram to be detected to obtain the quasi-periodic radiation detection result, including the following steps:

[0068] First, the image is input into a convolutional neural network. Here, the EfficiectNetV2 network is selected, which can convert the input image into high-level semantic features for use by the decoder module. Then an efficient channel attention (ECA) module is added, which can help us better extract the required features in complex backgrounds.

[0069] After adding the position encoding, the improved Transformer encoder and decoder structure is input. The decoder aligns the position in the feature map with the object query vector through the self-attention mechanism to predict the category and position of the target. The fully connected layer maps the feature vector to the output space of the target detection. Through a series of fully connected operations, activation functions and parameterized transformations, the final target detection result is generated.

[0070] A physical constraint on the proton cyclotron frequency was then added to the prediction to eliminate some erroneous prediction results.

[0071] Figure 1 A schematic diagram of the process of the method for detecting quasi-periodic radiation phenomena of the Zhang Heng-1 satellite based on DETR provided in an embodiment of the present disclosure.

[0072] Figure 2 For Figure 1 Corresponding flowchart of the DETR-based detection method for quasi-periodic radiation phenomena of the Zhangheng-1 satellite.

[0073] like Figure 1 As shown, at step S110, the input image is converted into high-level semantic features through a convolutional neural network to obtain a feature map.

[0074] In an embodiment of the present invention, the convolutional neural network is an EfficiectNetV2 network.

[0075] In an embodiment of the present invention, the conversion of the input image into high-level semantic features through a convolutional neural network includes: obtaining global features through deformable convolution in an encoder; wherein the encoder is also provided with a multi-layer deformable sparse attention fusion mechanism; and parsing obvious individual features through a decoder.

[0076] In the embodiment of the present invention, the acquiring of global features through deformable convolution in the encoder includes: calculating K offsets, and adding the K offsets to a reference point to obtain coordinates of K focus points.

[0077] More specifically, if Figure 3As shown, DETR obtains global features through the encoder in the Transformer, and the decoder parses obvious individual features, and has a high detection accuracy for overlapping targets during target detection. Deformable convolution is an effective way to focus on sparse spatial positioning. DeformableDETR combines the best sparse spatial sampling method of Deformable convolution with the relational modeling capability of Transformer, which not only accelerates the convergence of DETR, but also reduces the complexity of the problem. Therefore, the deformable attention mechanism is introduced into the model of the present invention. The key problem of the original Transformer attention is that it pays attention to all spatial positions. The Deformable attention module only pays attention to a small number of key sampling points near the reference point, regardless of the spatial size of the feature map. The query is used to calculate K offsets, which are then added to the reference point to calculate K "coordinates of the focus point", rather than H*W points of the entire feature map (where H is the height of the feature map, W is the width of the feature map, and H*W is all points of the feature map).

[0078] like Figure 4 As shown, in the model proposed in the present invention, the global sampling adopted by Multi-Head self-Attention in the original Transformer Encoder is replaced by the variable attention mechanism (Def-Attention) proposed in Deformable: a sparse sampling strategy to extract more refined features with less computational effort.

[0079] In addition, designing a multi-layer attention fusion mechanism in the Encoder can improve feature learning ability and enhance feature extraction effect.

[0080] More specifically, the process proceeds to step S120.

[0081] In step S120, target detection results are extracted from the high-level semantic features based on an efficient channel attention module; wherein the target detection results include position encoding.

[0082] Next, go to step S130.

[0083] At step S130, the feature map of the final output is mapped into a sequence form, then input into the encoder, and position encoding is added; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain the representation information of the feature, and the feature is mapped to the prediction space through multiple fully connected layers to predict the category and position of the target.

[0084] In the embodiment of the present invention, the position code is a fixed sinusoidal function.

[0085] In an embodiment of the present invention, the object query vector is a set of learnable vectors, each of which represents a potential target object or target category; these vectors are usually initialized to random values ​​and then optimized through back propagation during the training process.

[0086] The choice of prediction space is closely related to the characteristics of the dataset. During training, the model learns to adjust parameters in the prediction space to generate predictions that match the true target categories and attributes.

[0087] In an embodiment of the present invention, the method further comprises: adding physical constraints on the proton cyclotron frequency during prediction.

[0088] In an embodiment of the present invention, the physical constraint of the proton cyclotron frequency is added during prediction, including: calculating the proton cyclotron frequency of the predicted orbit by observing the magnetic field strength based on the following formula:

[0089]

[0090] Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass;

[0091] Interpolate the latitude value and frequency value to the number of pixels to obtain the position of the image where the pixel is located; the horizontal axis is the latitude and the vertical axis is the frequency;

[0092] When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

[0093] Table 1:

[0094]

[0095] More specifically, the proton cyclotron frequency of the predicted orbit was calculated using the magnetic field strength observed by the CDSM of the HPM in Zhang Heng-1.

[0096] The vertical axis of the image is 0-2.5kHZ from the bottom to the top, and the horizontal axis is the latitude value of about -69° to 69°. The latitude value and frequency value are interpolated to the number of pixels, so that the position of the pixel in the image can be obtained through the latitude value and frequency value (the proton cyclotron frequency value of the current latitude value can be calculated, and this frequency is reflected in the image as the number of pixels on the vertical axis, that is, the coordinate y, and the frequency value is the coordinate x). When the (x1, y1) of the center of the prediction box is obtained, it is verified whether the value of y1 at the x1 position is greater than the pixel value where the proton cyclotron frequency is located. If it is less, the prediction box is discarded.

[0097]

[0098] Satellite equipment can obtain the magnetic field strength B, and the proton charge q is 1.60217733×10 -19 C and the proton mass m is 1.672621637×10 -27 KG, such as Figure 5 As shown, the red line is the proton cyclotron frequency.

[0099] Table 2:

[0100]

[0101] Table 2 is a comparison of the reasoning speed and accuracy with other algorithms. It can be seen that the present invention is significantly superior to other existing algorithms in terms of reasoning speed and accuracy.

[0102] like Figure 6 The results are shown in Figure 1.

[0103] Compared to the original DETR, the proposed method improves on the accuracy by 1%. In addition, the number of parameters is reduced by 17%, making the model size only 34.27M. In terms of detection speed, the proposed model outperforms other algorithms and shows similar performance to YOLOv3. In summary, these findings support the conclusion that the proposed method provides superior performance in detecting quasi-periodic radiation events, not only surpassing the state of the art in terms of accuracy, but also providing significant detection speed. The combination of improved accuracy, reduced number of parameters, and enhanced detection speed positions this new model as a robust and efficient solution for the task of QP event detection.

[0104] In this study, we started with the original DETR model and performed comparative experiments on the same dataset at each improvement step. Through these experiments, the performance of the enhancements proposed by this application was evaluated and analyzed. By using the mAP metric, the results obtained from these experiments were measured and compared. The mAP-based evaluation verified that the improvements introduced by this application effectively enabled the model to extract the characteristics of quasi-periodic radiation more accurately. By conducting these comparative experiments on a consistent dataset, we were able to demonstrate the effectiveness of the modifications proposed by this application in enhancing the model to accurately capture the unique characteristics of quasi-periodic radiation.

[0105] like Figure 7 Shown are the results of detection on a completely new dataset.

[0106] In order to evaluate the practical applicability of the model of this application in future scenarios, the automatic detection model proposed in this application was used to detect quasi-periodic radiation events on data collected in June 2023. In 1918 images, 164 quasi-periodic radiation events were detected, of which 87 were correctly identified and 32 quasi-periodic radiation events were not identified. The results in June 2023 showed an accuracy of 53.0% and a recall of 81.5%. These findings show that the model of this application accurately identified most QP events.

[0107] Figure 8 The quasi-periodic radiation phenomenon detection system 800 of the Zhangheng-1 satellite based on DETR provided by the present invention is shown, including a feature extraction module 810, a prediction module 820 and a conversion module 830.

[0108] The feature extraction module 810 is used to convert the input image into high-level semantic features through a convolutional neural network to obtain a feature map; and,

[0109] Based on the efficient channel attention module, the input feature map is globally averaged pooled, and then a convolution operation is performed, and the channel weight is obtained through an activation function, and the channel weight is multiplied by the corresponding element of the feature map to obtain the final output feature map;

[0110] The prediction module 820 is used to map the feature map of the final output into a sequence form, which is then input into the encoder and positional encoding is added; the encoder globally models the sequence of feature map mappings through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain feature representation information, and the features are mapped to the prediction space through multiple fully connected layers to predict the category and position of the target.

[0111] In an embodiment of the present invention, the system further includes:

[0112] The constraint module is configured to add physical constraints on the proton cyclotron frequency when predicting, including:

[0113] The proton cyclotron frequency of the predicted orbit is calculated from the observed magnetic field strength based on the following formula:

[0114]

[0115] Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass;

[0116] Interpolate the latitude value and frequency value to the number of pixels to obtain the position of the image where the pixel is located; the horizontal axis is the latitude and the vertical axis is the frequency;

[0117] When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

[0118] See also Fig. 9 The present disclosure also provides an electronic device 90, which includes:

[0119] at least one processor; and,

[0120] a memory communicatively connected to the at least one processor; wherein,

[0121] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method based on DETR in the aforementioned method embodiment.

[0122] The embodiments of the present disclosure also provide a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the DETR-based Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method in the aforementioned method embodiment.

[0123] The embodiments of the present disclosure also provide a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer executes the DETR-based Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method in the aforementioned method embodiment.

[0124] Reference below Fig. 9, which shows a schematic diagram of the structure of an electronic device 90 suitable for implementing the embodiment of the present disclosure. The electronic device in the embodiment of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Fig. 9 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0125] like Fig. 9 As shown, the electronic device 90 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage device 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the electronic device 90 are also stored. The processing device 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0126] Typically, the following devices may be connected to the I / O interface 905: input devices 906 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 907 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 908 including, for example, a magnetic tape, a hard disk, etc.; and communication devices 909. The communication device 909 may allow the electronic device 90 to communicate with other devices wirelessly or by wire to exchange data. Although the electronic device 90 with various devices is shown in the figure, it should be understood that it is not required to implement or have all the devices shown. More or fewer devices may be implemented or have alternatively.

[0127] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 909, or installed from a storage device 908, or installed from a ROM 902. When the computer program is executed by the processing device 901, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed.

[0128] It should be noted that the computer-readable medium disclosed above may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may 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 the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, in which a computer-readable program code is carried. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium may also be any computer readable medium other than a computer readable storage medium, which may 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 the computer readable medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0129] The computer-readable medium may be included in the electronic device, or may exist independently without being installed in the electronic device.

[0130] The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains at least two Internet Protocol addresses; sends a node evaluation request including the at least two Internet Protocol addresses to a node evaluation device, wherein the node evaluation device selects an Internet Protocol address from the at least two Internet Protocol addresses and returns it; receives the Internet Protocol address returned by the node evaluation device; wherein the obtained Internet Protocol address indicates an edge node in a content distribution network.

[0131] Alternatively, the computer-readable medium carries one or more programs, and when the one or more programs are executed by the electronic device, the electronic device: receives a node evaluation request including at least two Internet Protocol addresses; selects an Internet Protocol address from the at least two Internet Protocol addresses; and returns the selected Internet Protocol address; wherein the received Internet Protocol address indicates an edge node in a content distribution network.

[0132] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0133] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0134] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The name of a unit does not limit the unit itself in some cases. For example, the first acquisition unit may also be described as a "unit for acquiring at least two Internet Protocol addresses".

[0135] It should be understood that various parts of the present disclosure may be implemented in hardware, software, firmware, or a combination thereof.

[0136] The above is only a specific implementation of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present disclosure should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A method for detecting quasi-periodic radiation phenomena of Zhangheng-1 satellite based on DETR, characterized in that: The method comprises the following steps: The input image is converted into high-level semantic features through a convolutional neural network to obtain a feature map; Based on the efficient channel attention module, the input feature map is globally averaged pooled, and then a convolution operation is performed, and the channel weight is obtained through an activation function, and the channel weight is multiplied by the corresponding element of the feature map to obtain the final output feature map; The feature map of the final output is mapped into a sequence form, and then input into the encoder, and position encoding is added; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain the representation information of the feature, and the feature is mapped to the prediction space through multiple fully connected layers to predict the category and position of the target; The method further includes: adding physical constraints on the proton cyclotron frequency during prediction, including: The proton cyclotron frequency of the predicted orbit is calculated from the observed magnetic field strength based on the following formula: Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass; According to the number of pixel points interpolated from the latitude value and the frequency value, the position of the image where the pixel point is located is obtained; wherein the horizontal axis is the latitude and the vertical axis is the frequency; When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

2. The method for detecting quasi-periodic radiation phenomena of Zhangheng-1 satellite based on DETR according to claim 1, characterized in that: The convolutional neural network is an EfficiectNetV2 network.

3. The method for detecting quasi-periodic radiation phenomena of Zhangheng-1 satellite based on DETR according to claim 1, characterized in that: The input image is converted into high-level semantic features through a convolutional neural network, including: The global features are obtained through deformable convolution in the encoder; wherein the encoder is also provided with a multi-layer deformable sparse attention fusion mechanism.

4. The method for detecting quasi-periodic radiation phenomena of Zhangheng-1 satellite based on DETR according to claim 3 is characterized in that: The method of obtaining global features through deformable convolution in the encoder includes: K offsets are calculated and added to the reference point to obtain the coordinates of the K focus points.

5. A DETR-based Zhangheng-1 satellite quasi-periodic radiation phenomenon detection system, characterized in that: The system comprises: A feature extraction module configured to convert an input image into high-level semantic features through a convolutional neural network to obtain a feature map; and, Based on the efficient channel attention module, the input feature map is globally averaged pooled, and then a convolution operation is performed, and the channel weight is obtained through an activation function, and the channel weight is multiplied by the corresponding element of the feature map to obtain the final output feature map; A prediction module is configured to map the feature map of the final output into a sequence form, and then input it into an encoder, and add position encoding; the encoder globally models the sequence of feature map mapping through a multi-layer self-attention mechanism and a feedforward neural network; based on the self-attention mechanism in the decoder, the position in the feature map of the final output is interacted with the object query vector to obtain feature representation information, and the feature is mapped to a prediction space through multiple fully connected layers to predict the category and position of the target; The constraint module is configured to add physical constraints on the proton cyclotron frequency when predicting, including: The proton cyclotron frequency of the predicted orbit is calculated from the observed magnetic field strength based on the following formula: Among them, ω is the proton cyclotron rate; q is the proton charge; B is the magnetic field strength; m is the proton mass; According to the number of pixel points interpolated from the latitude value and the frequency value, the position of the image where the pixel point is located is obtained; wherein the horizontal axis is the latitude and the vertical axis is the frequency; When the position of the image is at the center coordinate of the prediction box, verify whether the value of the vertical coordinate is greater than the pixel value of the proton cyclotron frequency at the horizontal coordinate position of the center of the prediction box; if it is less than, discard the prediction box.

6. An electronic device, characterized in that: The electronic device includes: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor. When the instructions are executed by the at least one processor, the at least one processor executes the DETR-based Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which, when executed by at least one processor, cause the at least one processor to execute the DETR-based Zhang Heng-1 satellite quasi-periodic radiation phenomenon detection method as described in any one of claims 1 to 4.

Images