Physical-communication domain consistency three-dimensional unmanned aerial vehicle abnormal behavior detection method and system

Abstract

The invention relates to a physical-communication domain consistency three-dimensional unmanned aerial vehicle abnormal behavior detection method and system. A plurality of communication monitoring stations are deployed in a to-be-monitored airspace, the arrival angle and the arrival time of wireless signals of the unmanned aerial vehicle are collected, the three-dimensional arrival angle and time difference information are fused, communication domain position estimation is obtained, and modeling is conducted on time synchronization errors. And meanwhile, the physical domain position estimation of the unmanned aerial vehicle is obtained through physical sensing equipment. And based on the positions of the communication domain and the physical domain and uncertainty information, establishing a consistency constraint discrimination mechanism, and realizing unmanned aerial vehicle abnormal behavior detection. According to the method, communication and physical perception are fused, the detection reliability and robustness are improved through consistency constraint judgment, and the method is suitable for a low-altitude airspace supervision scene.

Description

Technical Field

[0001] This invention relates to the fields of low-altitude airspace management, UAV monitoring and wireless communication sensing technology, and in particular to a three-dimensional UAV abnormal behavior detection method and system based on physical domain and communication domain consistency constraints for low-altitude airspace management. Background Technology

[0002] With the gradual opening of low-altitude airspace and the widespread application of drones in logistics, urban patrol, and emergency rescue, low-altitude airspace management faces challenges such as a surge in the number of drones, a complex flight environment, and rising safety risks. Illegal flights, communication interference, identity spoofing, and malicious deception pose a serious threat to the safety of low-altitude airspace operations.

[0003] Current UAV anomaly detection technologies primarily rely on information from a single sensing domain for judgment. One type of method is based on physical sensing means, such as radar, optical imaging, or infrared imaging, to monitor the spatial position and trajectory of the UAV. This type of method can directly reflect the physical state of the UAV, but it is susceptible to obstruction, multipath reflection, and weather conditions in complex urban environments, and the equipment deployment and maintenance costs are relatively high. Another type of method is based on communication sensing means, which detects and identifies the UAV by listening to its wireless communication signals and extracting features such as signal strength, angle of arrival, and time difference of arrival. This type of method has advantages such as being non-contact and having a wide coverage area, but existing communication sensing technologies mostly use two-dimensional planar models, ignore the altitude information of the UAV, and lack systematic modeling of time synchronization errors between communication monitoring stations, resulting in insufficient three-dimensional positioning accuracy and reliability.

[0004] Current technologies generally use physical perception results and communication perception results independently, lacking a constraint mechanism to ensure their statistical consistency. When drones exhibit communication anomalies, identity spoofing, or deception, relying solely on information from a single perception domain makes it difficult to identify abnormal targets in a timely and accurate manner, easily leading to false alarms or missed alarms.

[0005] Therefore, there is an urgent need for a method and system that can integrate communication and physical sensing information in three-dimensional space and detect abnormal behavior of UAVs through consistency constraints, so as to improve the safety and reliability of low-altitude airspace management. Summary of the Invention

[0006] The main objective of this invention is to address the technical problems of insufficient two-dimensional modeling accuracy, lack of time synchronization error modeling, and insufficient reliability of cross-sensory domain discrimination in existing UAV anomaly detection methods. It provides a three-dimensional UAV anomaly behavior detection method and system with physical-communication domain consistency constraints for low-altitude airspace management. By explicitly modeling the uncertainty of the three-dimensional positioning results in the communication domain and performing statistical consistency analysis with the perceived trajectory in the physical domain, reliable identification of communication anomalies, identity spoofing, and deception behaviors is achieved. While ensuring detection reliability, real-time requirements are met, making this method applicable to online monitoring of UAVs in low-altitude airspace and air traffic management systems.

[0007] The present invention provides a method for detecting abnormal behavior of three-dimensional unmanned aerial vehicles based on physical-communication domain consistency, comprising the following steps: Step 1: Obtain the 3D position estimation result of the UAV in the communication domain and its first covariance. The 3D position estimation in the communication domain is based on the 3D angle of arrival and time difference of arrival information of wireless signals collected by multiple communication monitoring stations. In the positioning calculation process, the statistical characteristics of the time synchronization error between communication monitoring stations are explicitly introduced into the observation model as an error term. Step 2: Obtain the 3D position estimation result of the UAV in the physical domain and its second covariance. The 3D position estimation in the physical domain is based on the perception data of radar or optical imaging equipment. Step 3: Calculate the Mahalanobis distance of the difference vector between the 3D position estimation results in the communication domain and the physical domain using the inverse matrix of the sum of the first covariance and the second covariance as weights. This distance is used as a consistency discrimination index. The Mahalanobis distance is compared with a discrimination threshold determined based on the chi-square distribution. If the Mahalanobis distance exceeds the discrimination threshold, the perception results in the communication domain and the perception results in the physical domain are statistically inconsistent. Step 4: When the determination result in Step 3 is inconsistent, it is determined that the UAV has abnormal behavior, and an abnormal alarm message is output.

[0008] Preferably, in step 1, the statistical characteristics of the time synchronization error are explicitly introduced into the observation model as an error term. Specifically, this includes converting the time synchronization error between each communication monitoring station into an equivalent distance error, and fusing the statistical variance of the equivalent distance error with the measurement noise variance of the arrival time difference to construct the comprehensive error term in the observation model.

[0009] Furthermore, in step 1, a nonlinear observation equation is constructed based on the three-dimensional angle of arrival, time difference of arrival, and the comprehensive error term, and a weighted least squares algorithm is used for iterative solution to obtain the three-dimensional position estimation result of the communication domain and its first covariance.

[0010] Preferably, step 2 specifically includes: The raw observation data of the UAV is acquired by radar or optical imaging equipment, including slant range, azimuth angle and pitch angle. The original observation data is transformed to obtain the physical domain position observation values ​​in the three-dimensional Cartesian coordinate system; Based on the physical domain position observations and the preset sensor measurement noise model, the three-dimensional position estimation result of the physical domain and its second covariance are determined.

[0011] Preferably, the discrimination threshold determined in step 3 based on the chi-square distribution is determined by setting a pre-set confidence level, determining the degrees of freedom of the chi-square distribution according to the dimension of the three-dimensional position coordinates, and querying the corresponding chi-square distribution critical value.

[0012] Furthermore, the discrimination threshold is selected directly from the chi-square distribution critical value table based on the system's requirements for the false alarm rate of anomaly detection and the chi-square distribution degrees of freedom that the Mahalanobis distance follows.

[0013] Preferably, in step 1, multiple communication monitoring stations are deployed with known three-dimensional spatial coordinates, the wireless communication signal is a signal transmitted or reflected by the UAV, and the three-dimensional angle of arrival includes azimuth and pitch angle.

[0014] Based on the same inventive concept, the present invention also designs an electronic device, comprising: One or more processors; memory for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement traffic flow prediction methods such as adaptive spatiotemporal coding and mask pre-training enhancement.

[0015] Based on the same inventive concept, the present invention also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described adaptive spatiotemporal coding and mask pre-training enhanced traffic flow prediction method.

[0016] Based on the same inventive concept, this invention also designs a three-dimensional UAV abnormal behavior detection system based on physical-communication domain consistency constraints, used to implement a three-dimensional UAV abnormal behavior detection method based on physical-communication domain consistency constraints, comprising: a communication domain positioning module, used to execute step 1, to obtain the three-dimensional position estimation result of the communication domain and its first covariance; a physical domain perception module, used to execute step 2, to obtain the three-dimensional position estimation result of the physical domain and its second covariance; a consistency discrimination module, used to execute step 3, to perform the cross-domain consistency statistical discrimination; and an anomaly alarm module, used to execute step 4, to output the anomaly alarm information.

[0017] Therefore, the present invention has the following advantages: 1. By adopting a three-dimensional communication perception model, the altitude information of the UAV is fully considered, which improves the positioning accuracy and reliability of the communication side and makes up for the shortcomings of traditional solutions in vertical space observation capabilities; 2. The time synchronization error of the communication monitoring station is explicitly modeled, and the uncertainty generated by the inter-station synchronization is identified as the inherent and random error of the system. It is then transformed into an equivalent propagation distance component and introduced into the observation equation. Automatic compensation for the synchronization residual is achieved by allocating a weight matrix, which enhances the robustness of the system in non-ideal deployment environments. 3. By constraining the consistency between the physical domain and the communication domain, reliable detection of abnormal UAV behavior can be achieved, reducing the probability of false alarms and missed alarms; 4. The system has a clear structure and adopts a weighted iterative solution and a chi-square distribution decision criterion to transform the complex nonlinear optimization problem into an efficient linearized iterative and statistical verification process. It has low computational complexity and can fully meet the real-time requirements of low-altitude airspace management for large-scale target monitoring. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall system structure of the present invention.

[0019] Figure 2 This is a flowchart of the physical-communication domain consistency constraint discrimination of the present invention.

[0020] Figure 3 This is a schematic diagram of the three-dimensional communication domain fusion positioning principle of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0022] Example 1: As attached Figure 1 , 3 As shown, this embodiment discloses a three-dimensional UAV abnormal behavior detection method based on physical-communication domain consistency constraints for low-altitude airspace management, including the following steps: Step 1: Three-dimensional positioning in the communication domain, including: Step 1-1: Construction Steps of Low-Altitude Airspace Communication Monitoring System Within the low-altitude airspace to be monitored, multiple communication monitoring stations are deployed, each with known three-dimensional spatial coordinates. These stations receive wireless communication signals transmitted or reflected by the UAV during flight. The stations process the received signals to obtain the angle of arrival (AHA) and time of arrival (TOA) information, including the azimuth and pitch angles, which characterize the signal's propagation direction in three-dimensional space.

[0023] Steps 1-2: Three-dimensional communication direction finding and time difference extraction steps The wireless communication signals received by each communication monitoring station in step 1-1 are processed. Specifically, each communication monitoring station receives radio frequency signals through an antenna array, digitally samples the signals, and then uses a spatial spectrum estimation algorithm to perform eigenvalue decomposition and peak search on the array signal covariance matrix, thereby calculating the three-dimensional angle of arrival measurement value corresponding to each communication monitoring station. The three-dimensional angle of arrival includes the azimuth and elevation angles of the signal source. The expression for the spatial spectral function is:

[0024] in, Represents the spatial spectral function. Indicates azimuth. Indicates pitch angle, This represents the array steering vector in the corresponding direction. Represents the noise subspace matrix. This represents the conjugate transpose operation. Simultaneously, each communication monitoring station collects and timestamps the received signal data in real time based on a unified high-precision clock reference. One communication monitoring station is selected as the reference station. Generalized cross-correlation processing is performed between the timestamped signal data from other monitoring stations and the signal data from the reference station. The time delay corresponding to the peak value of the cross-correlation function is calculated, thereby determining the signal arrival time difference between other monitoring stations and the reference station, and obtaining the measured value of the UAV communication signal arrival time difference. Through the above processing, three-dimensional angle of arrival measurement data and time difference of arrival measurement data of the UAV at multiple communication monitoring stations are obtained, providing observational information for subsequent three-dimensional communication fusion positioning.

[0025] Steps 1-3: Modeling steps for time synchronization uncertainty at communication monitoring stations The time synchronization error existing between communication monitoring stations is identified as both inherent system error and random jitter error. First, the time value of the time synchronization error is multiplied by the propagation speed of the wireless signal in the medium to calculate the corresponding equivalent propagation distance error. Next, this equivalent propagation distance error is statistically fused with the measurement noise error generated during the time difference of arrival (TDOA) extraction process. Specifically, the fusion process involves extracting the variance of the measurement noise error and the variance of the equivalent propagation distance error, then superimposing them to obtain the total error variance of the TDOA observation. Based on this total error variance, an TDOA error model incorporating time synchronization uncertainty is constructed. This model describes the actual observed TDOA as follows:

[0026] in, Indicates the first The observation distance difference between each communication monitoring station and the reference station and These represent the drones up to the [number]th [number]. The theoretical geometric distance between the communication monitoring station and the reference station This represents the measurement noise error term in the time difference of arrival extraction process. The equivalent propagation distance error term is represented by the time synchronization uncertainty; the total error variance of the arrival time difference observation is:

[0027] in, This represents the total error variance. The variance represents the measurement noise error. This represents the variance of the equivalent propagation distance error. Through this step, the time synchronization error between communication monitoring stations is explicitly modeled, that is, the time synchronization error is introduced into the observation equation as an independent error component with clear statistical characteristics, and corresponding weights are assigned to this error component during the solution process, so as to automatically compensate for or suppress the influence of the time synchronization error in the positioning solution.

[0028] Steps 1-4: 3D Fusion Positioning Steps in the Communication Domain Based on the 3D angle of arrival measurements obtained in steps 1-2 and the time difference of arrival measurements containing time synchronization uncertainties constructed in steps 1-3, a joint nonlinear observation model is established. This model uses the 3D position coordinates of the UAV as the state variables to be solved, and combines the angle measurement equations of each monitoring station with the hyperboloid equations of the time difference into a set of observation equations. A weighted least squares algorithm is used to iteratively solve the nonlinear observation model.

[0029] In the solution process, the geometric center coordinates of each communication monitoring station are set as the initial values ​​for iteration. An iteration step size threshold and a maximum number of iterations are preset based on the actual positioning scenario as criteria for algorithm convergence. A weighting matrix W is constructed using the reciprocal of the total error variance calculated in step 3 to reflect the reliability differences between different observation sources. Simultaneously, a first-order Taylor expansion of the observation equation is performed at the current estimated position to calculate the Jacobian matrix H of the observation function with respect to the position coordinates. Iterative corrections are then used to obtain the communication domain position estimation result of the UAV in three-dimensional space. Finally, based on the linearized Jacobian matrix H and the weighting matrix W, the covariance matrix P corresponding to the communication domain position estimation result is calculated. This matrix quantitatively describes the geometric distribution of the positioning error in three-dimensional space, and its calculation formula is: P = (H... T WH) -1 .in,( ) Indicates matrix transpose, ( )-1 This represents finding the inverse of a matrix.

[0030] Step 2: Physical Domain UAV Trajectory Perception Steps Radar or electro-optical tracking equipment deployed in low-altitude airspace is selected as the physical sensing information source to continuously detect UAV targets. Specifically, the target slant range, azimuth, and pitch angle measurement data output by the physical sensing equipment are acquired and transformed into position coordinates in a three-dimensional Cartesian coordinate system using a coordinate transformation algorithm, thereby obtaining the physical sensing trajectory position of the UAV in three-dimensional space. A physical domain observation uncertainty model is established for the physical sensing trajectory position. This model describes the three-dimensional position observation value output by the physical sensing equipment as a linear superposition of the UAV's true position and measurement noise, where the measurement noise is modeled as a random variable following a zero-mean Gaussian distribution to reflect the inherent observation error characteristics of the physical sensor. Based on the model, the UAV's physical domain position estimation result and its corresponding covariance matrix are determined. Here, it refers to constructing the measurement noise covariance matrix based on the hardware indicators such as the ranging accuracy and angle measurement accuracy of the physical sensing equipment. This covariance matrix is ​​used to quantitatively characterize the measurement uncertainty range and error distribution characteristics of the physical sensor in each coordinate axis direction of three-dimensional space under the current observation conditions.

[0031] Step 3: Physical-Communication Domain Consistency Constraint Determination Steps The 3D position estimation results and their covariance obtained in step 1 are fused with the physical domain position estimation results and their covariance obtained in step 2. Specifically, the position deviation vector between the communication domain position estimates and the physical domain position estimates is calculated, and their covariance matrices are superimposed to reflect joint uncertainty. Based on this position deviation vector and the superimposed covariance matrix, a Mahalanobis distance statistic characterizing the statistically significant deviation between the two spatial positions is constructed as a physical-communication domain consistency indicator, as shown in the appendix. Figure 2 As shown. Based on statistical hypothesis testing theory, the above consistency discrimination index follows a chi-square distribution; the degrees of freedom of the distribution are determined according to the dimension of the three-dimensional position coordinates, and combined with the system's requirements for the false alarm rate of anomaly detection, the corresponding chi-square distribution critical value is selected as the preset threshold, and then the consistency discrimination index is... With preset threshold A comparison is made. In this embodiment, k=3; combined with the system's false alarm rate for anomaly detection. The requirement, the preset threshold Determined according to the following right-tail probability formula:

[0032] in, Let x be the probability density function of a chi-square distribution with k degrees of freedom, where x is an integral variable; in practical applications, if a false alarm rate is preset... If the value is 0.01, the corresponding preset threshold is obtained by querying the chi-square distribution table with 3 degrees of freedom. The value was 11.34; subsequently, the consistency discrimination index was... With preset threshold Comparison: When the calculation results > If the communication positioning results are statistically significantly inconsistent with the physical perception trajectory, it can be determined that the UAV target is engaging in identity deception or abnormal trajectory behavior.

[0033] Step 4: Unmanned Aerial Vehicle Behavior Identification and Alarm Procedures Based on the consistency judgment results of step 3, and combined with the consistency changes of physical layer parameters such as the received signal strength and Doppler frequency shift monitored in real time, abnormal behavior of the UAV target is identified. Specifically, it determines whether the attenuation trend of the received signal strength matches the change in target distance, and whether the measured signal frequency shift matches the target's physical flight speed. When the consistency judgment indicators in step 3 show anomalies, or when logical paradoxes are detected in the above communication signal characteristics (such as a mismatch between signal strength and distance, or a mismatch between Doppler frequency shift and speed), it is comprehensively determined that the UAV target exhibits abnormal behavior. In this case, an anomaly alarm message is output for the low-altitude airspace management system to provide risk warnings, flight control, or further action.

[0034] Example 2 In this embodiment, taking a drone detection scenario in a core urban area as an example, the specific implementation process of the three-dimensional drone abnormal behavior detection method based on physical-communication domain consistency constraints described in this invention is explained in detail. In this application scenario, the system works collaboratively with a physical domain radar device with three-dimensional detection capabilities through three communication monitoring stations deployed in the perimeter area. In the initial processing timing stage, each communication monitoring station synchronously intercepts the radio frequency complex signal of the drone's downlink communication link and converts the signal into a digital IQ sequence format. A nanosecond-level timestamp is added to each frame of signal data using a local high-precision atomic clock. Subsequently, the central processing unit receives angle measurement data and timestamped signal packets from each station in parallel, extracts the observation vector with an azimuth angle of 120.5° and an elevation angle of 15.2° using spatial spectrum estimation, and simultaneously calculates the inter-station arrival time difference as 1.5 μs using a generalized cross-correlation function. In the data fusion processing sequence, the system automatically retrieves the pre-stored 10ns inter-station synchronization error parameter, converts it into a 3m equivalent propagation distance error component, and linearly superimposes it with the 4m² measurement noise variance in the time difference extraction process, thereby constructing a communication domain observation matrix that includes location uncertainty.

[0035] At the same time, the radar equipment completes real-time acquisition of the UAV's physical reflection characteristics and outputs the target's three-dimensional position estimation vector in the NED geographic coordinate system. (Unit: meters), the corresponding filter covariance matrix is During the decision-making phase, the system uses the three-dimensional position estimation results, including X, Y, and Z components, calculated in the communication domain. Aligning with the physical position vector output by the radar, the deviation vector in three-dimensional space is calculated as follows: By summing the estimation error covariance matrices of the communication domain and the physical domain, a weight matrix reflecting the joint uncertainty of the two domains is constructed, and then the Mahalanobis distance scalar index characterizing spatial consistency is calculated to be 15.6. The system selects 11.34, corresponding to a false alarm rate of 1%, as the chi-square distribution critical threshold based on the degrees of freedom of the three-dimensional coordinates. Since the calculated Mahalanobis distance scalar of 15.6 exceeds the threshold of 11.34, the processor immediately determines that the UAV is exhibiting abnormal behavior where its communication identity and physical trajectory do not match, and outputs an abnormal alarm message; if the calculated index does not exceed the threshold, the target identity is determined to be consistent, and real-time monitoring continues.

[0036] Example 3 Based on the same inventive concept, the present invention also provides an electronic device, including one or more processors; a storage device for storing one or more programs; and when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in Embodiment 1.

[0037] Since the device described in Embodiment 3 of this invention is the electronic device used in implementing the 3D UAV abnormal behavior detection method based on physical-communication domain consistency constraints in Embodiment 1 of this invention, those skilled in the art can understand the specific structure and variations of this electronic device based on the method described in Embodiment 1 of this invention, and therefore will not be repeated here. All electronic devices used in any method of this invention fall within the scope of protection of this invention.

[0038] Example 4 Based on the same inventive concept, the present invention also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the method described in Embodiment 1.

[0039] Since the computer-readable medium described in Embodiment 4 of this invention is the same computer-readable medium used in implementing the three-dimensional UAV abnormal behavior detection method based on physical-communication domain consistency constraints in Embodiment 1 of this invention, those skilled in the art can understand the specific structure and variations of this computer-readable medium based on the method described in Embodiment 1 of this invention, and therefore will not be repeated here. All computer-readable media used in any method of this invention are within the scope of protection of this invention.

[0040] Example 5 Based on the same inventive concept, this embodiment discloses a three-dimensional UAV abnormal behavior detection system with physical-communication domain consistency constraints for low-altitude airspace management, comprising: a communication monitoring station module for acquiring UAV wireless communication signals and extracting three-dimensional angle of arrival and time of arrival information; a communication domain three-dimensional positioning module for performing three-dimensional fusion positioning based on the angle of arrival and time of arrival difference, and outputting the position estimation result and its covariance; a physical domain perception module for acquiring the physical trajectory position of the UAV and its uncertainty information; a consistency discrimination module for performing consistency constraint discrimination on the position estimation results of the communication domain and the physical domain; and an anomaly identification and alarm module for outputting an abnormal behavior alarm based on the consistency discrimination result.

[0041] Since the system described in Embodiment 5 of this invention is the system used to implement the three-dimensional UAV abnormal behavior detection method based on physical-communication domain consistency constraints in Embodiment 1 of this invention, those skilled in the art can understand the specific structure and variations of this system based on the method described in Embodiment 1 of this invention, and therefore will not be repeated here. All systems used in any method of this invention's embodiments fall within the scope of protection of this invention.

[0042] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting abnormal behavior of a three-dimensional unmanned aerial vehicle (UAV) based on physical-communication domain consistency, characterized in that, Includes the following steps: Step 1: Obtain the 3D position estimation result of the UAV in the communication domain and its first covariance. The 3D position estimation result in the communication domain is based on the 3D angle of arrival and time difference of arrival information of wireless signals collected by multiple communication monitoring stations. In the positioning calculation process, the statistical characteristics of the time synchronization error between communication monitoring stations are explicitly introduced into the observation model as an error term. Step 2: Obtain the 3D position estimation result of the UAV in the physical domain and its second covariance. The 3D position estimation result in the physical domain is based on the perception data of radar or optical imaging equipment. Step 3: Calculate the Mahalanobis distance of the difference vector between the 3D position estimation results in the communication domain and the physical domain, using the inverse matrix of the sum of the first covariance and the second covariance as weights, as a consistency discrimination index. The Mahalanobis distance is compared with the discrimination threshold determined based on the chi-square distribution. If the Mahalanobis distance exceeds the discrimination threshold, it is determined that the communication domain perception result and the physical domain perception result are inconsistent in a statistical sense. Step 4: When the determination result in step 3 is inconsistent, it is determined that the UAV has abnormal behavior and an abnormal alarm message is output.

2. The method according to claim 1, characterized in that, In step 1, the statistical characteristics of the time synchronization error are explicitly introduced into the observation model as an error term. Specifically, this includes converting the time synchronization error between each communication monitoring station into an equivalent distance error, and fusing the statistical variance of the equivalent distance error with the measurement noise variance of the arrival time difference to construct the comprehensive error term in the observation model.

3. The method according to claim 2, characterized in that: In step 1, a nonlinear observation equation is constructed based on the three-dimensional angle of arrival, time difference of arrival, and comprehensive error term, and a weighted least squares algorithm is used for iterative solution to obtain the three-dimensional position estimation result of the communication domain and its first covariance.

4. The method according to claim 1, characterized in that, Step 2 specifically includes: The raw observation data of the UAV is acquired by radar or optical imaging equipment, including slant range, azimuth angle and pitch angle. The original observation data is transformed to obtain the physical domain position observation values ​​in the three-dimensional Cartesian coordinate system; Based on the physical domain position observations and the preset sensor measurement noise model, the three-dimensional position estimation result of the physical domain and its second covariance are determined.

5. The method according to claim 1, characterized in that: The discrimination threshold determined in step 3 based on the chi-square distribution is determined by setting a pre-set confidence level, determining the degrees of freedom of the chi-square distribution according to the dimension of the three-dimensional position coordinates, and querying the corresponding chi-square distribution critical value.

6. The method according to claim 5, characterized in that: The discrimination threshold is selected directly from the chi-square distribution critical value table based on the system's requirements for the false alarm rate of anomaly detection and the chi-square distribution degrees of freedom that the Mahalanobis distance follows.

7. The method according to claim 1, characterized in that: In step 1, multiple communication monitoring stations are deployed with known three-dimensional spatial coordinates. The wireless communication signal is a signal transmitted or reflected by the UAV. The three-dimensional angle of arrival includes azimuth and pitch angle.

8. An electronic device, characterized in that, include: At least one processor; A memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method of any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

10. A three-dimensional UAV abnormal behavior detection system based on physical-communication domain consistency, used to implement the method of any one of claims 1 to 7, characterized in that, include: A communication domain positioning module is used to perform step 1 and obtain the three-dimensional position estimation result of the communication domain and its first covariance. The physical domain perception module is used to perform step 2 and obtain the three-dimensional position estimation result of the physical domain and its second covariance. The consistency discrimination module is used to perform step 3, and to perform statistical discrimination of consistency between the communication domain and the physical domain. The abnormal alarm module is used to execute step 4 and output the abnormal alarm information.

Images