CN-122017783-A - Weak cyclostationary signal detection method based on unmanned aerial vehicle passive radar network

CN122017783ACN 122017783 ACN122017783 ACN 122017783ACN-122017783-A

Abstract

The invention discloses a weak cyclostationary signal detection method based on an unmanned aerial vehicle-mounted passive radar network, which comprises the steps of firstly establishing a receiving signal model of an active radar transmitting signal and an unmanned aerial vehicle passive radar, then establishing a binary hypothesis model based on cyclostationary generalized likelihood ratio detection by utilizing block cyclic matrix approximation, constructing generalized likelihood ratio detection statistics under a colored noise background by utilizing cyclostationary characteristics of target echoes, obtaining local detection results of each unmanned aerial vehicle, finally fusing the local detection results of each unmanned aerial vehicle based on Bayesian minimum risk criteria in a fusion center, outputting global detection judgment, and forming a detection flow of cyclostationary generalized likelihood ratio detection-Bayesian minimum risk fusion. The method of the invention utilizes the cyclostationary characteristic of the target echo to construct the detection statistic, can improve the overall detection accuracy and robustness under the conditions of low signal-to-noise ratio and complex colored noise, and is suitable for the cooperative passive radar detection scene of multiple unmanned planes.

Inventors

ZHANG LONG
XIONG KUI
YU XIANXIANG
CUI GUOLONG
KONG LINGJIANG
YANG XIAOBO

Assignees

电子科技大学

Dates

Publication Date: 20260512
Application Date: 20260227

Claims (5)

1. A weak cyclostationary signal detection method based on an unmanned aerial vehicle passive radar network comprises the following specific steps: s1, establishing an active radar transmitting signal and a receiving signal model of a passive radar of an unmanned aerial vehicle; S2, based on the step S1, a binary hypothesis model based on cyclostationary generalized likelihood ratio detection is established by utilizing block cyclic matrix approximation; s3, establishing cyclostationary generalized likelihood ratio detection statistics based on the binary hypothesis model established in the step S2; And S4, based on the step S3, establishing a Bayesian minimum risk fusion model, and fusing all detection results of the unmanned aerial vehicle to obtain a final system detection result.
2. The method for detecting weak cyclostationary signal based on unmanned airborne passive radar network according to claim 1, wherein the step S1 is specifically as follows: First, an active radar emission signal is set The expression is as follows: (1); Wherein, the A time series is represented by a sequence of time, Indicating the frequency modulation slope, A sequence of pulses is represented and, The number of pulses is indicated and, Representing the carrier frequency of the signal, The pulse width is indicated as such, Indicating the pulse repetition period of the pulse, The amplitude is represented by a value representing the amplitude, The specific expression of the rectangular function is as follows: (2); Then the first Target echo signal received by unmanned aerial vehicle The expression is as follows: (3); Wherein, the , Representing the total number of unmanned aerial vehicles, Indicating arrival of the received signal at the drone Is added to the amplitude decay of (a), Representing the time delay of the corresponding unmanned aerial vehicle.
3. The method for detecting weak cyclostationary signal based on unmanned airborne passive radar network according to claim 2, wherein the step S2 is specifically as follows: each unmanned aerial vehicle executes multiple detection operations in each detection period, and each detection operation corresponds to a duration time of Wherein The cycle period is represented by the number of cycles, Indicating the repetition number of the period, and counting the total number of the detection windows as , After the discretization, the product is obtained, Represent the first The personal unmanned plane is on the first Echo sampling sequences within the respective detection windows; Then the first The personal unmanned plane is on the first Binary hypothesis model in each detection window The expression is as follows: (4); Wherein, the A discrete time series is represented and is shown, Represent the first The personal unmanned plane is on the first The noise in the individual detection windows is detected, Represent the first The personal unmanned plane is on the first Multipath interference within the individual detection windows; Then define a signal vector The expression is as follows: (5); Wherein, the Represents a set of complex numbers, Representing matrix transpose operations, defining received signals by the same Noise and Interference signal The expression of the respective signal vector is as follows: (6); the binary hypothesis model is updated to the following expression: (7); At the position of In the period of The observed signal expression at the individual drones is as follows: (8); Wherein, the ; Setting covariance matrix Is defined as , Then, using the cyclostationary property, the expression is obtained as follows: (9); Wherein, the It is indicated that the desire is to be met, Represents the conjugate transpose of the object, Expressed under the assumption that As a result of the desire, The time delay is indicated as such, And (3) expressing the cycle period, deriving the expression as follows: (10); The binary hypothesis is updated to the following expression: (11); Wherein, the The gaussian distribution is represented by the formula, Representing a 0 matrix; transforming the signal form to meet the two-dimensional discrete Fourier transform processing requirement, and obtaining a transformed signal expression as follows: (12); then approximated by a block circulant matrix, i.e The concrete expression of the block circulation matrix is as follows: (13); reintroducing two transformation matrices And The covariance matrix is transformed by the two matrices, and the concrete expressions are as follows: (14); Wherein, the , Matrix of (a) After diagonalization, a diagonal matrix is obtained , The specific process expression is as follows: (15); Wherein, the Representing Cronecker product, diagonal matrix The dimension is Which is divided into Diagonal blocks Each sub-block corresponds to a space-time block From the slave The expression is as follows: (16); Wherein, the ; Then the inverse Fourier transform is carried out, and the change is carried out to obtain The expression is as follows: (17); Then reassembles all spatial covariance sub-blocks, reconstructing the block diagonal covariance matrix The structure corresponds to the original structure, and the expression is as follows: (18); Wherein, the Representing the block diagonal matrix obtained by splicing the matrix according to the block diagonal form, and obtaining the matrix by the same ; Observation vector of final detection system The expression is as follows: (19); Wherein, the The binary hypothesis model expression based on cyclostationary generalized likelihood ratio detection is as follows: (20)。
4. The method for detecting weak cyclostationary signal based on unmanned airborne passive radar network according to claim 3, wherein said step S3 is specifically as follows: First, the observation vector is calculated Divided into Segments, i.e. Maximum likelihood estimation The expression is as follows: (21); Setting cyclostationary-generalized likelihood ratio detection statistics The definition expression is as follows: (22); Wherein, the And expressing likelihood functions, according to a statistical signal processing theory, performing matrix operation, wherein a covariance matrix estimation expression is as follows: (23); (24); Wherein, the Representing the first of the matrix The number of blocks of the block is one, Representing identity matrix, detection statistics The simplified expression is as follows: (25); Wherein, the The decision expression is as follows: (26); Wherein, the Representation unmanned aerial vehicle The decision threshold of (1), the probability of detection The expression is as follows: (27); Wherein, the The probability calculation is represented by a graph of the probability, Representing the local signal-to-noise ratio, Indicating the number of samples to be tested, Representing the variance of the noise and, Representing a standard Q function, and false alarm probability The expression is as follows: (28)。
5. The method for detecting weak cyclostationary signal based on unmanned airborne passive radar network according to claim 4, wherein the step S4 is specifically as follows: at the fusion center, each unmanned aerial vehicle transmits a local detection result to the fusion center, and equivalent false alarm probability and equivalent detection probability analysis are adopted Equivalent detection probability of (a) And equivalent false alarm probability The expressions are as follows: (29); Wherein, the The method comprises the steps of representing the signal-to-noise ratio of a transmission channel, fusing the local detection results of all unmanned aerial vehicles at a fusion center by adopting a Bayesian minimum risk method to obtain a final detection result The expression is as follows: (30); Wherein, the Representing assumptions Is used to determine the prior probability of (c) for a given channel, The local detection result of the unmanned aerial vehicle is represented, Representing assumptions Is used to determine the joint probability of (1), Representing edge probability, then the decision criterion expression at the fusion center is as follows: (31); Wherein, the Representing assumptions Is used to determine the prior probability of (c) for a given channel, Representing assumptions And then the equivalent detection probability And equivalent false alarm probability Bringing into formula (31), the resulting expression is as follows: (32); Wherein, the Is expressed as The target is considered to be present by the individual unmanned aerial vehicle, Representing the decision threshold, if The target is considered to exist, otherwise, the target is not exist The expression is as follows: (33); Wherein, the The integral variable is represented by a value of the integral variable, Representing the non-central parameter of the device, Representing the bezier function of the model, The degree of freedom of chi-square distribution is represented; Finally, based on local detection and global fusion, signal detection is realized, and target detection probability is improved.

Description

Weak cyclostationary signal detection method based on unmanned aerial vehicle passive radar network Technical Field The invention belongs to the technical field of radar target detection, and particularly relates to a weak cyclostationary signal detection method based on an unmanned aerial vehicle passive radar network. Background In recent years, the low-altitude slow small target has increasingly increased activity under urban groups, coastlines and complex electromagnetic environments, the target has small scattering cross section and strong maneuverability, and is easy to be overlapped with ground clutter and interference signals, so that the detection performance of the existing methods such as energy detection, matched filtering and the like is obviously reduced under the condition of low signal-to-noise ratio. Meanwhile, the passive radar realizes 'passive hidden detection' by using external radiation sources such as broadcast televisions, communication base stations and the like, has the advantages of anti-radiation striking, flexible spectrum utilization and the like, is suitable for rapid deployment and large-scale monitoring of an unmanned aerial vehicle platform, but echoes of the passive radar are generally influenced by a radiation source modulation structure, propagation multipath, platform motion and nonstationary clutter, and have obvious correlation and non-Gaussian characteristics, so that a classical self-adaptive detector based on independent homodisperse Gaussian noise assumption is difficult to obtain stable gain. Cyclostationary signal detection can utilize the inherent cycle statistics characteristics of the irradiation source and echo, and realize noise suppression gain under the condition that noise and clutter are insensitive to the cycle frequency, and is considered as an important approach for passive detection of a weak target. However, in an unmanned plane scene, due to factors such as limited observation window, insufficient sampling data volume, unreliable time-frequency synchronization error and link reporting of each platform and the like, the estimation deviation of cyclic statistics is increased, and the balance between communication overhead and detection performance is needed when multiple unmanned planes cooperate, the existing method often has the problems of strong prior dependence, insufficient robustness to non-ideal factors, complex engineering implementation and the like. Therefore, it is necessary to provide a weak cyclostationary signal detection method suitable for an unmanned airborne passive radar network, and stable and reliable target detection is realized under the conditions of low signal-to-noise ratio, strong clutter interference and non-ideal cooperation. Disclosure of Invention In order to solve the technical problems, the invention provides a weak cyclostationary signal detection method based on an unmanned aerial vehicle passive radar network, which is used for coping with reconnaissance challenges in a low signal-to-noise ratio environment and improving the reliability and accuracy of a reconnaissance system. The technical scheme adopted by the invention is that the weak cyclostationary signal detection method based on the unmanned aerial vehicle passive radar network comprises the following specific steps: s1, establishing an active radar transmitting signal and a receiving signal model of a passive radar of an unmanned aerial vehicle; S2, based on the step S1, a binary hypothesis model based on cyclostationary generalized likelihood ratio detection is established by utilizing block cyclic matrix approximation; s3, establishing cyclostationary generalized likelihood ratio detection statistics based on the binary hypothesis model established in the step S2; And S4, based on the step S3, establishing a Bayesian minimum risk fusion model, and fusing all detection results of the unmanned aerial vehicle to obtain a final system detection result. Further, the step S1 specifically includes the following steps: First, an active radar emission signal is set The expression is as follows: (1); Wherein, the A time series is represented by a sequence of time,Indicating the frequency modulation slope,A sequence of pulses is represented and,The number of pulses is indicated and,Representing the carrier frequency of the signal,The pulse width is indicated as such,Indicating the pulse repetition period of the pulse,The amplitude is represented by a value representing the amplitude,The specific expression of the rectangular function is as follows: (2); Then the first Target echo signal received by unmanned aerial vehicleThe expression is as follows: (3); Wherein, the ,Representing the total number of unmanned aerial vehicles,Indicating arrival of the received signal at the droneIs added to the amplitude decay of (a),Representing the time delay of the corresponding unmanned aerial vehicle. Further, the step S2 specifically includes the following steps: and in e