Double-domain combined narrow-band radar moving target classification and identification method and related product

Abstract

The invention discloses a double-domain combined narrow-band radar moving target classification and identification method and related products. Processing a target echo signal received by the radar to obtain a distance Doppler spectrum; obtaining a micro-Doppler time-frequency spectrum based on a target echo signal received by a radar; respectively cutting the distance Doppler spectrum and the micro Doppler time-frequency spectrum to obtain distance Doppler domain characteristic data and micro Doppler domain characteristic data; and inputting the distance Doppler domain feature data and the micro Doppler domain feature data into a pre-constructed RFICNet to obtain a target classification result. According to the method, dual-domain data features of the distance Doppler domain and the micro-Doppler domain are combined, the method does not depend on complex echo data preprocessing operation and excessive prior physical assumptions, dual-domain joint feature information can make up for the deficiency of feature basis during classification, and the classification recognition precision of the narrow-band radar moving target is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of narrowband radar target recognition, and relates to a dual-domain combined narrowband radar moving target classification and recognition method and related products. Background Art

[0002] As a detection device widely used in military reconnaissance, civil monitoring and air traffic management, narrowband radar has inherent bandwidth limitations and low resolution, which have always been the key factors restricting its performance improvement. Due to the narrow bandwidth of narrowband radar, the target information it can obtain during the detection process is relatively limited, which greatly increases the difficulty of subsequent target classification and identification.

[0003] Among the existing narrowband radar moving target classification and recognition methods, most methods rely on manual feature extraction and parameter estimation based on physical models. These methods usually require complex preprocessing of radar echo signals to extract characteristic parameters that can reflect the characteristics of the target, such as speed, acceleration, shape information, etc. However, due to the resolution limitation of narrowband radar, these characteristic parameters are often not accurate or comprehensive enough, thus affecting the accuracy of classification and recognition.

[0004] In addition, existing classification and recognition methods often have difficulty in fully utilizing the characteristic information of the target in the radar echo when processing narrowband radar data. On the one hand, traditional signal processing methods may ignore certain details that have an important impact on classification and recognition when extracting features; on the other hand, existing machine learning algorithms may face problems such as high computational complexity and poor model generalization ability when processing high-dimensional data, thus limiting their effectiveness in practical applications. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem that various existing classification and recognition methods in the prior art are difficult to fully utilize narrow-band radar target feature information, resulting in inaccurate recognition of moving targets, and to provide a dual-domain joint narrow-band radar moving target classification and recognition method and related products.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: A first aspect of the present invention provides a dual-domain joint narrow-band radar moving target classification and recognition method, comprising the following steps: Process the target echo signal received by the radar to obtain the range Doppler spectrum; Based on the target echo signal received by the radar, a micro-Doppler time-frequency spectrum is obtained; The range Doppler spectrum and the micro-Doppler time-frequency spectrum are cut respectively to obtain the range Doppler domain characteristic data and the micro-Doppler domain characteristic data; Input the range Doppler domain feature data and micro-Doppler domain feature data into the pre-built RFICNet to obtain the classification and recognition results; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

[0007] Further, the target echo signal received by the radar is processed to obtain a range Doppler spectrum, specifically: Perform frequency mixing on the target echo signal received by the radar to obtain a baseband signal; The baseband signal is processed to obtain the range Doppler spectrum.

[0008] Further, the target echo signal received by the radar is mixed to obtain a baseband signal, specifically: For linear frequency modulated pulse radar, the target echo signal received by the radar is digitally down-converted to obtain a baseband signal; For linear frequency modulation continuous wave radar, the target echo signal received by the radar is de-slanted to obtain a baseband signal.

[0009] Further, the micro-Doppler time-frequency spectrum is obtained based on the target echo signal received by the radar, specifically: Perform frequency mixing on the target echo signal received by the radar to obtain a baseband signal; The baseband signal is processed to obtain a range pulse spectrum; Based on the range pulse spectrum, the micro-Doppler time-frequency spectrum is obtained.

[0010] Further, the micro-Doppler time-frequency spectrum is obtained based on the range pulse spectrum, specifically: The slow time dimension data corresponding to the range unit of the target in the range pulse spectrum is selected, and the selected slow time dimension data is transformed into a time-frequency spectrum to obtain a micro-Doppler time-frequency spectrum; the range unit of the target is the range unit of the target obtained after the range Doppler spectrum is processed by constant false alarm rate detection and point trace condensation.

[0011] Furthermore, the range Doppler spectrum and the micro-Doppler time-frequency spectrum are cut respectively to obtain the range Doppler domain feature data and the micro-Doppler domain feature data, specifically: After the range Doppler spectrum is subjected to constant false alarm rate detection, point trace condensation is performed to obtain the range unit and Doppler unit of the target; The range unit and Doppler unit corresponding to the condensed target in the range Doppler spectrum are used as the center to cut the range Doppler spectrum to obtain the range Doppler domain feature data; The frequency data with the largest amplitude in the micro-Doppler time-frequency spectrum is used as the center to cut the micro-Doppler time-frequency spectrum to obtain a cut micro-Doppler time-frequency spectrum; The cropped micro-Doppler time-frequency spectrum is stretched and downsampled to obtain micro-Doppler domain feature data; the micro-Doppler domain feature data has the same dimension as the range Doppler domain feature data.

[0012] A second aspect of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned dual-domain joint narrow-band radar moving target classification and recognition method when executing the computer program.

[0013] A third aspect of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-mentioned dual-domain joint narrow-band radar moving target classification and recognition method.

[0014] A fourth aspect of the present invention provides a computer program product, which includes computer instructions, and the computer instructions instruct a computer to execute the above-mentioned dual-domain joint narrow-band radar moving target classification and recognition method.

[0015] A fifth aspect of the present invention provides a dual-domain joint narrow-band radar moving target classification and recognition system, comprising: The moving target detection module is used to process the target echo signal received by the radar to obtain the range Doppler spectrum; The time-frequency analysis module is used to obtain the micro-Doppler time-frequency spectrum based on the target echo signal received by the radar; A dual-domain data preprocessing module is used to cut the range Doppler spectrum and the micro-Doppler time-frequency spectrum respectively to obtain range Doppler domain feature data and micro-Doppler domain feature data; The target classification and recognition module is used to input the range Doppler domain feature data and the micro-Doppler domain feature data into the pre-built RFICNet to obtain the target classification result; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

[0016] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a dual-domain joint narrow-band radar moving target classification and recognition method. The method combines the range Doppler domain feature data and the micro Doppler domain feature data in a dual-domain joint manner. The range Doppler feature provides information such as the distance, velocity information, amplitude, range dimension extension characteristics, Doppler dimension extension characteristics and background clutter characteristics of the target in the radar line of sight direction, while the micro Doppler time-frequency spectrum feature reveals the Doppler frequency shift difference caused by the relative motion of different parts of the target. The combination of the two can more accurately describe the fusion feature information of the target; can more comprehensively capture the motion characteristics and structural information of the target object, thereby realizing more accurate classification of the moving target; and at the same time, by cutting the range Doppler spectrum and the micro Doppler time-frequency spectrum, key feature data is extracted, thereby effectively reducing the interference of redundant information on the classification and recognition process, not only improving the data processing efficiency, but also enhancing the representativeness of the feature data, so that the classification and recognition result is more reliable.

[0017] Furthermore, the present invention selects data by taking the target centroid in the range Doppler spectrum as the center and the frequency data with the largest amplitude in the micro-Doppler time-frequency spectrum as the center, which can ignore the interference caused by useless position information, reduce the amount of calculation, and avoid the problem of decreased accuracy caused by the sensitivity of the convolutional neural network in RFICNet to the position information of the two-dimensional input feature data.

[0018] Furthermore, the present invention selects a certain size of effective area as data sample for both range Doppler domain and micro-Doppler domain feature data, avoiding interference caused by too much irrelevant data, greatly reducing the input data volume of RFICNet, improving the efficiency of data sample learning and training and recognition and classification, and also making the radar target classification method more versatile. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A schematic diagram of a dual-domain joint narrow-band radar moving target intelligent classification and recognition method provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the structure of the radar fusion information classification and recognition network in an embodiment of the present invention; Figure 3 The confusion matrix of the classification results of the verification experiment of the embodiment of the present invention; Figure 4 It is a block diagram of the dual-domain combined narrow-band radar moving target intelligent classification and recognition method of the present invention; Figure 5 This is a block diagram of the dual-domain combined narrow-band radar moving target intelligent classification and recognition system of the present invention.

[0021] Among them: 501-moving target detection module; 502-time-frequency analysis module; 503-dual-domain data pre-module; 504-target classification and recognition module. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention described and marked in the drawings here can be arranged and designed in various different configurations.

[0023] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0024] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0025] The present invention is further described in detail below in conjunction with the accompanying drawings: See also Figure 4 An embodiment of the present invention discloses a dual-domain joint narrow-band radar moving target classification and recognition method, comprising the following steps: S1, processes the target echo signal received by the radar to obtain the range Doppler spectrum; S2, based on the target echo signal received by the radar, obtain the micro-Doppler time-frequency spectrum; S3, cutting the range Doppler spectrum and the micro-Doppler time-frequency spectrum respectively to obtain range Doppler domain feature data and micro-Doppler domain feature data; S4, inputting the range Doppler domain feature data and the micro-Doppler domain feature data into the pre-built RFICNet to obtain the classification and recognition results; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

[0026] See also Figure 1 An embodiment of the present invention discloses a dual-domain joint narrow-band radar moving target classification and recognition method, comprising the following steps: S1, processing the target echo signal received by the radar to obtain the range Doppler spectrum; based on the range Doppler spectrum, obtaining the range unit and Doppler unit information of the target.

[0027] S101, mixing the target echo signal received by the radar to obtain a baseband signal. In this embodiment, the echo data collected by the FMCW (Frequency Modulated Continuous Wave) radar is used for analysis. For the FMCW radar, it is assumed that the target echo signal received by the radar is , the radar system transmits a signal , then the baseband signal of the target echo signal received by the radar after mixing is:

[0028] in, LPF Represents a low-pass filter.

[0029] S102, using a 2D-FFT (Two-Dimensional Fast Fourier Transform) method to implement MTD (Moving Target Detection) on the baseband signal to obtain a range Doppler spectrum; the range Doppler spectrum is represented by 4096 The matrix is ​​512, where the rows are distance cells, the columns are Doppler frequencies, and the values ​​represent amplitudes.

[0030] S103, the range Doppler spectrum is processed by CFAR (Constant False Alarm Rate) detection and point condensation to obtain the range unit and Doppler unit information of the target. CA-CFAR (Cell Averaging CFAR) is used for constant false alarm rate detection for single target scenes, and SO-CFAR (Smallest Of CFAR) is used for constant false alarm rate detection for multi-target scenes. Point condensation adopts the point condensation method based on the mass center method.

[0031] S2, based on the target echo signal received by the radar, obtain the micro-Doppler time-frequency spectrum of the target echo.

[0032] S201, performing 1D-FFT (One Dimensional Fast Fourier Transform) transformation on the baseband signal in step S101 to obtain a range pulse spectrum.

[0033] S202, based on the target distance unit obtained in step S103, select the slow time dimension data corresponding to the target distance unit in the range pulse spectrum as the target signal , for the target signal Perform time-frequency transformation. In this embodiment, the short-time Fourier transform method is used to perform time-frequency transformation on the target signal to obtain the micro-Doppler time-frequency spectrum of the target echo. The time-frequency transformation method is described as follows:

[0034] in, is the angular frequency, is the analysis window function.

[0035] The micro-Doppler time-frequency spectrum is described as a micro-Doppler domain time-frequency spectrum matrix, wherein the rows of the matrix are time, the columns are Doppler frequencies, and the values ​​represent amplitudes.

[0036] S3, cutting the range Doppler spectrum and the micro-Doppler time-frequency spectrum respectively to obtain range Doppler domain feature data and micro-Doppler domain feature data; S301, for the range Doppler spectrum in step S1, based on the range unit and Doppler unit information of the target obtained in S103, the range unit and Doppler unit corresponding to the condensed target in the range Doppler spectrum are taken as the center, and a fixed-size area is selected for cutting to generate range Doppler domain feature data, wherein the range Doppler domain feature data is described as a matrix, the rows of the matrix are distance units, and the columns are Doppler units, and the matrix is ​​used as a data sample of the range Doppler domain for dual-domain joint identification. The present invention obtains range Doppler domain feature data by selecting a fixed-size effective area with the range unit and Doppler unit corresponding to the condensed target as the center, thereby removing meaningless distance and speed information and excessive background clutter information in the target type judgment, and reducing the size of the two-dimensional data at the same time. The range Doppler domain feature data includes target information such as average energy characteristics, distance extension characteristics, Doppler extension characteristics, and background clutter characteristics of the target.

[0037] S302, for the micro-Doppler time-frequency spectrum generated in step S202, the Doppler frequency data with the largest amplitude is used as the sample center, and a fixed-size area is selected to cut the micro-Doppler time-frequency spectrum to obtain micro-Doppler domain feature data; the micro-Doppler domain feature data is stretched and appropriately downsampled so that the micro-Doppler domain feature data to be input into RFICNet (Radar Fusion Information Classification Network) is consistent in dimension with the range Doppler domain feature data; the micro-Doppler domain feature data that has been stretched and appropriately downsampled is used as a sample of the micro-Doppler domain in the dual-domain joint training.

[0038] The stretching and downsampling processing described in this embodiment results in that the rows of the micro-Doppler domain time-frequency spectrum matrix (i.e., the micro-Doppler domain feature data) have no specific meaning, but can reflect the time dimension information; the columns of the micro-Doppler domain time-frequency spectrum matrix have no specific meaning, but can reflect the frequency dimension information; the stretching operation makes the frequency extension characteristics reflected by the micro-Doppler domain time-frequency spectrum matrix as obvious as possible, and the downsampling operation reduces the sample data size, thereby improving the efficiency of narrow-band radar target classification and recognition.

[0039] S4, constructs a radar fusion information classification and recognition network (RFICNet) for radar target classification and recognition.

[0040] See also Figure 2 ,The RFICNet constructed by the present invention consists of an input module, a feature extraction module, a spatial attention module, and a classification and recognition module.

[0041] The input module is used to normalize the input dual-domain joint two-dimensional data so that the data range is [0,1].

[0042] Two feature extraction modules are used to extract features of different scales. In the first feature extraction module, the convolution kernel size of the convolution layer is 3×3, the number of convolution kernels is 10, the maximum pooling is adopted, the pooling kernel size is 2×2, the pooling operation step is 2, and the ReLU activation function is used. In the second feature extraction module, the convolution kernel size of the convolution layer is 3×3, the number of convolution kernels is 2, the maximum pooling is adopted, the pooling kernel size is 2×2, the pooling operation step is 2, and the ReLU activation function is used.

[0043] The two spatial attention modules are placed after the feature extraction module, and global maximum pooling and global average pooling are performed on the input feature map respectively. The downsampled output after pooling is concat-operated to obtain the context information feature map. After the dimensionality is reduced by 3×3 convolution kernel convolution operation, the spatial attention weight is generated and weighted operation is performed with the feature map of the input spatial attention module.

[0044] Finally, the classification recognition module uses three fully connected layers to classify the input feature map. The first fully connected layer has a size of 256 and uses the ReLU activation function. The second fully connected layer has a size of 256 and uses the ReLU activation function. The third fully connected layer has a size of 64 and uses the ReLU activation function. Finally, the Softmax layer is used to output the probability distribution of category prediction.

[0045] In the RFICNet, the input module performs data normalization processing, uniformly maps the two-dimensional data to the interval [0, 1], so that the features of different dimensions have a certain degree of numerical comparability, which can improve the convergence speed and classification accuracy of the network; the feature extraction module increases the receptive field of neurons through convolution operations, realizes the extraction of semantic information from relatively detailed to abstract, extracts more representative features through pooling, reduces the dimension of features, reduces the size of feature maps, but does not change the number of channels, and the ReLU activation function increases the nonlinear relationship of the network and reduces the interdependence of parameters; the spatial attention module enables the network to adaptively focus on important local areas of the two-dimensional space, thereby improving the effectiveness of feature extraction; the classification recognition module recombines all local features extracted by the feature extraction module and the spatial attention module through a weight matrix, and nonlinearly maps the feature map obtained by the previous process into a one-dimensional feature vector, which contains all the extracted feature information, and the Softmax layer inputs the feature activation value of the input two-dimensional image in each category, and outputs the probability corresponding to each category.

[0046] S5, input the dual-domain joint training data (i.e., range Doppler domain feature data and micro-Doppler domain feature data) into RFICNet (Radar Fusion Information Classification Network) to obtain the trained RFICNet. When classifying and identifying unknown types of targets, the echo signal of the moving target is processed by S1-S3 and then input into the trained RFICNet to obtain the target classification result.

[0047] In order to demonstrate the superiority of the present invention, this embodiment verifies the dual-domain combined narrow-band radar moving target classification and recognition method of the present invention by using the echo data of low-altitude drones, pedestrians and vehicles.

[0048] The narrowband radar system collects target echo data of low-altitude drones, pedestrians and vehicles. After processing through steps S1-S3, training samples of range Doppler domain feature data and training samples of micro-Doppler domain feature data, i.e., dual-domain joint training sample data, are generated. Corresponding category labels are set for the sample data corresponding to each target. At the same time, all the dual-domain joint training samples obtained are divided into training set, validation set and test set in a ratio of 7:2:1 for training and testing the radar fusion information classification and recognition network.

[0049] The UAV flies at a relatively low altitude, making it difficult to distinguish it through the pitch and azimuth information estimated by radar target parameters; at the same time, the speed of the UAV and vehicles is relatively slow, and is similar to the radial speed of pedestrians relative to the radar, making it difficult to distinguish them through simple target speed information.

[0050] In the range Doppler domain training samples, the vehicle echo power is usually large due to the high RCS (Radar Cross Section). According to the radar equation, the relationship between the received echo power and RCS (Radar Cross Section) can be expressed as:

[0051] in, is the radial distance between the radar and the target, is the system loss, is the transmission power, is the radar transmitter gain, is the radar receiver gain, is the emission wavelength, is the signal processing system gain.

[0052] It can be understood that the echo of the vehicle has a more obvious ductility in the distance dimension. This is because the target volume of the vehicle is large and will occupy multiple distance units; but it occupies very few units in the Doppler dimension, because for the vehicle, its parts that can produce micro-Doppler effect, namely the wheels, occupy a small proportion of the entire vehicle body. The target motion state of a moving vehicle can be divided into two parts, namely the translation of the body and the rotation of the wheels. The translation of the body produces the Doppler effect, and the rotation of the wheels produces the micro-Doppler effect. In an actual moving vehicle, it is difficult to observe the micro-Doppler component because the rubber material of the tire has a lower backscattering and the wheel accounts for a much smaller proportion of the size of a conventional vehicle than the body.

[0053] At the same time, the range Doppler domain samples of pedestrians show more irregularities. This is because the RCS (Radar Cross Section) of the human body at different angles has large differences, resulting in variability in amplitude. The movement of body parts when walking causes the echo to have different degrees of extension in the Doppler dimension. At the same time, because pedestrians walk slowly, their echoes often appear near static clutter, which is reflected in the fact that the unit where the pedestrian target echo is located is often accompanied by clutter covering the entire range dimension.

[0054] Compared with pedestrians and vehicles, drones have a smaller RCS (Radar Cross Section), which is reflected in the lower amplitude of the range Doppler domain characteristics.

[0055] In the micro-Doppler domain time-frequency spectrum training samples, the micro-Doppler domain feature data of the vehicle target does not show obvious micro-Doppler phenomenon. There is only an obvious Doppler component caused by the translation of the vehicle body. The sample is reflected as a straight spectral line in the center. At the same time, because the vehicle has a large echo power, the time-frequency spectrum line also presents a higher amplitude, and the obvious energy is diffused in the frequency range.

[0056] The motion state of the UAV target is divided into movement in space and rotation of the rotor. The micro-Doppler effect of the UAV target is generated by the rotation of the rotor. The rotor of the UAV generally has a high reflection coefficient. For traditional narrow-band radars, the point scattering model can be used to analyze its electromagnetic scattering. Each rotor on the UAV is regarded as a scattering center, and each scattering center is independent of each other. Therefore, the total scattering component of the UAV rotor can be regarded as the linear sum of the scattered echoes of each blade.

[0057] Assume that the initial distance from the rotor center to the radar is , the height is , the azimuth relative to the radar line of sight is , the pitch angle is The angular velocity of the rotor is , then the distance from the blade to the rotor center is The radar echo signal generated by a scattering point can be expressed as:

[0058] in, is the wavelength, is the initial phase, The Doppler frequency is caused by the radial velocity of the target relative to the radar.

[0059] The blade echo can be considered as the superposition of various scattering points. The radar echo signal is removed from the carrier frequency and the blade length is Integrating the above formula gives the radar echo signal of a single blade:

[0060] If each rotor is The radar echo signal of a single rotor is:

[0061] It can be seen that the echo signal of a single rotor of the UAV is expressed in the time domain as Function consistent periodic flickering, period in the time domain It depends on the rotor speed and the number of blades and can be expressed as:

[0062] in , is a constant whose value depends on the number of blades of a single rotor ,when When odd , Even number .

[0063] The radar echo signal of a single rotor is converted to the time-frequency domain by short-time Fourier transform to obtain the time-frequency spectrum of a single rotor echo. The time-frequency domain modulation characteristics of a single rotor are manifested as periodic spectral line fluctuations centered on the Doppler frequency shift. The micro-Doppler characteristics of the UAV are reflected as the superposition of the micro-Doppler characteristics of multiple rotors.

[0064] In this embodiment, the UAV adopts a four-rotor UAV, and the Doppler component caused by the spatial movement of the main body of the fuselage occupies the main part of the time-frequency spectrum and has higher energy. Since the UAV rotor has a high rotation speed during the flight, the time-frequency spectrum presents a continuous periodic micro-Doppler feature, and its graphic characteristics are reflected as periodic fluctuations similar to a sine function. At the same time, because the UAV is small in size and has weak backscattering, some micro-Doppler features are submerged in the noise background.

[0065] For the radar echo signal of pedestrian targets, the base point of the spine is defined as the height of the human body. The 58% position of the human body is taken as the origin to establish the human body coordinate system, which is used to describe the human body walking model. In the far-field plane wave case of narrow-band low-resolution radar, the electromagnetic scattering of the human body can also be described according to the point scattering model.

[0066] The motion state of pedestrians can be roughly modeled as 12 scattering points based on the Boulic global walking model, which are the head ( ),trunk( )、Left and right upper arms( , )、Left and right thighs( , )、Left and right thighs( , )、Left and right calves( , )、Left and right foot( , ).

[0067] Simplified, the motion trajectory of the scattering point is used to represent the motion of the human body, and the electromagnetic scattering intensity of each part is represented by its own RCS. The motion of pedestrians can be divided into the overall translation of the human body and the relative motion between each part and the center of mass of the human body. The motion of each part of the human body can be considered to have obvious periodicity. The motion of each part of the human body can be simply described as the left and right swing of the trunk and the swing of the limbs. The following sine function can be used to describe the pedestrian's first Parts of Relative motion on the axis relative to the origin of the established human coordinate system:

[0068] in, For the Micro-movement parts The swing amplitude, For the Micro-movement parts The oscillation frequency, For the Micro-movement parts The initial phase of movement is the initial angle between the corresponding micro-motion part and the torso.

[0069] According to the Boulic empirical model of human walking, it can be considered that the swing frequency of the head and trunk during walking is twice the movement frequency of the human body, and the other micro-movement parts have the same movement frequency as walking, that is:

[0070] Assume that the radar is the origin of the radar coordinate system and the initial distance from the pedestrian's centroid is , the azimuth is , the pitch angle is , the radial velocity of the pedestrian relative to the radar is For each of the above-mentioned micro-movement parts, the positions in the created radar coordinate system can be considered as follows:

[0071] Furthermore, the human body The time delay of the radar echo signal of a micro-movement part can be expressed as:

[0072] Assuming that the mutual occlusion of various parts of the human body can be ignored, the following approximate expression can be used to express the radar echo signal caused by the micro-motion effect of the human body:

[0073] in, Indicates the number of micro-movement parts, i.e., scattering points, indicating the The RCS of the scattering point, is the wavelength of the radar carrier.

[0074] The above radar echo signal is subjected to short-time Fourier transform to obtain the micro-Doppler domain matrix of the pedestrian target echo; it can be understood that in the micro-Doppler domain matrix of this embodiment, the time-frequency characteristics of the pedestrian target are mainly caused by the swing of the torso, which is the main basis for the micro-Doppler domain matrix to be used for classification. The micro-Doppler matrix caused by the swing of the limbs is basically submerged by the background noise due to its small reflection coefficient and susceptibility to occlusion.

[0075] The dual-domain joint training samples of low-altitude drones, pedestrians and vehicles are input into RFICNet for training. During the training process of RFICNet, the learning rate It has a significant impact on the training speed and stability of the model. The learning rate is dynamically adjusted during the training process to control the decrease of the learning rate during the training process, so that the model can more accurately approach the optimal solution in the later stage of training; batch size The optimizer determines the number of samples used in each iteration of the model. The optimizer determines the algorithm used to update the model weights. In this example, the validation set is used to adjust the hyperparameters during training. The initial learning rate is set to ;use The learning rate adjuster will adjust , ; The batch size is set to 32, and small batch gradient descent is used to improve training efficiency and generalization ability.

[0076] After training, the validation set is used for verification, and the classification accuracy is calculated based on the To evaluate the classification accuracy of the method of the present invention, the validation set of this embodiment It is 95.3%, so it can be considered that the dual-domain joint narrow-band radar moving target classification and recognition method of the present invention has good performance.

[0077] The test set is used to verify the generalization ability of RFICNet, and the classification accuracy is Evaluate the performance of the detection model. The expression is as follows:

[0078] Test set It can reach 94.4%. The test results show that the present invention can fit the training data well, has good generalization ability, and shows good detection performance. The test set classification recognition confusion matrix is ​​as follows: Figure 3 As shown, each column of the confusion matrix represents the predicted category, and the total number of each column represents the number of data predicted to be in that category; each row of the confusion matrix represents the true category of the data, and the total number of data in each row represents the number of data instances in that category; the value in each column represents the number of true data predicted to be in that category.

[0079] In summary, the present invention provides a dual-domain joint narrow-band radar moving target classification and recognition method. By performing dual-domain joint training on the feature data of the range Doppler domain and the micro-Doppler domain of the radar echo signal, a radar fusion information classification and recognition network based on a two-dimensional convolutional neural network is constructed. The network achieves good narrow-band radar target classification performance. Compared with the traditional narrow-band radar target classification method, the dual-domain joint narrow-band radar moving target classification and recognition method of the present invention does not require complex artificial feature extraction based on physical assumptions, does not require a large amount of mathematical statistics, and intelligently extracts dual-domain joint features, which can avoid the problem of insufficient feature extraction or too many redundant features for classification, and has higher classification accuracy and efficiency.

[0080] See also Figure 5 One embodiment of the present invention provides a dual-domain joint narrow-band radar moving target classification and recognition system, including: The moving target detection module 501 is used to process the target echo signal received by the radar to obtain the range Doppler spectrum; The time-frequency analysis module 502 is used to obtain a micro-Doppler time-frequency spectrum based on the target echo signal received by the radar; The dual-domain data preprocessing module 503 is used to cut the range Doppler spectrum and the micro-Doppler time-frequency spectrum respectively to obtain the range Doppler domain characteristic data and the micro-Doppler domain characteristic data; The target classification and identification module 504 is used to input the range Doppler domain feature data and the micro-Doppler domain feature data into the pre-built RFICNet to obtain the target classification result; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

[0081] In one embodiment of the present invention, the present invention further provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It is understandable that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device, and can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that more specific examples (non-exhaustive list) of the computer-readable storage medium here include: an electrical connection with one or more wires, a portable 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.

[0082] Computer readable storage media also include data signals propagated in baseband or as part of a carrier wave, which carry readable program codes. Such propagated data signals can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium can also be any readable medium other than a readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the readable storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0083] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0084] The processor may load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the dual-domain combined narrow-band radar moving target classification and recognition method in the above embodiment. An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned dual-domain joint narrow-band radar moving target classification and recognition method when executing the computer program. The memory may be any type of storage device, such as a random access memory (RAM), a read-only memory (ROM), a hard disk drive (HDD), a solid-state drive (SSD), or a cloud storage. In this embodiment, a computer program for implementing a photovoltaic power generation power prediction method is stored in the memory. The processor is responsible for executing the computer program stored in the memory. The processor may be any type of computing device, such as a central processing unit (CPU), a graphics processing unit (GPU), a field programmable gate array (FPGA), or an application-specific integrated circuit (ASIC). In this embodiment, the processor executes a computer program to implement a dual-domain joint narrow-band radar moving target classification and recognition method.

[0085] One embodiment of the present invention provides a computer program product, comprising a series of computer instructions. These instructions are designed to instruct a computer to perform a series of operations to implement dual-domain joint narrow-band radar moving target classification and recognition. These instructions can be loaded into any type of computer system, including but not limited to personal computers, servers, cloud computing platforms, etc., to execute the dual-domain joint narrow-band radar moving target classification and recognition method. The computer instructions in the computer program product of this embodiment have the main function of instructing the computer to perform the following steps to implement the dual-domain joint narrow-band radar moving target classification and recognition method.

[0086] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A dual-domain joint narrow-band radar moving target classification and recognition method, characterized in that: The following steps are involved: Process the target echo signal received by the radar to obtain the range Doppler spectrum; Based on the target echo signal received by the radar, a micro-Doppler time-frequency spectrum is obtained; The range Doppler spectrum and the micro-Doppler time-frequency spectrum are cut respectively to obtain the range Doppler domain characteristic data and the micro-Doppler domain characteristic data; Input the range Doppler domain feature data and micro-Doppler domain feature data into the pre-built RFICNet to obtain the classification and recognition results; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

2. The dual-domain joint narrow-band radar moving target classification and recognition method according to claim 1 is characterized in that: The target echo signal received by the radar is processed to obtain a range Doppler spectrum, specifically: Perform frequency mixing on the target echo signal received by the radar to obtain a baseband signal; The baseband signal is processed to obtain the range Doppler spectrum.

3. The dual-domain joint narrow-band radar moving target classification and recognition method according to claim 2 is characterized in that: The target echo signal received by the radar is mixed to obtain a baseband signal, specifically: For linear frequency modulated pulse radar, the target echo signal received by the radar is digitally down-converted to obtain a baseband signal; For linear frequency modulation continuous wave radar, the target echo signal received by the radar is de-slanted to obtain a baseband signal.

4. The dual-domain joint narrow-band radar moving target classification and recognition method according to claim 1 is characterized in that: The micro-Doppler time-frequency spectrum is obtained based on the target echo signal received by the radar, specifically: Perform frequency mixing on the target echo signal received by the radar to obtain a baseband signal; The baseband signal is processed to obtain a range pulse spectrum; Based on the range pulse spectrum, the micro-Doppler time-frequency spectrum is obtained.

5. The dual-domain joint narrow-band radar moving target classification and recognition method according to claim 4 is characterized in that: The micro-Doppler time-frequency spectrum is obtained based on the range pulse spectrum, specifically: The slow time dimension data corresponding to the range unit of the target in the range pulse spectrum is selected, and the selected slow time dimension data is transformed into a time-frequency spectrum to obtain a micro-Doppler time-frequency spectrum; the range unit of the target is the range unit of the target obtained after the range Doppler spectrum is processed by constant false alarm rate detection and point trace condensation.

6. The dual-domain joint narrow-band radar moving target classification and recognition method according to claim 1 is characterized in that: The range Doppler spectrum and the micro-Doppler time-frequency spectrum are cut respectively to obtain the range Doppler domain feature data and the micro-Doppler domain feature data, specifically: After the range Doppler spectrum is subjected to constant false alarm rate detection, point trace condensation is performed to obtain the range unit and Doppler unit of the target; The range unit and Doppler unit corresponding to the condensed target in the range Doppler spectrum are used as the center to cut the range Doppler spectrum to obtain the range Doppler domain feature data; The frequency data with the largest amplitude in the micro-Doppler time-frequency spectrum is used as the center to cut the micro-Doppler time-frequency spectrum to obtain a cut micro-Doppler time-frequency spectrum; The cropped micro-Doppler time-frequency spectrum is stretched and downsampled to obtain micro-Doppler domain feature data; the micro-Doppler domain feature data has the same dimension as the range Doppler domain feature data.

7. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the dual-domain joint narrow-band radar moving target classification and recognition method according to any one of claims 1 to 6 is implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the dual-domain joint narrow-band radar moving target classification and recognition method according to any one of claims 1 to 6 is implemented.

9. A computer program product, comprising computer instructions, characterized in that: The computer instructions instruct the computer to execute the dual-domain joint narrow-band radar moving target classification and recognition method described in any one of claims 1-6.

10. A dual-domain joint narrow-band radar moving target classification and recognition system, based on the dual-domain joint narrow-band radar moving target classification and recognition method according to claim 1, characterized in that: include: The moving target detection module is used to process the target echo signal received by the radar to obtain the range Doppler spectrum; The time-frequency analysis module is used to obtain the micro-Doppler time-frequency spectrum based on the target echo signal received by the radar; A dual-domain data preprocessing module is used to cut the range Doppler spectrum and the micro-Doppler time-frequency spectrum respectively to obtain range Doppler domain feature data and micro-Doppler domain feature data; The target classification and recognition module is used to input the range Doppler domain feature data and the micro-Doppler domain feature data into the pre-built RFICNet to obtain the target classification result; The pre-built RFICNet includes an input module, two feature extraction modules, two spatial attention modules and a classification recognition module; the input module is connected to the feature extraction module, and each feature extraction module is connected to a spatial attention module; the spatial attention module is connected to the classification recognition module; the input module is used to normalize the input data; the feature extraction module adopts maximum pooling; the spatial attention module includes maximum pooling and average pooling; the classification recognition module adopts three fully connected layers.

Images