High-precision radar target recognition system based on orthogonal waveform design and convolutional neural network

Abstract

A high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network, comprising a radar orthogonal waveform design module, a waveform generation module, a transmitting antenna module, a receiving antenna module, a waveform receiving module and a convolutional neural network module; by designing orthogonal radar transmitting pulse train to increase receiving entropy, so that the radar echo pulse train received by the receiving antenna contains sufficient target information and the target and clutter and multipath interference characteristics present strong difference, and then the convolutional neural network can extract the target key features from the radar echo pulse train received by the antenna while suppressing the clutter and multipath interference. The present application directly processes the radar echo in the electromagnetic domain, efficiently extracts the features, realizes high-precision recognition, integrates the traditional multi-step signal processing into one step, simplifies the process, and may reduce the complexity of the receiver hardware, which is of great significance to improve the performance of electronic information systems.

Description

Technical Field

[0001] This invention relates to radar target recognition technology and deep learning technology, and in particular to a high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network. Background Technology

[0002] Radar technology can detect targets at all times, in all weather conditions, and at long distances, making it an important tool for target identification. Convolutional neural networks (CNNs), due to their superior feature extraction capabilities, have been widely used in target identification tasks. By processing radar echoes to obtain carriers of target information, such as one-dimensional or two-dimensional images, neural networks are then used to extract and identify features from these carriers, thereby outputting the target type. For example, Jinhu Li, Fangzheng Zhang, Yu Xiang, and Shilong Pan, "Towards small target recognition with photonics-based high-resolution radar range profiles," Opt. Express 29, 31574-31581, 2021, used a photonic deskewing radar system to transmit signals with a bandwidth of 8 GHz to detect targets. After pulse compression of the received radar echoes to obtain one-dimensional images, a convolutional neural network was used to extract and identify features from these one-dimensional images, achieving a target identification accuracy of up to 97.16%. Anyi Deng, Na Qian, Shiyu Hua, Jun Wan, Zhenbin Lv, and WeiwenZou, "High-resolution ISAR imaging based on photonic receiving for high-accuracy automatic target recognition," Opt. Express 30, 20580-20588, 2022. This study utilizes a photonic radio frequency direct acquisition radar system to detect complex targets by transmitting signals with a bandwidth of 8 GHz. Multiple received radar echoes are pulse-compressed and coherently accumulated to obtain a two-dimensional image. A convolutional neural network is then used to extract features from the two-dimensional image and perform recognition, achieving a recognition accuracy of 95%. However, these studies were conducted in an ideal environment free of clutter and multipath interference.

[0003] As the electromagnetic environment becomes increasingly complex, radar echoes inevitably contain clutter and multipath interference in addition to target-scattered echoes. These interference factors create false targets in the target information carrier, severely affecting target identification results. Therefore, effective preprocessing is necessary before inputting the target information carrier into a convolutional neural network to remove false targets. However, these preprocessing algorithms are not only highly complex but also have limited applicability to certain types of clutter or interference (M. Azimifar, and M. A. Sebt, Clutter Suppression in Passive ISAR using Compressive Sensing, in IEEE Transactions on Aerospace and Electronic Systems, 2024).

[0004] Therefore, it is of great significance to study a simpler, more efficient and more accurate radar target identification method in clutter and multipath interference environments. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a high-precision radar target recognition system based on orthogonal waveform design and convolutional neural networks. By designing the radar transmitted pulse train as an orthogonal signal set, a strong difference between target characteristics and clutter and multipath interference characteristics is constructed. This enables the convolutional neural network to extract key target features from radar echo pulse trains contaminated by clutter and multipath interference and suppress clutter and multipath interference, thereby achieving high-precision target recognition.

[0006] The present invention is achieved using the following technical solution.

[0007] A high-precision radar target recognition system based on orthogonal waveform design and convolutional neural networks, characterized by including:

[0008] The radar orthogonal waveform design module is used to design orthogonal signal sets to increase the information entropy of radar transmitted signals.

[0009] A waveform generation module, connected to the radar orthogonal waveform design module, is used to generate orthogonal pulse trains designed by the radar orthogonal waveform design module.

[0010] A transmitting antenna module, connected to the waveform generation module, is used to transmit the orthogonal pulse train as an electromagnetic signal.

[0011] The receiving antenna module is used to receive radar echo pulse trains;

[0012] A waveform receiving module, connected to the receiving antenna module, is used to convert the received radar echo pulse train into a digital signal;

[0013] A convolutional neural network module, connected to the waveform receiving module, is used to identify target types based on the converted radar echo pulse train digital signal;

[0014] The orthogonal pulse train signal designed by the radar orthogonal waveform design module, after being transmitted by the waveform generation module and the antenna transmission module, can be scattered on the object surface and received by the receiving antenna module and the waveform receiving module to form a radar echo pulse train dataset. This dataset contains sufficient target information, and the target characteristics are strongly different from clutter and multipath interference characteristics, so as to provide the convolutional neural network module with the necessary information. By training the convolutional neural network module, it can learn and extract key target features from the dataset, while suppressing clutter and multipath interference, thereby achieving high-precision target recognition.

[0015] A method for target recognition using the above system is described. This method includes the following steps:

[0016] 1) Training phase:

[0017] During the training phase, it is necessary to adjust the orthogonal pulse train designed by the radar orthogonal waveform design module to increase the transmission entropy; design the hyperparameters of the convolutional neural network architecture; efficiently train the convolutional neural network using optimization algorithms; train the convolutional neural network using a dataset constructed from radar echo pulse trains received by the receiving antenna module; when the target recognition accuracy of the convolutional neural network is low (e.g., <90%), adjust the radar orthogonal waveform design module and the convolutional neural network module, ultimately enabling the convolutional neural network to extract key target features from radar echo pulse trains contaminated by clutter and multipath interference and suppress clutter and multipath interference, achieving high-precision target recognition based on radar echoes. The parameters of the radar orthogonal waveform design module and the convolutional neural network module are then fixed.

[0018] 2) Application stage:

[0019] Based on the orthogonal pulse train designed by the radar orthogonal waveform design module during the training phase and the parameters of the convolutional neural network module, in the application phase, the transmitting antenna module transmits the orthogonal pulse train signal designed by the radar orthogonal waveform design module generated by the signal generation module, and the receiving antenna module receives the radar echo pulse train, which serves as the input of the convolutional neural network. The convolutional neural network outputs the target type.

[0020] Compared with the prior art, the present invention has the following advantages:

[0021] 1. This invention increases the information entropy of the transmitted signal, i.e., the transmission entropy, by designing the radar transmitted pulse train as an orthogonal signal set. By constructing a strong difference between the target characteristics and clutter and multipath interference characteristics, the convolutional neural network can correctly identify the target in the presence of clutter and multipath interference. This invention combines the multiple steps of pulse compression, clutter and multipath interference suppression, and feature extraction in existing radar target identification into one step, simplifying the processing steps.

[0022] 2. This invention trains a convolutional neural network to automatically learn target features from datasets with strong differences and suppresses clutter and multipath interference. Compared with traditional clutter and multipath interference suppression methods, this method is simpler, more efficient and more versatile.

[0023] 3) The radar orthogonal waveform design module significantly increases the information entropy (i.e., transmission entropy) of the radar transmitted signal by designing orthogonal signal sets. This design method enhances the uniqueness and distinguishability of the signal, enabling the radar system to more accurately distinguish different targets, thereby improving the accuracy of target identification. The convolutional neural network module, through deep learning and feature extraction techniques, can extract key features of the target from the radar echo pulse train and effectively suppress clutter and multipath interference. This capability allows the system to accurately identify targets in complex environments, further improving the accuracy of identification.

[0024] 4) Convolutional neural networks (CNNs) have low computational and spatial complexity, making them suitable for large-scale data processing and real-time applications. In radar target recognition systems, this characteristic enables the system to quickly process large amounts of radar echo data, achieving real-time target detection and recognition. By optimizing the radar orthogonal waveform design and convolutional neural network architecture, the system can achieve rapid response and real-time application while maintaining recognition accuracy. This is of great significance for fields such as military, aviation, and transportation, as it can provide decision-makers with timely and accurate target information. Attached Figure Description

[0025] Figure 1 This is a schematic diagram illustrating an embodiment of the high-precision radar target recognition system based on radar orthogonal waveform design and convolutional neural networks according to the present invention.

[0026] Figure 2 This is a schematic diagram of an embodiment of radar orthogonal pulse train, where a shows the Welch Costas sequence encoding, and b shows the time-frequency diagram of 10 orthogonal signals combined with the Costas frequency hopping signal and the wideband linear frequency modulation (LFM) signal;

[0027] Figure 3 A schematic diagram of an embodiment of a convolutional neural network architecture.

[0028] Figure 4This is a target feature distribution map under different radar transmission waveforms in one embodiment. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0030] This embodiment provides a radar target recognition system. By designing the orthogonal pulse train waveform emitted by the radar and combining it with the powerful feature extraction capability of the convolutional neural network, the system achieves high-precision target recognition in clutter and multipath interference environments. Figure 1 A block diagram illustrating an embodiment of the system of the present invention is shown, mainly including the following: a radar orthogonal waveform design module 1, a waveform generation module 2, a transmitting antenna module 3, a receiving antenna module 4, a waveform receiving module 5, and a convolutional neural network module 6. The output of the waveform generation module 2 is connected to the input of the transmitting antenna module 3, the output of the receiving antenna module 4 is connected to the input of the waveform receiving module 5, and the output of the waveform receiving module 5 is connected to the input of the convolutional neural network module 6. The radar waveform design module 1 is responsible for designing orthogonal radar transmit pulse train waveforms, ensuring that the waveform information entropy is sufficiently large to enhance the characteristic differences between the target and clutter / multipath interference. The waveform generation module 2 is an arbitrary waveform generator, capable of generating corresponding orthogonal waveform signals based on the output of the radar orthogonal waveform design module. The transmitting antenna module 3 includes a multi-band antenna for transmitting the electromagnetic signals generated by the waveform generation module 2. The receiving antenna array module 4 is also equipped with a multi-band antenna for receiving radar echo pulse trains, ensuring stable signal reception over a wide spectrum. The waveform receiving module 5 is responsible for receiving the radar echo captured by the receiving antenna module 4 and converting it into a digital signal. This waveform receiving module can be an electrical system or an optoelectronic hybrid system. The aforementioned convolutional neural network 6 is implemented on a computer and is used to extract key target features from the radar echo pulse train received by the receiving module, and to suppress clutter and multipath interference, thereby achieving high-precision target identification.

[0031] The high-precision radar target recognition method based on radar orthogonal waveform design and convolutional neural network described above is divided into two stages: training stage and application stage, which are described in detail below.

[0032] Training phase:

[0033] First waveform design: The orthogonal pulse train waveform is designed using the radar orthogonal waveform design module 1, so that the radar echo pulse train received by the receiving antenna contains sufficient target information and exhibits a strong difference between the target and clutter and multipath interference characteristics. This enables the convolutional neural network module to extract key target features from the radar echo pulse train received by the antenna while suppressing clutter and multipath interference. Figure 2 This is a schematic diagram of an embodiment of a radar quadrature pulse train. Figure 2 a demonstrates the Welch Costas sequence encoding. Figure 2 b shows the time-frequency plot of 10 orthogonal signals combining the Costas frequency hopping signal with a wideband linear frequency modulation (LFM) signal, with the LFM slopes of the 10 orthogonal signals alternating between positive and negative.

[0034] The generation, transmission, and reception of the second orthogonal pulse train signal: The waveform generation module 2 generates the pulse train signal, the transmitting antenna module 3 transmits the pulse train signal, and the receiving antenna module 4 and the waveform receiving module 5 receive the radar echo pulse train.

[0035] The third dataset was constructed by using radar echo pulse trains to train the convolutional neural network module 6.

[0036] Fourth, optimize the convolutional neural network architecture, hyperparameters, and training methods to verify whether the network can achieve high-precision target recognition. For example... Figure 3 As shown, this embodiment includes an input layer, four convolutional layers, four max pooling layers, two fully connected layers, and an output layer.

[0037] Fifth, when the optimized convolutional neural network still cannot achieve the ideal output, the radar orthogonal waveform design module 1 is optimized so that after repeating the above steps one to four, the convolutional neural network module 6 can achieve high-precision target recognition. Figure 4 This is a target feature distribution map under different radar transmission waveforms in one embodiment. Figure 4 As can be seen, convolutional neural networks can learn the key features of targets more arbitrarily from datasets constructed from orthogonal radar echo pulse trains. The features of the two types of targets have obvious linear separability and are easy to be correctly identified.

[0038] Application phase:

[0039] The orthogonal pulse train obtained by the radar orthogonal waveform design module 1 during the training phase, along with the parameters of the convolutional neural network module 6, are used. The waveform generation module 2 and the transmitting antenna module 3 generate the designed radar pulse train and transmit it. The receiving antenna module 4 receives the radar echo pulse train. The waveform receiving module 6 converts the radar echo pulse train into data or signals adapted to the input of the convolutional neural network module 6. The convolutional neural network module 6 extracts key target features from the radar echo pulse train and suppresses clutter and multipath interference, achieving high-precision target type identification.

[0040] This invention utilizes radar orthogonal waveform design and convolutional neural networks to process radar echoes in the electromagnetic domain, efficiently extracting target features. It combines the multi-step processes of pulse compression, clutter suppression, and feature extraction required in traditional radar target identification into a single step, simplifying the processing flow and even reducing the complexity of the receiver hardware system. This invention can be widely applied in the field of target identification.

Claims

1. A high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network, characterized in that, The application relates to a radar orthogonal waveform design module (1) for designing an orthogonal signal set to increase the information entropy of a radar transmitting signal; a waveform generation module (2) connected to the radar orthogonal waveform design module for generating an orthogonal pulse train designed by the radar orthogonal waveform design module; a transmitting antenna module (3) connected to the waveform generation module for transmitting the orthogonal pulse train as an electromagnetic signal; a receiving antenna module (4) for receiving a radar echo pulse train; a waveform receiving module (5) connected to the receiving antenna module for converting the received radar echo pulse train into a digital signal; and a convolutional neural network module (6) connected to the waveform receiving module for identifying a target type based on the converted radar echo pulse train digital signal. The radar orthogonal waveform design module (1) designed orthogonal pulse train signals can be scattered on the surface of an object after being transmitted by the waveform generation module (2) and the antenna transmitting module (3), and the scattered signals can be received by the receiving antenna module (4) and the waveform receiving module (5) and form a data set of a radar echo pulse train, the data set contains sufficient target information, and the target characteristics are strongly different from the clutter and multipath interference characteristics, so as to supply the convolutional neural network module (6); by training the convolutional neural network module (6), the convolutional neural network module (6) can learn and extract target key features from the data set while suppressing clutter and multipath interference, so as to realize high-precision target identification. The training of the convolutional neural network module (6) comprises the following steps: Adjusting the radar orthogonal waveform design module to design an orthogonal radar pulse train to increase the transmitting entropy; Designing and optimizing the convolutional neural network architecture and hyperparameters; Training the convolutional neural network by using a data set constructed by a radar echo pulse train until the identification precision requirement is met. The waveform generation module (2) can generate any orthogonal waveform train in a wide spectrum range. The transmitting antenna module (3) can realize signal transmission in a wide spectrum range.

2. The high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network according to claim 1, characterized in that, The receiving antenna module (4) can realize signal reception in a wide spectrum range. The waveform receiving module (5) is a device capable of converting the orthogonal radar echo pulse train received by the antenna into a digital signal. ​ ​ 3. The high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network according to claim 2, characterized in that, ​ 4. The high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network according to claim 2, characterized in that, ​ 5. The high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network according to claim 2, characterized in that, ​ 6. The high-precision radar target recognition system based on orthogonal waveform design and convolutional neural network according to claim 2, characterized in that, ​

Images