CN-119493111-B - Double-domain combined narrow-band radar moving target classification and identification method and related products

CN119493111BCN 119493111 BCN119493111 BCN 119493111BCN-119493111-B

Abstract

The invention discloses a double-domain combined narrow-band radar moving target classification recognition method and related products, which are used for processing target echo signals received by a radar to obtain a distance Doppler spectrum, obtaining a micro Doppler time spectrum based on the target echo signals received by the radar, respectively cutting the distance Doppler spectrum and the micro Doppler time spectrum to obtain distance Doppler domain characteristic data and micro Doppler domain characteristic data, and inputting the distance Doppler domain characteristic data and the micro Doppler domain characteristic data into a pre-constructed RFICNet to obtain a target classification result. According to the method, the characteristics of the double-domain data of the distance Doppler domain and the micro Doppler domain are combined, complex echo data preprocessing operation and excessive priori physical assumption are not relied on, the deficiency of characteristic basis in classification can be made up by the double-domain combined characteristic information, and the classification recognition precision of the narrow-band radar moving target is improved.

Inventors

LIU YIZHI
YUAN WEIMING
SHEN KAILUN

Assignees

西安电子科技大学

Dates

Publication Date: 20260512
Application Date: 20241105

Claims (7)

1. The double-domain combined narrow-band radar moving target classification and identification method is characterized by comprising the following steps of: processing target echo signals received by a radar to obtain a range-Doppler spectrum; Acquiring a micro Doppler time spectrum based on a target echo signal received by a radar; Cutting the range-Doppler spectrum and the micro-Doppler time spectrum respectively to obtain range-Doppler domain characteristic data and micro-Doppler domain characteristic data; Inputting the range-Doppler domain characteristic data and the micro-Doppler domain characteristic data into RFICNet constructed in advance to obtain a classification and identification result; The pre-constructed RFICNet comprises an input module, two feature extraction modules, two spatial attention modules and a classification recognition module, wherein the input module is connected with the feature extraction modules, each feature extraction module is connected with one spatial attention module, the spatial attention modules are connected with the classification recognition module, the input module is used for normalizing input data, the feature extraction modules adopt maximum pooling, the spatial attention modules comprise maximum pooling and average pooling, and the classification recognition module adopts three full connection layers; The target echo signal received based on the radar obtains a micro Doppler time spectrum, which specifically comprises the following steps: mixing the target echo signals received by the radar to obtain baseband signals; processing the baseband signal to obtain a distance pulse spectrum; The micro Doppler time spectrum is obtained based on the distance pulse spectrum, and the micro Doppler time spectrum is obtained based on the distance pulse spectrum specifically as follows: Selecting slow time dimension data corresponding to a distance unit of a target in a distance pulse spectrum, and performing time-frequency transformation on the selected slow time dimension data to obtain a micro Doppler time spectrum, wherein the distance unit of the target is a distance unit of the target obtained by constant false alarm rate detection and trace condensation processing of the distance Doppler spectrum; the method comprises the steps of cutting a range-Doppler spectrum and a micro-Doppler time spectrum respectively to obtain range-Doppler domain characteristic data and micro-Doppler domain characteristic data, wherein the specific steps are as follows: Performing constant false alarm rate detection on the range-Doppler spectrum, and then performing trace point condensation to obtain a range unit and a Doppler unit of a target; Cutting the range-Doppler spectrum by taking a range unit and a Doppler unit corresponding to the condensed target in the range-Doppler spectrum as centers to obtain range-Doppler domain characteristic data; Cutting the micro Doppler time spectrum by taking frequency data with the largest amplitude in the micro Doppler time spectrum as a center to obtain a cut micro Doppler time spectrum; And stretching and downsampling the cut micro Doppler time spectrum to obtain micro Doppler domain feature data, wherein the micro Doppler domain feature data is consistent with the range Doppler domain feature data in dimension.
2. The method for classifying and identifying the moving target of the narrow-band radar combined by the double domains according to claim 1, wherein the method is characterized in that the target echo signals received by the radar are processed to obtain a range-doppler spectrum, and specifically comprises the following steps: mixing the target echo signals received by the radar to obtain baseband signals; and processing the baseband signal to obtain a range-Doppler spectrum.
3. The method for classifying and identifying the moving target of the narrow-band radar by combining the two domains according to claim 2, wherein the mixing processing is performed on the target echo signal received by the radar to obtain a baseband signal, specifically: For a linear frequency modulation pulse radar, performing digital down-conversion processing on a target echo signal received by the radar to obtain a baseband signal; and for the linear frequency modulation continuous wave radar, performing declining treatment on a target echo signal received by the radar to obtain a baseband signal.
4. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the two-domain combined narrowband radar moving object classification method of any of claims 1-3 when the computer program is executed.
5. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when executed by a processor, implements the two-domain combined narrowband radar moving object classification identification method of any of claims 1-3.
6. A computer program product comprising computer instructions for instructing a computer to perform the two-domain combined narrowband radar moving object classification method of any of claims 1-3.
7. A dual-domain combined narrowband radar moving object classification and identification system, based on the dual-domain combined narrowband radar moving object classification and identification method of claim 1, characterized by comprising: The moving target detection module is used for processing target echo signals received by the radar to obtain a range-Doppler spectrum; The time-frequency analysis module is used for obtaining a micro Doppler time spectrum based on a target echo signal received by the radar; The dual-domain data preprocessing module is used for respectively cutting the range Doppler spectrum and the micro Doppler time spectrum to obtain range Doppler domain characteristic data and micro Doppler domain characteristic data; The target classification and identification module is used for inputting the range Doppler domain characteristic data and the micro Doppler domain characteristic data into RFICNet constructed in advance to obtain a target classification result; The pre-constructed RFICNet comprises an input module, two feature extraction modules, two spatial attention modules and a classification recognition module, wherein the input module is connected with the feature extraction modules, each feature extraction module is connected with one spatial attention module, the spatial attention modules are connected with the classification recognition module, the input module is used for normalizing input data, the feature extraction modules adopt maximum pooling, the spatial attention modules comprise maximum pooling and average pooling, and the classification recognition module adopts three full connection layers.

Description

Double-domain combined narrow-band radar moving target classification and identification method and related products Technical Field The invention belongs to the technical field of narrow-band radar target recognition, and relates to a double-domain combined narrow-band radar moving target classification recognition method and related products. Background The narrow-band radar is used as a detection device widely applied to a plurality of fields such as military reconnaissance, civil monitoring, air traffic management and the like, and the inherent bandwidth limitation and lower resolution are always key factors for restricting the performance improvement. Because the bandwidth of the narrow-band radar is narrow, the target information which can be acquired in the detection process is relatively limited, and the difficulty in classifying and identifying the target in the follow-up process is greatly increased. Among existing methods for classifying and identifying moving targets of narrow-band radar, most methods rely on artificial feature extraction and parameter estimation based on a physical model. These methods typically require complex preprocessing of the radar echo signal to extract characteristic parameters that reflect the characteristics of the target, such as speed, acceleration, shape information, etc. However, due to the resolution limitations of narrowband radars, these feature parameters are often not sufficiently accurate or comprehensive, thereby affecting the accuracy of classification recognition. In addition, in the existing classification and identification method, when narrowband radar data is processed, it is often difficult to comprehensively utilize characteristic information of a target in radar echo. On one hand, the traditional signal processing method possibly ignores some detail information which has important influence on classification recognition when extracting features, and on the other hand, the existing machine learning algorithm possibly faces the problems of high computational complexity, poor model generalization capability and the like when processing high-dimensional data, so that the effect of the method in practical application is limited. Disclosure of Invention The invention aims to solve the technical problem that the prior various classification and identification methods are difficult to comprehensively utilize the characteristic information of a narrow-band radar target to cause inaccurate identification of a moving target, and provides a double-domain combined narrow-band radar moving target classification and identification method and related products. In order to achieve the purpose, the invention is realized by adopting the following technical scheme: the invention provides a double-domain combined narrow-band radar moving target classification and identification method, which comprises the following steps: processing target echo signals received by a radar to obtain a range-Doppler spectrum; Acquiring a micro Doppler time spectrum based on a target echo signal received by a radar; Cutting the range-Doppler spectrum and the micro-Doppler time spectrum respectively to obtain range-Doppler domain characteristic data and micro-Doppler domain characteristic data; Inputting the range-Doppler domain characteristic data and the micro-Doppler domain characteristic data into RFICNet constructed in advance to obtain a classification and identification result; The pre-constructed RFICNet comprises an input module, two feature extraction modules, two spatial attention modules and a classification recognition module, wherein the input module is connected with the feature extraction modules, each feature extraction module is connected with one spatial attention module, the spatial attention modules are connected with the classification recognition module, the input module is used for normalizing input data, the feature extraction modules adopt maximum pooling, the spatial attention modules comprise maximum pooling and average pooling, and the classification recognition module adopts three full connection layers. Further, the processing the target echo signal received by the radar to obtain a range-doppler spectrum specifically includes: mixing the target echo signals received by the radar to obtain baseband signals; and processing the baseband signal to obtain a range-Doppler spectrum. Further, the mixing processing is performed on the target echo signal received by the radar to obtain a baseband signal, which specifically includes: For a linear frequency modulation pulse radar, performing digital down-conversion processing on a target echo signal received by the radar to obtain a baseband signal; and for the linear frequency modulation continuous wave radar, performing declining treatment on a target echo signal received by the radar to obtain a baseband signal. Further, the target echo signal received based on the radar obtains a micro-doppler time spectrum, which s