Radar raw domain target recognition system based on multi-view antenna distribution design and convolution processing

Abstract

A radar target recognition system based on multi-view antenna distribution design and convolutional processing is proposed. This system increases the received information entropy by designing a multi-view receiving antenna array and establishes a target feature extraction algorithm based on a convolutional neural network. The invention constructs strong differences between target characteristics and clutter / multipath interference characteristics in the radar's original domain data, achieving clutter and multipath interference suppression functions found in mainstream radar target recognition paradigms. Simultaneously, leveraging the powerful feature extraction capabilities of convolutional computation, a feature extraction algorithm is designed to distinguish the characteristics of the target from clutter, multipath interference, and different targets, ultimately achieving high-precision radar target recognition. This invention is expected to address the challenges of complex algorithms in current high-precision radar target recognition, such as clutter suppression, coherence accumulation, and imaging, providing new insights into the field of radar target recognition.

Description

Technical Field

[0001] This invention relates to the field of radar target recognition, and more specifically, to a radar native target recognition system based on multi-view antenna distribution design and convolution processing in the presence of clutter and multipath interference. Background Technology

[0002] Radar Target Recognition (RTR) is a target detection technology that analyzes radar signals to determine the type of target. This process involves extracting key information from radar echoes and matching it with a pre-set target database to achieve accurate target identification and classification.

[0003] Traditional radar target recognition technology relies on analog-to-digital converters to sample radar echoes. After processing such as pulse compression and coherent accumulation, target information carriers such as one-dimensional images, two-dimensional images, and range-Doppler images are obtained. Feature extraction and recognition algorithms are then based on these information carriers for subsequent processing. Current research focuses on the processing of images or other information carriers after radar imaging (A.Deng, N.Qian, S.Hua, J.Wan, Z.Lv, and W.Zou, High-resolution ISAR imaging based on photonic receiving for high-accuracy automatic target recognition, Optics Express, 2022, 30(12):20580-20588.), while the processing of the original radar signal directly remains insufficient.

[0004] In addition to containing target scattering echoes, radar echoes inevitably contain clutter and multipath interference, which form false targets in the target information carrier and seriously affect the target identification results. Therefore, before feature extraction, it is necessary to preprocess the radar echoes, one-dimensional images or two-dimensional images to remove false targets.

[0005] To improve identification accuracy, increasing the number of receiving antennas and optimizing their layout has proven to be an effective strategy. Multi-antenna systems can not only capture target echoes from different perspectives, improving azimuth resolution, but also enhance target features through spatial diversity, suppressing interference and thus improving the signal-to-noise ratio. For example, Gerard Lachapelle et al. in 2016 used array antennas and beamforming technology to effectively reduce clutter and multipath interference (Reference: Vagle N et al., 2016); the German Aerospace Center (DLR) in 2020 used filtering to remove multipath signals by utilizing the difference in signal strength received by multiple antennas (Reference: Zorn S et al., 2020).

[0006] Furthermore, multi-view receiving antennas can further improve detection accuracy. A 2018 study by the Italian Naval Research Center showed that the more uniformly distributed the antenna is and the wider the angle around the target, the higher the azimuth resolution (Reference: Leembo L et al., 2018). However, in practical applications, due to the uncertainty of the target's position, it is difficult to ideally place the antenna around the target. To address this, the University of Quebec in Canada proposed an adaptive radar receiver placement mechanism in the same year, aiming to maximize the signal-to-noise ratio, reduce the bit error rate, and improve positioning accuracy (Reference: Kilani MB et al., 2018).

[0007] Although there has been considerable research on optimizing received signals by changing the distribution of receiving antennas, the following challenges remain: while multi-antenna reception can improve the signal-to-noise ratio, subsequent digital signal processing is still required to completely suppress noise and interference; existing research focuses on simple radar functions, such as early warning and positioning, while research on complex radar target recognition functions is still insufficient; at the same time, methods such as beamforming still require post-processing of the raw radar data when dealing with clutter and multipath interference, which increases processing delay and power consumption. Summary of the Invention

[0008] This invention addresses the aforementioned shortcomings of existing technologies by providing a radar target recognition system based on multi-view antenna distribution design and convolution processing under clutter and multipath interference conditions. It increases the received information entropy by designing a multi-view receiving antenna array and establishes a target feature extraction algorithm based on convolution calculation. After the signal echo containing clutter and multipath interference is received by the multi-view antenna array, signals from different angles interact in the feature extraction module based on convolution calculation to highlight signal features, thereby achieving the purpose of removing clutter and multipath interference and ultimately realizing high-precision radar target recognition.

[0009] The technical solution of the present invention is as follows:

[0010] A radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing, characterized by including:

[0011] Waveform transmitter module, used to generate and transmit radar signals;

[0012] A multi-view antenna array receiver module, comprising multiple receiving antenna arrays and multiple analog-to-digital converters, is used to receive radar signals from multiple views and convert them into digital signals;

[0013] A single-antenna receiver module, comprising a receiving antenna and an analog-to-digital converter, is used to receive radar signals from space and convert them into digital signals;

[0014] A feature extractor module based on a convolutional neural network is used to extract high-dimensional, abstract features from the data output by the multi-view antenna array receiver module and the single antenna receiver module.

[0015] The recognition classifier module is used to classify and recognize targets based on the extracted features;

[0016] The decision module is used to compare the recognition results of the network trained under different antenna configurations and trigger the feedback optimization module.

[0017] The feedback optimizer module is used to analyze the receiving performance of the multi-view antenna array receiver based on the output of the decision module, and optimize the arrangement of the multi-view receiving antenna array.

[0018] The output terminals of the multi-view antenna array receiver module and the single-antenna receiver module are respectively connected to the input terminal of the convolutional neural network-based feature extractor module. The output terminal of the convolutional neural network-based feature extractor module is connected to the input terminal of the recognition classifier module. The output terminal of the recognition classifier module is connected to the input terminal of the decision module. The output terminal of the decision module is connected to the input terminal of the feedback optimizer module. The output terminal of the feedback optimizer module is connected to the input terminal of the multi-view antenna array receiver module, forming a closed-loop optimization system.

[0019] The waveform transmitter is a broadband signal transmitting device based on multiple or a single antenna.

[0020] The multi-view antenna array receiver is a radar data receiving module based on multiple receiving antenna arrays.

[0021] The analog-to-digital converter in the multi-view antenna array receiver is a data acquisition board or an oscilloscope.

[0022] The single-antenna receiver is a radar data receiving module based on a single receiving antenna.

[0023] The feature extractor based on the convolutional neural network is a computer program that implements the convolutional network.

[0024] The identification classifier is a computer program based on nonlinear computation and a fully connected layer.

[0025] The aforementioned feedback optimizer is a computer program that implements algorithms for various multi-antenna array configurations based on different receiving performance characteristics such as anti-interference performance and target feature extraction capability.

[0026] A high-precision radar target recognition method is achieved by utilizing the aforementioned radar in-domain target recognition techniques based on multi-view antenna distribution design and convolutional processing. This method includes two stages: training and applying a feature extractor and a recognition classifier based on a convolutional neural network. The steps are as follows:

[0027] 1) Training phase:

[0028] The convolutional neural network-based feature extractor and classifier can only achieve high-precision identification of the radar original domain signals output by the multi-view antenna array receiver and the single-antenna receiver after training. When the waveform transmitter transmits a broadband signal, the signal propagates through space, and the radar echoes, carrying clutter and multipath interference, are received by the multi-view antenna array receiver and the single-antenna receiver respectively, resulting in two sets of radar original domain data with different antenna configurations. This radar original domain data is used as the training set for the convolutional neural network-based feature extractor and classifier. An optimization algorithm is then used to train the convolutional neural network-based feature extractor and classifier, enabling them to learn the features between different targets and between targets and clutter and multipath interference, thus achieving classification and identification. Simultaneously, the decision-maker compares the recognition results of the network trained on datasets obtained from different antenna configurations. If the recognition accuracy of the network trained on the data received by the multi-view antenna array receiver is lower than that of the network trained on the data received by the single-antenna receiver, the feedback optimizer analyzes the reception performance of the multi-view antenna array receiver and re-optimizes the arrangement of the multi-view receiving antennas according to the algorithm. Otherwise, training is complete, ultimately achieving high-precision radar target recognition.

[0029] 2) Application phase:

[0030] After the waveform transmitter transmits a broadband signal, the multi-view antenna array receiver receives the radar echo, which is accompanied by clutter and multipath interference, propagating through space. The received signal is then input into the convolutional neural network-based feature extractor and classifier, which have been trained during the training phase. The network outputs a high-accuracy recognition result, achieving high-precision radar target identification.

[0031] Based on the above technical features, the present invention has the following advantages:

[0032] 1. A multi-view antenna array is used to receive radar signals contaminated by clutter and multipath interference. By combining multi-view reception with a feature extractor based on a convolutional neural network, the target features are amplified, thereby suppressing clutter and multipath interference.

[0033] 2. The feature extractor and classifier based on convolutional neural networks is an end-to-end, simple, fast and efficient radar target recognition method based on target feature learning. Its feature is that it directly processes the radar original domain signal, avoiding complicated signal processing procedures.

[0034] This invention is based on the design of a multi-view antenna array to construct a strong difference between the target characteristics and the clutter and multipath interference characteristics, forming an innovative method to suppress clutter and multipath interference. This allows for the reception of contaminated radar echoes in the electromagnetic original domain, followed by direct feature extraction based on convolutional neural networks to obtain the key features of the target. Then, a classification and recognition algorithm is executed to achieve high-precision target recognition. Attached Figure Description

[0035] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0036] Figure 1 This is a block diagram of a radar original domain target recognition system based on multi-view antenna distribution design and convolution processing in one embodiment of the present invention.

[0037] Figure 2 This is a flowchart of the feedback optimization module in one embodiment of the present invention.

[0038] Figure 3 This is a schematic diagram of the structure of a feature extractor module and a recognition classifier module based on a convolutional neural network in one embodiment of the present invention.

[0039] Figure 4 This is the training effect of the feature extractor module and the recognition classifier module based on the convolutional neural network in one embodiment of the present invention. Detailed Implementation

[0040] The following is a detailed description of an embodiment of the radar original domain target recognition technology based on multi-view antenna distribution design and convolution processing of the present invention: This embodiment is implemented under the premise of the technical solution of the present invention, and provides detailed implementation methods and specific operation processes, but the protection scope of the present invention should not be limited to the following embodiment.

[0041] Figure 1 This diagram illustrates the module composition of a radar in-domain target recognition system based on multi-view antenna distribution design and convolution processing, according to an embodiment of the present invention. The present invention mainly includes the following operations:

[0042] Design the placement of each antenna in a multi-antenna receiver array;

[0043] The transmitting antenna transmits a signal to the target. After receiving the signal echoes contaminated by clutter and multipath interference, each receiving antenna processes the data to obtain the joint received data.

[0044] Set up a single receiving antenna to receive the echo from the target detected by the transmitting antenna under the same conditions;

[0045] A neural network based on convolutional computation is used to process the joint received data obtained from the multi-antenna receiving array and the data received from a single antenna, respectively.

[0046] The identification results obtained from the data obtained by the multi-antenna receiving array and the single receiving antenna are compared and judged.

[0047] If the multi-antenna receiving array performs worse or fails to meet accuracy requirements, the spatial arrangement of the multi-antenna receiving array should be re-optimized; otherwise, the array can be used to achieve high-precision target identification.

[0048] like Figure 1 As shown, a radar in-domain target recognition system based on multi-view antenna distribution design and convolutional processing comprises the following components: a waveform transmitter module 1, a multi-view antenna array receiver module 2, a single-antenna receiver module 3, a feature extractor module 4 based on a convolutional neural network, a recognition classifier module 5, a decision module 6, and a feedback optimizer module 7. Its construction process is divided into two stages: a training stage and an application stage. In the training stage, the multi-view antenna array receiver module 2 is optimized, and the feature extractor module 4 and the recognition classifier module 5 based on the convolutional neural network are trained to enable high-precision target recognition. In the application stage, the optimized multi-view antenna array receiver module 2, the trained feature extractor module 4, and the recognition classifier module 5 are applied to perform high-precision radar target recognition. In this embodiment, the waveform transmitter module 1 generates a linear frequency modulated signal with a bandwidth of 1 GHz using an arbitrary waveform generator, then modulates it to a 10 GHz frequency band using a mixer, and transmits the signal into space through an antenna. The receiving antenna of the multi-view antenna array receiver module 2 is positioned at a 90° angle with the line connecting the transmitter and the target as the axis of symmetry. The receiving antenna of the single-antenna receiver module 3 is placed next to the transmitting antenna. The feature extractor module 4 based on a convolutional neural network is a convolutional neural network, and the recognition classifier module 5 is a fully connected layer; its structural diagram is shown below. Figure 3As shown in the figure, the feature extractor module 4 based on the convolutional neural network includes an input layer, two one-dimensional convolutional layers, and two max-pooling layers; the recognition classifier module 5 consists of a fully connected layer and an output layer. The two modules used in this embodiment are implemented in a personal computer. The decision module 6 is a comparator based on recognition accuracy implemented in a personal computer. It determines whether the receiving antenna arrangement in the multi-view antenna array receiver module 2 is optimal by comparing the accuracy of the data obtained from the multi-view antenna array receiver module 2 and the data obtained from the single-antenna receiver module 3 after training. The processing flowchart of the feedback optimizer module 7 is shown below. Figure 2 As shown, the optimal antenna arrangement is determined using the simulated annealing algorithm. The objective function (system energy) is jointly determined by the peak-to-side-lobe ratio (PSLR) of the received signal and the full width at half maximum (FWHM) of the target lobe. PSLR corresponds to the intensity of clutter and multipath interference on the received signal; a higher PSLR indicates lower clutter and multipath interference. FWHM corresponds to the clarity of the radar system's target detection; a lower FWHM indicates a narrower target lobe, resulting in more detectable scattering points and a higher azimuth resolution. First, initial coefficients a = 1 and b = -1 are set, and the objective function (system energy) E = a·PSLR + b·FWHM is set. The optimal antenna arrangement can then be calculated using the simulated annealing algorithm. If, after passing through the decision module 6, the current antenna arrangement is deemed to result in low recognition accuracy, the analysis determines whether the low accuracy is due to clutter and multipath interference (case 1) or unclear target features (case 2). For case 1, increase the proportion of peak-to-sidelobe ratio (PSLR) in the objective function, i.e., increase 'a'; for case 2, increase the proportion of full width at half maximum (FWHM) in the objective function, i.e., decrease 'b' (increase the absolute value of 'b').

[0049] The module connections and module functions in the two stages of the construction process are described below.

[0050] 1) Training phase:

[0051] First, a microwave source generates a 10GHz carrier signal, and an arbitrary waveform generator generates a 1GHz linear frequency modulated signal. This signal is modulated onto the 10GHz carrier frequency by a mixer. Second, two receiving antennas are set, and the receiving antennas of the multi-view antenna array receiver module 2 are positioned at 90° angles to the line connecting the transmitter and the target. Third, two different types of targets, circular and triangular, are set. By rotating the target, the antennas can detect target features at different angles and obtain a dataset. For each type of target, one target echo is obtained every 0.18°, resulting in a total of 4000 sets of data for both types of targets. These are randomly divided into training and test sets at a 3:1 ratio. Fourth, the collected training set is input into the convolutional neural network-based feature extractor module 4 and the recognition classifier module 5 for training. Specifically, for the data obtained by the single-antenna receiver module 3, the received data from the two antennas are concatenated to obtain a one-dimensional vector. For the single-antenna receiver module 3, since it receives data from only one antenna, to ensure that the input data to the neural network has the same dimension, the received data from the single antenna is copied and concatenated to obtain a one-dimensional vector. The Adam optimization algorithm is chosen to reduce the absolute error between the network input and the network reference, and the learning rate of the optimization algorithm is set to 0.0001. During 30,000 iterations, the network parameters are continuously adjusted. Finally, the feature extractor module 4 and the recognition classifier module 5 based on the convolutional neural network, after training, can learn the features of different targets and achieve high-precision radar target recognition.

[0052] Figure 4 Figures show the training results of the feature extractor module 4 based on the convolutional neural network and the recognition classifier module 5. Figure (a) shows the training results obtained from data collected by the multi-view antenna array receiver module 2 when the number of receiving antennas is two, and Figure (b) shows the training results obtained from data collected by the single-antenna receiver module 3. It can be seen that when training the network with data from the multi-view antenna array receiver module 2, the network loss can drop to a minimum of 0.0344, and the network converges quickly; while when training the network with data from the single-antenna receiver module 3, the network loss can only drop to a minimum of about 0.159, and the network converges more slowly, only approaching convergence at the end of training.

[0053] 2) Application phase:

[0054] After the waveform transmitter transmits a broadband signal, the multi-view antenna array receiver receives the radar echoes propagating through space, carrying clutter and multipath interference. The received signals are then input into the convolutional neural network-based feature extractor and classifier, which have been trained during the training phase. For a test set of 1000 sets, the recognition accuracy of the network output layer is approximately 99.4%, achieving high-precision radar target recognition.

[0055] Any matters not covered in the above embodiments of the present invention are well-known in the art.

[0056] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing, characterized in that, include: Waveform transmitter module (1) is used to generate and transmit radar signals; The multi-view antenna array receiver module (2) includes multiple receiving antenna arrays and multiple analog-to-digital converters for receiving radar signals from multiple views and converting them into digital signals; The single-antenna receiver module (3) includes a receiving antenna and an analog-to-digital converter for receiving radar signals from space and converting them into digital signals; The feature extractor module (4) based on the convolutional neural network is used to extract high-dimensional, abstract features from the data output by the multi-view antenna array receiver module (2) and the single antenna receiver module (3); The identification classifier module (5) is used to classify and identify targets based on the extracted features; The decision module (6) is used to compare the recognition results of the network trained under different antenna configurations and trigger the feedback optimizer module (7); Feedback optimizer module (7) is used to analyze the receiving performance of the multi-view antenna array receiver based on the output of the decision module (6) and optimize the arrangement array of the multi-view receiving antennas. The output of the multi-view antenna array receiver module (2) and the output of the single antenna receiver module (3) are respectively connected to the input of the feature extractor module (4) based on the convolutional neural network. The output of the feature extractor module (4) based on the convolutional neural network is connected to the input of the recognition classifier module (5). The output of the recognition classifier module (5) is connected to the input of the decision module (6). The output of the decision module (6) is connected to the input of the feedback optimizer module (7). The output of the feedback optimizer module (7) is connected to the input of the multi-view antenna array receiver module (2), forming a closed-loop optimization system.

2. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 1, characterized in that, The waveform transmitter module (1) is a broadband signal transmitting device based on multiple or a single antenna.

3. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 1 or 2, characterized in that, The waveform transmitter module (1) includes an arbitrary waveform generator for generating radar baseband signals, a microwave source for generating high-frequency carrier signals, a mixer for upconverting radar baseband signals to carrier frequencies, an amplifier for boosting signal power in a microwave link, and a transmitting antenna for transmitting radar signals into space.

4. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 1, characterized in that, The multi-view antenna array receiver module (2) is a radar data receiving module based on multiple receiving antenna arrays.

5. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 4, characterized in that, The analog-to-digital converter in the multi-view antenna array receiver module (2) is a data acquisition board or an oscilloscope.

6. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 1, characterized in that, The single-antenna receiver module (3) is a radar data receiving module based on a single receiving antenna.

7. The radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing according to claim 1, characterized in that, The feature extractor module (4) based on the convolutional neural network is a computer program that implements the convolutional neural network; the recognition classifier module (5) is a computer program based on nonlinear calculation and fully connected layers; the feedback optimizer module (7) is a computer program that implements algorithms for various multi-antenna array configurations based on anti-interference performance, target feature extraction capability and other receiving performance.

8. A method for achieving high-precision radar target recognition using the radar original-domain target recognition system based on multi-view antenna distribution design and convolution processing as described in any one of claims 1-7, characterized in that, Includes the following steps: A waveform transmitter transmits a broadband signal, a multi-view antenna array receiver receives the radar echo, and the signal is input into a pre-trained feature extractor and classifier based on a convolutional neural network to achieve high-precision radar target identification. The training of the feature extractor and recognition classifier based on the convolutional neural network includes: After the waveform transmitter module (1) transmits a broadband signal, the signal propagates through space. The radar echo with clutter and multipath interference is received by the multi-view antenna array receiver module (2) and the single antenna receiver module (3) respectively, and two sets of radar original domain data with different antenna configurations are obtained. The radar original domain data is used as the training set for the feature extractor module (4) and the recognition classifier module (5) based on the convolutional neural network. The feature extractor module (4) and the recognition classifier module (5) based on the convolutional neural network are trained using an optimization algorithm, so that the feature extractor module (4) and the recognition classifier module (5) based on the convolutional neural network can learn the features between different targets and between the target and clutter and multipath interference, and realize classification and recognition. The decision module (6) compares the recognition results of the network trained on the datasets obtained from different antenna configurations. If the recognition accuracy of the network trained on the data received by the multi-view antenna array receiver module (2) is lower than that of the network trained on the data received by the single antenna receiver module (3), the feedback optimizer module (7) will analyze the receiving performance of the multi-view antenna array receiver module (2) and re-optimize the arrangement array of the multi-view receiving antennas according to the algorithm; otherwise, the training is completed, and high-precision radar target recognition is finally achieved.

Images