Low-altitude unmanned aerial vehicle target detection method based on spectrum anomaly recognition

Abstract

The invention relates to the technical field of unmanned aerial vehicle detection, in particular to a low-altitude unmanned aerial vehicle target detection method based on spectrum anomaly recognition, which comprises the following steps: S1, collecting radio frequency signals of a monitoring area, and generating a plurality of time domain sampling sequences; s2, performing complex number self-correlation operation on the time domain sampling sequence, and extracting dynamic characteristics with rotation invariance; and S3, determining the target existence of the unmanned aerial vehicle according to the trajectory closing index of the dynamic characteristics in the phase space. According to the method, passive, non-contact and high-precision detection of a low-altitude unmanned aerial vehicle target is achieved, compared with a traditional energy detection or frequency spectrum scanning method, rotor signals can still be stably extracted in a non-line-of-sight, multi-path reflection or lightning noise environment, the closed structure of a rotor track is effectively recognized, and the detection accuracy is improved. And the problem that the traditional'head and tail point distance 'is easy to misjudge is avoided.

Description

Technical Field

[0001] The present invention relates to the field of drone detection technology, and in particular to a low-altitude drone target detection method based on spectrum anomaly recognition. Background Art

[0002] With the increasing presence of low-altitude small drones in sensitive areas such as airports and power stations, the security threat posed by illegal drone intrusions urgently requires effective monitoring and intervention methods. Existing drone detection technologies primarily include radar detection, optical recognition, and radio frequency signal sensing. Among these, passive detection methods based on radio frequency signals have become a key research and engineering practice due to their low cost, flexible deployment, and independence from weather and lighting conditions.

[0003] However, existing RF detection methods generally rely on energy detection, spectrum template matching, or feature extraction from known communication protocols. For example, some methods determine the presence of a target by identifying the carrier frequency, bandwidth, or modulation mode of a drone's image transmission signal. However, these methods are extremely sensitive to spectrum variations and are prone to misjudgment or missed detection in scenarios with multi-source interference or frequency-hopping communications. Furthermore, spectrum energy algorithms inherently discard the phase information in the signal, whereas phase structure often contains richer and more stable dynamic characteristics.

[0004] For example, during the operation of a rotorcraft drone, its rotating blades periodically modulate the reflected signal, causing the received signal to exhibit a specific rotational behavior in the complex plane. This physical phenomenon has been underutilized in traditional techniques. Existing literature has largely focused on extracting features such as amplitude, envelope, and frequency, while ignoring the deterministic phase trajectory patterns induced by rotor motion. Especially in environments with strong interference, highly reliable identification based solely on energy signatures is difficult, and robust identification methods based on physical mechanisms are lacking. Summary of the Invention

[0005] The present invention provides a low-altitude UAV target detection method based on spectrum anomaly recognition. It can extract phase dynamic structure characteristics from radio frequency signals and combine them with physical properties such as the stability and closure of their rotational trajectories to perform target identification. This new UAV detection method can effectively improve the anti-interference detection capability of low-altitude UAVs, reduce the false alarm rate, and have multi-target distinction and model adaptation capabilities.

[0006] A low-altitude UAV target detection method based on spectrum anomaly recognition includes the following steps: S1: Collect radio frequency signals in the monitoring area and generate a complex time domain sampling sequence; S2: performing a complex autocorrelation operation on the time domain sampling sequence to extract dynamic features with rotation invariance; S3: Determine the existence of the UAV target based on the trajectory closure index of the dynamic characteristics in the phase space.

[0007] Optionally, collecting the radio frequency signal of the monitoring area in S1 includes collecting the radio frequency signal of the monitoring area through a dual-channel orthogonal down-conversion receiver, wherein the local oscillator frequency sweep covers the operating frequency band of the drone.

[0008] Optionally, the S1 further includes synchronously sampling the in-phase branch and the quadrature branch signals in baseband processing to obtain a time domain sampling sequence in a complex form.

[0009] Optionally, the S1 further includes adjusting the sampling gain through an automatic gain control circuit so that the modulus standard deviation of the time domain sampling sequence is within a stable range.

[0010] Optionally, extracting the dynamic features with rotation invariance in S2 specifically includes: S21: performing a delayed complex autocorrelation operation on the complex time-domain sampling sequence to generate a complex autocorrelation function sequence; S22: extracting a phase derivative sequence of the complex autocorrelation function sequence; S23: Calculate the cumulative displacement of the phase derivative sequence on the complex plane as a dynamic feature of rotational invariance.

[0011] Optionally, S21 specifically includes processing the complex signal of the received time domain sampling sequence, comparing the phase difference between the signal at the current moment and the historical signal, constructing a phase difference sequence that changes with time, calculating the rate of change of the phase difference sequence with time, that is, the phase change rate, reflecting the change in the rotation speed of the signal in the complex plane, accumulating the phase change rate within the time interval, and generating a complex autocorrelation function sequence.

[0012] Optionally, the S22 specifically includes performing a phase angle extraction operation on the generated complex autocorrelation function sequence, calculating the phase angle of the complex autocorrelation value corresponding to each moment, using sliding difference to perform difference operation on the phase angles of two adjacent moments, and normalizing them with a time interval to form a continuous phase change rate sequence, the phase change rate sequence is the phase derivative sequence, which numerically represents the instantaneous rotation speed of the signal trajectory in the complex plane, and the S23 performs an integral operation on the phase derivative sequence on the time axis based on the obtained phase derivative sequence, and calculates its cumulative phase change in a continuous time window, the integral operation is the time accumulation of the signal phase rotation speed to reflect the total number of rotations around its complex trajectory, the integral operation window covers multiple rotor cycles, and the integral operation result is the rotational invariance dynamic characteristic.

[0013] Optionally, the dynamic feature includes a real part and an imaginary part, and S3 includes drawing the phase space trajectory of the dynamic feature in a complex plane coordinate system with the real part as the horizontal axis and the imaginary part as the vertical axis.

[0014] Optionally, S3 quantifies the path structure integrity of the phase space trajectory by vector cross product and integration, and calculates the trajectory closure index of the phase space trajectory.

[0015] Optionally, determining the existence of the drone target includes multiple criteria determination: S31, the trajectory closure index is greater than a predetermined closure threshold; S32, the trajectory enclosed area is greater than the lower limit of the minimum effective rotor enclosed area; S33, the trajectory rotation direction consistency is greater than a predetermined percentage; When the S31-S33 conditions are met, it is determined that a drone target exists.

[0016] Beneficial effects of the present invention: The present invention constructs a time-domain sampling sequence in complex form and completely retains the signal phase information, effectively overcoming the problem of existing methods relying on amplitude or spectrum intensity and being susceptible to interference. It utilizes the phase rotation law caused by the motion of the UAV rotor and maps it into dynamic trajectory characteristics on the complex plane. Without relying on target cooperative equipment, tag information or communication protocols, it achieves passive, non-contact, and high-precision detection of low-altitude UAV targets. Compared with traditional energy detection or spectrum scanning methods, the present invention can still stably extract rotor signals in non-line-of-sight, multipath reflection or lightning noise environments.

[0017] This paper uses the phase derivative of drone rotor reflection signals as a key dynamic quantity and proposes a mechanism for extracting a "rotational invariance feature." This method uses the time-integrated integral of the phase change rate to accurately characterize the number of signal rotations in complex space. This feature exhibits inherent amplitude robustness and immunity to frequency drift, effectively distinguishing rotor targets from other non-rotating interference sources (such as Wi-Fi pulses, multipath echoes, and lightning noise). This method maintains high detection accuracy even for drone targets in low signal-to-noise ratio environments.

[0018] In the judgment stage, the present invention proposes a triple verification mechanism of "closure index-area-rotation direction consistency" based on the complex phase space trajectory. By calculating the degree of closure and directional stability of the trajectory in the complex plane, a judgment standard with geometric and physical interpretation is constructed. It can effectively identify the closed structure of the rotor trajectory and avoid the problem of easy misjudgment of the traditional "head and tail point distance". BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention; Figure 2 Schematic diagram of multiple criteria for determining the existence of drone targets according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Those skilled in the art may also implement some known technologies in other alternative ways. The accompanying drawings are only for describing the embodiments in more detail and are not intended to limit the present invention in any specific way.

[0022] like Figure 1-Figure 2 As shown, a low-altitude UAV target detection method based on spectrum anomaly recognition includes the following steps: S1: Collect the radio frequency signal in the monitoring area and generate a complex time domain sampling sequence.

[0023] S11, frequency band sweep receiving mechanism: The RF signal of the monitoring area is collected by a dual-channel orthogonal down-conversion receiver, where the local oscillator frequency Linear sweep is performed in the range of 2.4-5.8GHz to cover the common drone operating frequency bands.

[0024] Currently, commercially available low-altitude drones (such as DJI, Parrot, FPV drones, etc.) generally use the following frequency bands for remote control, image transmission, and data link communication: 2.4GHz ISM band: widely used in remote control links, compatible with WiFi image transmission; 5.8GHz ISM band: mainly used for high-definition video transmission and low-latency fly-through flight control; Intermediate frequency bands (such as 3.5 and 4.9 GHz): Used by some customized or frequency-hopping devices and have regional and confrontational purposes.

[0025] Therefore, setting the sweep range to 2.4–5.8 GHz can effectively cover the RF transmission frequency bands of most drones and have broad-spectrum detection capabilities.

[0026] Modern drones often use FHSS (frequency hopping spread spectrum) or multi-frequency cooperative communication technology. Its frequency will dynamically switch within the above range. Through linear sweeping, different frequency points can be monitored at each moment to capture the instantaneous signal of frequency hopping. Sweeping can achieve spectrum separation of multiple drones existing at the same time and effectively deal with multi-machine interference; combined with the time window length (such as one rotor cycle) for extracting complex phase trajectories in subsequent steps, ensure that the sweep dwell time is sufficient for feature recognition.

[0027] S12, orthogonal baseband signal generation: In baseband processing, the in-phase (I) branch and the quadrature (Q) branch signals are collected simultaneously to form a complex time domain sampling sequence: ;in, is the sampling value of the in-phase channel, which comes from The mixing result is Indicates the orthogonal channel sampling value, which comes from The mixing result is is an imaginary unit representing the phase shift of the quadrature component.

[0028] S13, dynamic gain control mechanism: adjust the receiving gain in real time through the automatic gain control circuit (AGC) to keep the standard deviation of the complex sequence modulus value stable within the effective analysis range, that is: ;in, represents the amplitude of the complex sequence, , Indicates the calculation of the standard deviation of the amplitude within the sliding time window; The dynamic adjustment logic is as follows: If Var , then reduce the AGC gain; If Var , then increase the AGC gain.

[0029] The subsequent S2 of the present invention relies on performing autocorrelation operations on complex time domain signals to extract phase dynamic trajectories. However, in the actual measurement environment, if the signal amplitude fluctuates too much, amplitude artifacts will appear in the autocorrelation results, thereby interfering with the phase change rate. Extraction. The motion of the drone's rotor causes a continuous and smooth phase rotation trajectory, while the amplitude variation is an independent modulation process. If amplitude fluctuations are not suppressed, nonlinear perturbations will appear in the autocorrelation results, obscuring the original phase closure trajectory. Depending on the drone model, distance, and obstruction conditions, the received signal strength may fluctuate by multiple orders of magnitude. Therefore, the gain of the receiving link needs to be dynamically adjusted to avoid signal oversaturation (distortion) or undersaturation (overwhelmed by noise). Therefore, S13 describes the introduction of an automatic gain control (AGC) mechanism during signal reception, which is used to adjust the gain of the receiving channel in real time to ensure that the amplitude fluctuation of the acquired complex time-domain sampled signal remains within a reasonable range. Specifically, by calculating the standard deviation of the complex signal modulus (i.e., amplitude) within a sliding time window, the system automatically reduces the gain when the amplitude fluctuation exceeds the upper threshold of 1.2V; when it falls below the lower threshold of 0.5V, the gain is increased.

[0030] In the present invention, the signal input to the complex autocorrelation module is guaranteed to have uniform amplitude statistical characteristics, so that the phase rotation feature has a stable recognition basis within different sampling windows, reducing misjudgments due to amplitude non-stationarity (such as short-term pulse signals from non-UAV sources that may cause strong but spurious phase disturbances). When sweeping to different frequency points, since the signal strength varies with the frequency band, the AGC ensures that the amplitudes of the signals collected at different frequency points are all within the effective analysis range, which helps to unify the subsequent processing standards.

[0031] S2: Performing a complex autocorrelation operation on the time domain sampling sequence to extract dynamic features with rotation invariance.

[0032] S21, time-delay autocorrelation operation: complex time domain sampling sequence Execute fixed delay Complex autocorrelation calculation of the autocorrelation function sequence : ; in, is the autocorrelation delay (the length of the delay), is the complex sampling sequence output by S1, is its complex conjugate, represents the sampling time interval, Indicates the number of sampling points corresponding to the delay, which is related to the delay The corresponding sample point offset number, , is the total number of sampling points in the sliding window, The delay is The time variable is The time-varying autocorrelation function value of is the current sample index (from 0 to N-1).

[0033] S21 actually processes the received complex signal first, comparing the phase difference between the current signal and the earlier signal. This comparison process will be carried out continuously to construct a "phase difference sequence" that changes with time, which indicates how the signal phase rotates with time; then calculates the rate of change of this phase difference with time, that is, the phase change rate. This step reflects how fast the signal rotates on the complex plane, similar to "how fast the rotor is rotating"; finally, this phase change rate is accumulated within a time interval to obtain a quantity representing "how many revolutions in total". This quantity is a global feature that can clearly determine whether the signal comes from a target with a stably rotating rotor (such as a drone), because only this type of target will produce regular and continuous phase rotation.

[0034] In summary, this calculation process first examines how the signal's phase changes, then the rate of change, and then accumulates these changes to ultimately generate a key signature that can identify whether the rotor is rotating. This signature is independent of signal strength and unaffected by frequency drift, making it a core technology for drone identification.

[0035] S22, phase derivative extraction, calculation of autocorrelation function The time direction phase change rate of is obtained to obtain the phase derivative sequence: ; in, represents the argument (phase) of a complex number, The discrete intervals representing the phase derivative (usually the same as the sampling interval the same as or an integral multiple thereof), For the current time point The phase derivative value, that is, the instantaneous phase change rate.

[0036] S22 observes how the phase (that is, the angle of the complex number on the complex plane) changes over time from the continuous autocorrelation results. Specifically, it takes the phase values ​​at the current moment and the next moment, calculates the difference between them, and then divides it by the time interval between them. This gives the speed of phase change.

[0037] In other words, the speed of phase rotation is determined during continuous observation. This speed of change is called the "phase derivative," which reflects whether there is a stable, continuous rotation trend in the signal. If so, it indicates that the signal contains regular phase rotation caused by the periodic mechanical motion of the rotor, a characteristic that is crucial for drone identification.

[0038] S23, the rotation invariance feature is constructed by integrating the phase derivative sequence on the time axis to obtain the cumulative trajectory length of the phase rotation, which is recorded as the rotation invariance dynamic feature: ;in, 、 Indicates the start and end boundary time of the integration time window. It is recommended to cover multiple rotor cycles (such as z3). It represents the rotation invariance characteristic quantity, that is, the integral result of the phase derivative, which represents the total amount of cumulative phase rotation of the rotation trajectory calculated on the complex plane, and theoretically satisfies ,in Indicates the number of complete rotations, represents a complete circle rotation (i.e. 360 degrees), represents the number of complete rotations that the signal completes in the complex plane, which is an integer. represents the set of integers, including positive integers, negative integers and zero, Indicates the small time step used for integration or derivative on the time axis.

[0039] For example, if the signal rotates three times around the origin in the complex plane, then ; If it only turns one and a half circles, then ,at this time , but since the present invention focuses on identifying “closed trajectories”, the ideal goal is near and is an integer; as time goes on, It will grow in a step-by-step manner, with each increase Corresponds to one complete rotor cycle.

[0040] The rotational invariance characteristic in this context refers to the total phase accumulated by a signal rotating around the origin in the complex plane, reflecting whether the signal exhibits continuous, regular, and consistent rotational behavior. When the drone's rotor rotates at high speed, the RF signal reflected by it appears at the receiving end as a phase trajectory that continuously rotates at a constant speed in the complex plane. This rotational behavior is not affected by the following factors: The starting phase of the signal (because we only care about the number of revolutions and don’t care where it starts); The amplitude fluctuation of the signal (because only the direction change is considered, not the intensity); Slight frequency changes or frequency hopping (because it is not dependent on a specific frequency); Therefore, this feature has good rotation invariance and has significant target discrimination ability: The accumulated phase of a drone signal shows a linear increase or step-wise transition, such as 2π for one revolution and 4π for two revolutions, indicating a "closed" trajectory. The phase changes of an interference signal, on the other hand, show irregular oscillations or repeated reversals, failing to accumulate into a complete circle. Therefore, the "rotational invariance characteristic," the cumulative number of phase rotations, is used to determine whether the signal has the stable circular trajectory of rotorcraft motion and is a key indicator of drone presence.

[0041] The core of this invention is to utilize the phase rotation characteristics caused by the motion of the drone rotor, which manifests as a stable and regular circular trajectory in the complex plane. Traditional methods only focus on amplitude and frequency characteristics, ignoring this critical phase information. Therefore, this invention extracts the physical trajectory characteristics of this dynamic process through S21-S23: S21 performs autocorrelation processing on the received complex signal sequence after a fixed time delay. This involves performing a complex product operation on each current signal point with the signal point preceding it by a certain time delay, and then averaging the results. This operation effectively enhances periodic structures while suppressing non-periodic interference and noise. The output of this operation is a time-varying complex number sequence that reflects the cumulative phase effect of the rotor's periodic motion. S22 extracts the phase (i.e., the angle of the complex number in the complex plane) of the complex autocorrelation sequence obtained in the previous step point by point. The phase sequence is then differentiated and divided by the time interval to obtain the rate of phase change, or the phase derivative. Essentially, this calculates the instantaneous rotational speed of the signal, which can be understood as the speed at which the rotor "circles" in the complex plane. For drone targets, this rotational speed is approximately stable, while environmental interference manifests as irregular jumps. S23 accumulates or integrates these phase derivatives over a longer period of time to obtain the total phase change, or the rotation trajectory length. This quantity mathematically corresponds to the total number of revolutions around the origin in the complex plane. If the signal comes from a drone, due to the stable motion of the rotor, its cumulative trajectory is a closed circle or a regular spiral in the complex plane, resulting in a stable and neat step increase in the integral value. Represents one complete rotation of the rotor.

[0042] This method utilizes phase structure rather than amplitude / frequency: the phase rotation of the drone rotor reflection signal is a direct reflection of its physical motion, while the amplitude and frequency are subject to strong interference factors such as occlusion, multipath, and power variations. Therefore, describing its dynamic behavior through phase derivatives is more robust than traditional spectrum / energy features. This method uses time-delay autocorrelation to amplify periodic signals (repeated fluctuations caused by rotor rotation) while naturally filtering out aperiodic structures (such as lightning pulses and WiFi bursts), improving the signal-to-noise ratio. Compared to instantaneous phase values, the total number of rotations after integration is insensitive to factors such as starting phase and amplitude jitter, demonstrating excellent robustness and cross-scenario consistency. By calculating derivatives and integrals over a sliding window, it can automatically adapt to slight changes in rotor speed while maintaining trajectory closure analysis.

[0043] The unique dynamic signal trajectory of a drone is extracted as a basis for determining its unique existence, providing a quantifiable indicator for subsequent determination steps (trajectory closure), thereby improving the ability to identify targets in low signal-to-noise ratio, non-line-of-sight, or complex environments. This solution introduces "rotational invariance" as a drone identification feature. Unlike traditional spectrum detection and energy detection, this solution focuses on the "trajectory behavior" of rotor motion in complex signals. This achieves an amplitude-independent, frequency-drift-robust, and structurally robust identification mechanism. Even if a drone's power decreases, its frequency jumps, or its rotor speed changes slightly, it can be effectively detected as long as its rotational structure remains unchanged. Compared to traditional methods, this solution maintains stable identification capabilities in complex electromagnetic environments such as urban multipath scenarios and near-field WiFi interference.

[0044] In summary, this solution separates the phase rotation law carried by the UAV rotor motion from the mixed signal through the "physical trajectory extraction" path of complex autocorrelation + phase derivative + phase integral, providing a highly robust technical means for low-altitude UAV target detection, which is the technical foundation of the core recognition mechanism of this invention.

[0045] In the present invention, the kinetic characteristics is a complex function that varies with time and has the form: ;in: represents the projection of the complex number locus on the real axis, which is called the real part; Im represents the projection of the complex number trajectory onto the imaginary axis, which is called the imaginary part; is the imaginary unit, representing the vertical direction in the complex plane; The phase derivative The complex characteristic trajectory after integration over time reflects the cumulative behavior of phase changes in complex space.

[0046] Is a complex number, consisting of two parts: the real part (Re) and the imaginary part (Im); Re yes One of the components of , responsible for representing its component in the direction of the real axis, if Understood as a "moving point" on the complex plane, then Re It is its position coordinate in the horizontal direction.

[0047] S3: Determine the existence of the UAV target based on the trajectory closure index of the dynamic characteristics in the phase space.

[0048] S31, construct phase space trajectory: the rotationally invariant dynamic feature sequence extracted from S2 Mapped to the complex plane coordinate system, the phase space trajectory is plotted with the real part as the horizontal axis and the imaginary part as the vertical axis: Horizontal axis: Re represents the real part of the complex number characteristic; Vertical axis: Im The imaginary part trajectory points representing the complex features are connected in time sequence to form the phase space evolution trajectory of the target signal on the complex plane.

[0049] S32, calculate the trajectory closure index: the phase space trajectory path is calculated by integration The structural integrity of the for: ; in, is the position vector of a point on the trajectory path, is the differential displacement vector of the point, represents the vector cross product between adjacent trajectory segments, For the entire trajectory path The projected area enclosed by the represents a closed integral along the trajectory path.

[0050] The "trajectory closure index" is calculated to determine whether the signal trajectory has a stable, periodic rotation closure feature in a quantitative, disturbance-resistant and physically interpretable way. This feature is one of the core manifestations of low-altitude UAV rotor signals. Therefore, a geometric measurement method is needed that can not only reflect the shape but also resist interference and speed changes.

[0051] The common method to judge "closure" is to compare the distance between the starting point and the end point of the trajectory. However, in actual scenarios: The rotor speed may fluctuate slightly, causing the last turn of the trajectory to not land accurately back to the starting point; The signal is affected by multipath, obstruction, etc., causing the tail trajectory to deviate; This method is highly sensitive to the number of trajectory cycles and start and end times, has poor robustness and a high misjudgment rate.

[0052] The core of this closed index is to treat the complex trajectory as a two-dimensional vector path, perform a vector cross product operation on each path segment, and integrate and accumulate the results. Its principle is related to the classic moment integral and area measurement: The modulus of the cross product represents the “local rotation area” enclosed by two adjacent trajectory points; Integrating the total amount of all cross products is equivalent to measuring the “structural symmetry” or “rotational integrity” of the entire path; Dividing by the area actually enclosed by the trajectory can normalize the tightness of closure of circular and non-circular paths.

[0053] Therefore, the above calculation will not fail due to starting point offset or slight abnormality in a certain section, and can stably reflect the overall rotor motion characteristics.

[0054] The ratio of the path integral result to the equivalent circular area is calculated to achieve unit scale unification. If the trajectory is highly closed and close to a circle, the closure index is close to 1. If the trajectory shows non-rotor structural characteristics such as divergence, reentry, and oscillation, the closure index is much less than 1.

[0055] S33, set target recognition criteria: Combine the trajectory geometry and dynamic direction characteristics to define the following three judgment conditions: 1. Closure index threshold: ; Used to measure whether the trajectory is close to a circle or a stable closed structure.

[0056] 2. Track area threshold: ;in is the lower limit of the minimum effective rotor enclosed area, set as: ; Used to filter trajectory deviations caused by slight disturbances or random noise.

[0057] 3. Rotation direction consistency: The proportion of adjacent vectors in the trajectory that maintain consistent rotation directions must meet the following requirements: consistency > 90%. The sign of the vector cross product is used to determine whether the trajectory maintains unidirectional rotation and filter out AC power supply disturbances.

[0058] This invention visualizes complex space trajectories, reducing complex phase change processes to geometric structures, and graphically distinguishing drone signals from background interference. The closure index is superior to the traditional "start-end distance" method. Even under conditions of rotor speed fluctuations or non-uniform motion, as long as the trajectory structure is closed, the closure index remains stable, whereas traditional methods are prone to misjudgment. Drone rotors have clear mechanical steering patterns, while background noise or lightning interference typically manifests as non-directional or bidirectional jumps. Incorporating directional consistency judgment can significantly reduce the false alarm rate.

[0059] This paper provides high-dimensional structural features based on complex trajectories for accurate identification of the presence of periodic rotor-like motion signals. By mapping rotationally invariant dynamic features into complex plane phase space and designing a triple verification mechanism involving a closure index, an area threshold, and directional consistency, it achieves high-precision identification of periodic UAV rotor signals.

[0060] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0061] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A low-altitude UAV target detection method based on spectrum anomaly recognition, characterized in that: The following steps are involved: S1: Collect radio frequency signals in the monitoring area and generate a complex time domain sampling sequence; S2: performing a complex autocorrelation operation on the time domain sampling sequence to extract dynamic features with rotation invariance; S3: Determine the existence of the UAV target based on the trajectory closure index of the dynamic characteristics in the phase space.

2. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 1 is characterized in that: The collecting of radio frequency signals in the monitoring area in S1 includes collecting radio frequency signals in the monitoring area through a dual-channel orthogonal down-conversion receiver, wherein the local oscillator frequency sweep covers the operating frequency band of the drone.

3. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 1 is characterized in that: The S1 further includes synchronously sampling the in-phase branch and the quadrature branch signals in baseband processing to obtain a time domain sampling sequence in a complex form.

4. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 1 is characterized in that: The S1 further includes adjusting the sampling gain through an automatic gain control circuit so that the modulus standard deviation of the time domain sampling sequence is within a stable range.

5. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 1 is characterized in that: The dynamic features extracted from S2 with rotation invariance specifically include: S21: performing a delayed complex autocorrelation operation on the complex time-domain sampling sequence to generate a complex autocorrelation function sequence; S22: extracting a phase derivative sequence of the complex autocorrelation function sequence; S23: Calculate the cumulative displacement of the phase derivative sequence on the complex plane as a dynamic feature of rotational invariance.

6. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 5 is characterized in that: The S21 specifically includes processing the complex signal of the received time domain sampling sequence, comparing the phase difference between the current signal and the historical signal, constructing a phase difference sequence that changes with time, calculating the rate of change of the phase difference sequence with time, that is, the phase change rate, which reflects the change in the rotation speed of the signal in the complex plane, accumulating the phase change rate within a time interval, and generating a complex autocorrelation function sequence.

7. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 6 is characterized in that: The S22 specifically includes performing a phase angle extraction operation on the generated complex autocorrelation function sequence, calculating the phase angle of the complex autocorrelation value corresponding to each moment, using sliding difference to perform a difference operation on the phase angles of two adjacent moments, and normalizing them with a time interval to form a continuous phase change rate sequence. The phase change rate sequence is a phase derivative sequence, which numerically represents the instantaneous rotation speed of the signal trajectory in the complex plane. The S23 is based on the obtained phase derivative sequence, and performs an integral operation on the phase derivative sequence on the time axis to calculate its cumulative phase change within a continuous time window. The integral operation is the time accumulation of the signal phase rotation speed to reflect the total number of rotations around its complex trajectory. The integral operation window covers multiple rotor cycles, and the integral operation result is the rotational invariance dynamic characteristic.

8. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 1 is characterized in that: The dynamic characteristics include a real part and an imaginary part, and S3 includes drawing the phase space trajectory of the dynamic characteristics in a complex plane coordinate system with the real part as the horizontal axis and the imaginary part as the vertical axis.

9. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 8 is characterized in that: The S3 quantifies the path structure integrity of the phase space trajectory through vector cross product and integration, and calculates the trajectory closure index of the phase space trajectory.

10. The low-altitude UAV target detection method based on spectrum anomaly recognition according to claim 9 is characterized in that: The determination of the existence of the UAV target includes multiple criteria: S31, the trajectory closure index is greater than a predetermined closure threshold; S32, the trajectory enclosed area is greater than the lower limit of the minimum effective rotor enclosed area; S33, the trajectory rotation direction consistency is greater than a predetermined percentage; When the S31-S33 conditions are met, it is determined that a drone target exists.

Images