CN-122028171-A - Unmanned aerial vehicle multimode fusion radio positioning and signal recognition method and system

CN122028171ACN 122028171 ACN122028171 ACN 122028171ACN-122028171-A

Abstract

The application discloses a multimode fusion radio positioning and signal identifying method and system of an unmanned aerial vehicle, which relate to the technical field of radio positioning and detection, wherein a power spectrum sequence and unmanned aerial vehicle communication signals are obtained by continuously monitoring a airspace radio environment in a monitoring area through each receiving station, and spectrum characteristics are extracted from the power spectrum sequence; the method comprises the steps of acquiring three-dimensional motion tracks of an unmanned aerial vehicle according to arrival time of receiving stations of the same transmitting event, acquiring echo signals generated by unmanned aerial vehicle rotors and body motions through a radar, extracting micro Doppler features from the echo signals, acquiring protocol related modulation features, cyclic spectrum features and protocol features according to unmanned aerial vehicle communication signals, carrying out multi-mode feature fusion on the spectrum features, the micro Doppler features, the modulation features, the cyclic spectrum features and the protocol features to acquire depth joint features, identifying unmanned aerial vehicle categories according to the depth joint features to acquire category labels, and constructing unmanned aerial vehicle feature signatures by the category labels, the three-dimensional motion tracks and the depth joint features. The application can improve the accuracy, robustness and supervision capability of unmanned aerial vehicle positioning and signal detection in complex environments.

Inventors

WANG GUANJUN
NIU NING
XIA YINGJUN
LIU MALIANG
YANG QIAN

Assignees

海南大学

Dates

Publication Date: 20260512
Application Date: 20260128

Claims (10)

1. A method of multi-mode fusion radio positioning and signal recognition for an unmanned aerial vehicle, characterized in that it is applied to a scene comprising an unmanned aerial vehicle, a radar and a plurality of receiving stations, said method comprising: Continuously monitoring a space domain radio environment in a monitoring area through each receiving station, acquiring a power spectrum sequence and an unmanned aerial vehicle communication signal, and extracting spectrum characteristics from the power spectrum sequence; For the same transmitting event, acquiring the arrival time of each receiving station with synchronous time; Positioning the space position of the unmanned aerial vehicle according to the arrival time of each receiving station, and acquiring the three-dimensional motion trail of the unmanned aerial vehicle in the monitoring area; Acquiring echo signals generated by the unmanned aerial vehicle rotor wing and the body motion through a radar, and extracting micro Doppler features according to the echo signals; Acquiring a baseband signal according to the unmanned aerial vehicle communication signal, and performing maximum likelihood-based modulation recognition on the baseband signal to acquire modulation characteristics; extracting a cyclic frequency from the unmanned aerial vehicle communication signal, and extracting a cyclic spectrum characteristic according to the cyclic frequency; carrying out multi-mode feature fusion on the spectrum feature, the micro Doppler feature, the modulation feature, the cyclic spectrum feature and the protocol feature to obtain a depth joint feature; identifying the category of the unmanned aerial vehicle according to the depth joint characteristics, and obtaining a category label; And constructing unmanned aerial vehicle characteristic signatures by the category labels, the three-dimensional motion trail and the depth joint characteristics.
2. The method of claim 1, wherein the extracting spectral features from the sequence of power spectra comprises: Performing energy detection and threshold judgment on the power spectrum sequence to obtain the distribution condition of the unmanned aerial vehicle communication signals on a time-frequency plane so as to obtain a frequency hopping track; and in each frequency hopping period, counting at least one parameter including frequency hopping period, frequency hopping bandwidth, channel duty ratio, power average value and power variance, and constructing the frequency characteristic.
3. The method of claim 1, wherein locating the spatial position of the drone based on the arrival time of each receiving station, and wherein obtaining the three-dimensional motion trajectory of the drone within the monitored area comprises: Selecting reference stations from the receiving stations, and respectively acquiring the arrival time difference between each receiving station and the reference station; Positioning the space position of the unmanned aerial vehicle based on a TDOA/AOA fusion positioning algorithm according to the arrival time difference, the space coordinates of each receiving station and the electromagnetic wave propagation speed; and smoothing and predicting the space position of the unmanned aerial vehicle at continuous moments by using Kalman filtering to obtain the three-dimensional motion trail.
4. The method of claim 1, wherein said extracting micro-doppler features from said echo signals comprises: Performing short-time Fourier transform on the echo signals to obtain time-frequency distribution, and calculating micro Doppler energy spectrum; and extracting at least one parameter including rotor fundamental frequency, frequency multiplication component, energy peak position, frequency bandwidth and energy time-varying track from the micro Doppler energy spectrum, and constructing the micro Doppler characteristic.
5. The method of claim 1, wherein said extracting cyclic spectral features from said cyclic frequencies comprises: And calculating cyclic autocorrelation and cyclic spectral density under a plurality of cyclic frequencies, selecting the spectral peak position, spectral energy and noise contrast at the non-zero cyclic frequency, and constructing the cyclic spectral features.
6. The method of claim 1, wherein extracting protocol features from multi-frame data of the drone communication signal comprises: acquiring an autocorrelation function of the baseband signal, and determining a frame boundary; And constructing the protocol feature based on at least one statistical feature parameter of the frame boundary, which is included in the multi-frame data in a statistical address field, a frame length field, a load field, an encryption field and an extension field.
7. The method of claim 1, wherein the multi-modal feature fusion of the spectral features, the micro-doppler features, the modulation features, the cyclic spectral features, and the protocol features to obtain multi-modal deep joint features comprises: And constructing a dynamic characteristic weight distribution sub-network according to real-time scene conditions of the monitoring area, and carrying out weighted characteristic fusion based on the characteristic weight distribution sub-network to obtain the depth joint characteristics, wherein the scene conditions comprise electromagnetic environment noise level and interference type.
8. The method of claim 1, wherein the deep joint feature identifies a class of unmanned aerial vehicle, and obtaining a class tag comprises: The method comprises the steps of inputting depth joint features into a first stage classifier, wherein the first stage classifier adopts a decision tree structure, takes information entropy and information gain as node division criteria, obtains target features for eliminating interference, inputting the target features into a second stage classification model, and the second stage classification model comprises one structure or a combination of the two structures in a convolutional neural network structure and a Transformer structure and is used for identifying unmanned aerial vehicle and non-unmanned aerial vehicle types, judging the unmanned aerial vehicle types and outputting class labels.
9. The method of claim 8, wherein the method further comprises: And performing countermeasure training on the second-stage classification model by adopting a generated countermeasure network.
10. A system for multi-mode fusion radio positioning and signal recognition of an unmanned aerial vehicle, applied to a scene comprising an unmanned aerial vehicle, a radar and a plurality of receiving stations, said system comprising: the frequency spectrum monitoring module is used for continuously monitoring the space domain radio environment in a monitoring area through each receiving station, acquiring a power spectrum sequence and unmanned aerial vehicle communication signals, and extracting frequency spectrum characteristics from the power spectrum sequence; The time synchronization module is used for acquiring the arrival time of each receiving station in time synchronization for the same transmitting event; the radio positioning and tracking module is used for positioning the space position of the unmanned aerial vehicle according to the arrival time of each receiving station and acquiring the three-dimensional motion trail of the unmanned aerial vehicle in the monitoring area; The radar signal processing module is used for acquiring echo signals generated by the unmanned aerial vehicle rotor wing and the body motion through a radar, and extracting micro Doppler features according to the echo signals; The protocol analysis module is used for acquiring a baseband signal according to the unmanned aerial vehicle communication signal, carrying out maximum likelihood-based modulation identification on the baseband signal, and acquiring modulation characteristics; The feature fusion module is used for carrying out feature fusion on the frequency spectrum feature, the micro Doppler feature, the modulation feature, the cyclic spectrum feature and the protocol feature to obtain a depth joint feature; the classification and identification module is used for identifying the category of the unmanned aerial vehicle according to the depth joint characteristics and obtaining a category label; and the database module is used for constructing unmanned aerial vehicle characteristic signatures from the category labels, the three-dimensional motion trail and the depth joint characteristics.

Description

Unmanned aerial vehicle multimode fusion radio positioning and signal recognition method and system Technical Field The application relates to the technical field of radio positioning and detection, in particular to a multimode fusion radio positioning and signal identification method and system for an unmanned aerial vehicle. Background In recent years, unmanned aerial vehicles are widely applied in various fields such as mapping, electric power inspection and the like, the civil number is exponentially increased, but the problems of black flight, disturbance, and the like form serious threats to airspace safety and national safety, and efficient and accurate individual identification of unmanned aerial vehicles becomes a core bottleneck of airspace management. In the existing identification method, visual and audio identification is not enough in stability due to environmental restriction, radar cost is high and is easy to interfere, and radio frequency spectrum sensing is mainstream due to low cost and high robustness. However, spectrum sensing is dependent on STFT and FFT to extract surface features, deep information is difficult to mine, recognition rate is suddenly reduced in a multipath interference, strong noise or signal camouflage scene, a single mode is difficult to meet requirements, and a multi-source information fusion technology is needed. The existing multimode fusion scheme is limited to double-mode combination, and has the defects that information conflict is caused by feature level splicing, unique electronic identity cannot be generated due to lack of protocol deep analysis, and the like, and the recognition performance is deteriorated when the scheme faces attack due to insufficient recognition precision of a camouflage interference scene. Disclosure of Invention Based on the above, it is necessary to provide a method and a system for multi-mode fusion radio positioning and signal recognition of an unmanned aerial vehicle, which can improve the robustness of complex scenes. In a first aspect, the present application provides a method for multimode fusion of radio positioning and signal recognition of an unmanned aerial vehicle, applied to a scene including the unmanned aerial vehicle, a radar and a plurality of receiving stations, the method comprising: Continuously monitoring the airspace radio environment in a monitoring area through each receiving station, acquiring a power spectrum sequence and unmanned aerial vehicle communication signals, and extracting spectrum features from the power spectrum sequence; For the same transmitting event, acquiring the arrival time of each receiving station with synchronous time; Positioning the space position of the unmanned aerial vehicle according to the arrival time of each receiving station, and acquiring the three-dimensional motion trail of the unmanned aerial vehicle in the monitoring area; Acquiring echo signals generated by the unmanned aerial vehicle rotor wing and the body motion through a radar, and extracting micro Doppler features according to the echo signals; acquiring a baseband signal according to the unmanned aerial vehicle communication signal, and carrying out maximum likelihood-based modulation recognition on the baseband signal to acquire modulation characteristics; carrying out multi-mode feature fusion on the spectrum features, the micro Doppler features, the modulation features, the cyclic spectrum features and the protocol features to obtain depth joint features; Identifying the class of the unmanned aerial vehicle according to the depth joint characteristics, and obtaining class labels; and constructing unmanned aerial vehicle characteristic signatures by combining the category labels, the three-dimensional motion tracks and the depth. In one embodiment, extracting spectral features from the power spectrum sequence includes: Performing energy detection and threshold judgment on the power spectrum sequence to obtain the distribution condition of unmanned aerial vehicle communication signals on a time-frequency plane so as to obtain a frequency hopping track; and in each frequency hopping period, counting at least one parameter including frequency hopping period, frequency hopping bandwidth, channel duty ratio, power average value and power variance, and constructing frequency characteristics. In one embodiment, locating the spatial position of the unmanned aerial vehicle according to the arrival time of each receiving station, and acquiring the three-dimensional motion trail of the unmanned aerial vehicle in the monitoring area includes: Selecting reference stations from the receiving stations, and respectively acquiring the arrival time difference between each receiving station and the reference station; Positioning the space position of the unmanned aerial vehicle based on a TDOA/AOA fusion positioning algorithm according to the arrival time difference, the space coordinates of each receiving station and the electromagn