Method and system for generating strategy decision of vehicle-mounted unmanned aerial vehicle interference detection device

Abstract

This invention discloses a strategy decision generation method and system for vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment, comprising the following steps: Step S1: Multimodal data acquisition step, in the vehicle's driving or parked state, continuously acquiring raw perception data of the target airspace through a vehicle-mounted multimodal sensor group; the vehicle-mounted multimodal sensor group includes at least a radio monitoring direction finding unit and an electro-optical tracking unit; Step S2: Feature parameter extraction step, preprocessing the raw perception data, separating the target signals suspected to be UAVs, and extracting multidimensional feature parameters for each target signal; the multidimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the electro-optical image. This invention is capable of...

Description

Technical Field

[0001] This invention relates to the field of data analysis, specifically to a method and system for generating strategy decisions for vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment. Background Technology

[0002] In the field of vehicle-mounted drone interference detection technology, existing technologies face several critical challenges in drone threat detection and interference decision-making: The electromagnetic environment in vehicle scenarios is complex, background noise fluctuates, and the vehicle's driving / parking states cause dynamic changes in attitude and heading, making traditional fixed-threshold threat assessment methods unsuitable for complex environments and prone to false alarms or missed detections; There is a lack of scientific quantitative criteria for the correlation between drone target points and trajectories, making it difficult to effectively filter transient noise interference and accurately select persistent targets that conform to physical kinematic constraints, resulting in insufficient reliability of trajectory data for subsequent decision-making; The interference strategy does not consider the co-frequency interference of the vehicle's electronic equipment, easily causing disruptions to vehicle communication and detection... The equipment used for detection and other purposes malfunctions due to self-interference; after interference is implemented, there is a lack of effective online performance evaluation and dynamic fine-tuning mechanisms, and ineffective interference can easily consume vehicle resources continuously. Moreover, the strategy adjustment lacks intelligence and is difficult to adapt to the dynamic changes of UAV targets; when facing multiple UAV targets, there is no scientific method for scheduling interference resources, and it is impossible to achieve the optimal allocation of interference beam illumination order and dwell time under limited time and space resources. This can easily lead to resource conflicts and untimely interception of high-threat targets. The overall interference detection strategy decision-making lacks intelligence, accuracy, and adaptability, making it difficult to meet the actual application requirements of vehicle-mounted platforms for efficient and reliable response to UAV threats. Therefore, a strategy decision generation method and system for vehicle-mounted UAV interference detection equipment is proposed. Summary of the Invention

[0003] The present invention solves the above-mentioned technical problems through the following technical solution, and the present invention includes the following steps:

[0004] Step S1: Multimodal data acquisition step, in the vehicle driving or parked state, continuously acquire raw perception data of the target airspace through the vehicle-mounted multimodal sensor group; the vehicle-mounted multimodal sensor group includes at least a radio monitoring direction finding unit and an optoelectronic tracking unit;

[0005] Step S2: Feature parameter extraction step, preprocessing the raw sensing data, separating the target signals suspected to be UAVs, and extracting multi-dimensional feature parameters for each target signal; the multi-dimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image;

[0006] Step S3: Target association and matching step. The multidimensional feature parameters extracted in the current period are matched with the locally stored historical trajectory database that is dynamically built and updated in real time as the device runs. Based on the sliding time window, persistent targets that meet the physical kinematic constraints are selected and transient noise interference is filtered out. If the current period is the first time the UAV target is detected and there is no matching historical trajectory data, a new trajectory to be confirmed is created for the target and it enters the continuous observation state.

[0007] Step S4: Dynamic threat assessment step. For persistent targets that have been successfully associated, their multidimensional feature parameters are input into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a lightweight decision tree structure that has been pre-trained and finalized in terms of overall structure and core decision rules. Its node splitting judgment threshold is dynamically adjusted by quantifying the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of photoelectric image through a preset function relationship to suppress false alarms.

[0008] Step S5: Interference strategy invocation step. If the real-time threat level exceeds the preset intervention threshold, the corresponding interference waveform parameters are invoked from the preset interference strategy library according to the target's current master control feature parameters. The master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters.

[0009] Step S6: The interference execution control step is to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate a beam pointing servo control command and a waveform generator excitation signal to drive the interference equipment to perform directional tracking and interference on the target.

[0010] Furthermore, the association matching based on the sliding time window in step S3 specifically includes:

[0011] Step S31: Construct a sliding window with a length of N scan cycles, and store the multidimensional feature parameters of all detection points within the window as a temporary sequence;

[0012] Step S32: Use the nearest neighbor data association algorithm to associate the current period's points with existing trajectories in the historical trajectory database. The association criterion is the weighted Euclidean distance between feature parameters.

[0013] Step S33: For a new point that is not associated with any historical trajectory, create a new trajectory to be confirmed and enter the continuous observation state; if the number of points of the trajectory to be confirmed exceeds the threshold within M consecutive cycles, and its motion trajectory conforms to the preset uniform acceleration physical model, then activate it and move it into the historical trajectory database; where M is a positive integer less than N.

[0014] Furthermore, the method for dynamically adjusting the node splitting threshold of the lightweight threat assessment network in step S4 is as follows:

[0015] Real-time acquisition of the root mean square value of background noise output from the radio monitoring direction finding unit and the signal-to-noise ratio of the image output from the photoelectric tracking unit;

[0016] By substituting the root mean square value of background noise and the image signal-to-noise ratio into a pre-calibrated functional relationship, a threat determination threshold for the current environment is dynamically generated.

[0017] The signal-to-noise ratio (SNR) of the target signal is compared with a dynamically generated threat assessment threshold. If the SNR of the target is higher than the threat assessment threshold, it is determined to be a potential threat and allowed to proceed to the subsequent decision-making process; otherwise, it is classified as environmental noise or a false alarm and filtered out.

[0018] Furthermore, the pre-defined functional relationship is specifically as follows:

[0019] The dynamically generated threat determination threshold is obtained by adding the baseline threshold to the first product term and then subtracting the second product term. The first product term is the product of the noise boosting coefficient and the root mean square value of the background noise, and the second product term is the product of the image confirmation weighting coefficient and the image signal-to-noise ratio. This functional relationship is expressed by the following formula:

[0020] ;

[0021] in, Threat determination thresholds are dynamically generated. η is the baseline threshold, α is the noise boosting coefficient, η is the root mean square value of background noise, β is the image confirmation weighting coefficient, and ρ is the image signal-to-noise ratio.

[0022] Furthermore, after calling the corresponding interference waveform parameters in step S5, a co-frequency interference avoidance step is also included:

[0023] Read the operating frequencies of other communication or detection tasks currently being performed by this vehicle;

[0024] The center frequency of the interference signal to be selected is compared with the operating frequency of other tasks of this vehicle, and the absolute value of the difference between the two is calculated.

[0025] If the absolute value is less than the preset protection bandwidth, the frequency avoidance mechanism is triggered. In the interference strategy library, a backup interference waveform of the same type as the target but with a different frequency is selected, or the interference signal is actively notched through modulation to protect the normal operation of the vehicle's electronic equipment.

[0026] Furthermore, following step S6, an online evaluation of the interference effect and a strategy fine-tuning step are also included:

[0027] Step S61: After the interference signal is emitted, continuously monitor the changes in the downlink signal strength of the target UAV and the offset of the motion trajectory in the photoelectric image;

[0028] Step S62: Construct an effectiveness evaluation function. This function calculates the effectiveness value of the current interference strategy by weighted summation of the signal strength attenuation rate and the trajectory deviation angular rate. The specific calculation process is as follows: First, set a weight factor, which is set according to whether the target is mainly image transmission link interference or navigation link interference; then multiply the signal strength attenuation rate by the weight factor to obtain the first weighted value; then multiply the trajectory deviation angular rate by a factor complementary to the weight factor to obtain the second weighted value; finally, add the first weighted value and the second weighted value to obtain the effectiveness value. Specifically:

[0029] ;

[0030] Where Q is the performance value and γ is the weighting factor. The attenuation rate of the signal strength. The angular rate of trajectory deviation;

[0031] Step S63: If the performance value of multiple consecutive evaluation cycles is lower than the preset performance threshold under the current interference parameters, the current interference strategy is determined to be ineffective, and the system will automatically switch to the next candidate strategy of the same type in the interference strategy library until the performance value recovers.

[0032] Furthermore, in step S63, the process switches to the next candidate policy of the same type, specifically employing an ε-greedy policy based on reinforcement learning for exploration and utilization:

[0033] The target signal strength and motion trajectory parameters monitored at the current moment are taken as the current state s;

[0034] Initialize an action-value table for this target model, recording the cumulative performance value under different state-action pairs;

[0035] In the current state s, set an exploration probability ε; select the interference action with the highest value in the current action-value table with probability 1-ε, and randomly select other interference actions with probability ε.

[0036] Based on the performance values ​​fed back after the actions are performed, the action-value table is updated online to achieve adaptive evolution of the strategy.

[0037] Furthermore, when the vehicle platform detects multiple incoming targets, the method also includes an interference resource scheduling step based on spatiotemporal conflict detection:

[0038] Extract azimuth, elevation, and threat levels of multiple targets;

[0039] The beam dwell time and mechanical rotation range of the jamming device are modeled as constraints to determine the total available time window within a scheduling cycle, denoted as ΔT.

[0040] By solving a 0-1 integer programming problem, the illumination order and dwell time allocation of the interfering beams are determined within a total time window ΔT, maximizing the sum of the weighted threat levels of the intercepted targets. Specifically, a decision variable xi represents whether target i is illuminated in this round of scheduling, with xi taking the value 0 or 1. An objective function is established to maximize the sum of the threat levels of all illuminated targets, while constraints are set such that the sum of the dwell time required by all illuminated targets and the beam adjustment stabilization time does not exceed the total time window ΔT. The mathematical model of the 0-1 integer programming problem is expressed as follows:

[0041] ;

[0042] ;

[0043] Where xi is a decision variable, indicating whether target i is illuminated in this round of scheduling. When xi=1, it means that target i is selected for illumination, and when xi=0, it means that it is not selected.

[0044] Ti represents the threat level of target i;

[0045] Adjust the stabilization time for beam pointing; The dwell time for interference targeting target i; The total time window within the scheduling period; n is the total number of targets;

[0046] Based on the calculation results of 0-1 integer programming, the optimal multi-target interference timing instruction is output.

[0047] The vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment strategy decision generation system includes:

[0048] A multimodal sensor array, deployed on an onboard platform, includes at least a radio monitoring and direction finding unit and an optoelectronic tracking unit, used to continuously acquire raw perception data of the target airspace while the vehicle is in motion or parked.

[0049] The data preprocessing module, connected to the multimodal sensor group, is used to preprocess the raw sensing data, separate the target signals suspected to be UAVs, and extract multidimensional feature parameters for each target signal; the multidimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image;

[0050] The association matching module, connected to the data preprocessing module, is used to perform association matching between the multi-dimensional feature parameters extracted in the current period and the locally stored historical trajectory database based on a sliding time window, to filter out persistent targets that meet physical kinematic constraints and to filter out transient noise interference.

[0051] The threat assessment module, connected to the association matching module, is used to input the multi-dimensional feature parameters of successfully associated persistent targets into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a decision tree structure, and its node splitting threshold is dynamically adjusted based on the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of the photoelectric image to suppress false alarms.

[0052] The strategy generation module, connected to the threat assessment module, is used to call the corresponding interference waveform parameters from the preset interference strategy library based on the target's current master control feature parameters when the real-time threat level exceeds the preset intervention threshold; where the master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters.

[0053] The interference control module, connected to the strategy generation module, is used to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate servo control commands for beam pointing and excitation signals for the waveform generator, thereby driving the interference equipment to perform directional tracking and interference on the target.

[0054] Compared with existing technologies, this invention has the following advantages: The vehicle-mounted UAV interference detection equipment strategy decision generation method and system continuously acquires raw perception data of the target airspace through a vehicle-mounted multimodal sensor group. Combined with sliding time window correlation matching and nearest neighbor data correlation algorithms, it can accurately filter persistent UAV targets that conform to physical kinematic constraints, effectively filter transient noise interference, and improve the accuracy of target detection. Relying on a lightweight threat assessment network using a decision tree structure, and dynamically adjusting the node splitting threshold based on the root mean square value of background noise and the signal-to-noise ratio of the photoelectric image, it can accurately calculate the real-time threat level of the target, effectively suppressing false alarms and making threat assessment more suitable for the complex vehicle environment. By determining the main control feature parameters through principal component analysis and calling the corresponding interference waveform parameters from the interference strategy library, while adding a co-frequency mutual interference avoidance step, it can protect the normal operation of the vehicle's electronic equipment while accurately matching the interference strategy, avoiding self-interference problems. After the interference is implemented, it constructs... The performance evaluation function enables online assessment of interference effects, and combined with an ε-greedy strategy based on reinforcement learning, it adaptively fine-tunes and evolves interference strategies. This allows for timely detection of failed strategies and rapid switching, ensuring the continuity and effectiveness of interference. For multi-target attack scenarios, a 0-1 integer programming model is constructed for interference resource scheduling based on spatiotemporal conflict detection. This achieves optimal allocation of interference beam illumination order and dwell time, maximizing the sum of weighted threat levels of intercepted targets and improving resource utilization efficiency and interception effectiveness for multi-target interference. The overall system integrates perception, analysis, evaluation, decision-making, and control, adapting to different vehicle states, including driving and parking. Through the collaborative work of various modules, the strategy decision-making for vehicle-mounted UAV interference detection becomes more intelligent, accurate, and efficient. Furthermore, the lightweight network structure and fast algorithm processing meet the real-time requirements of vehicle platforms, significantly improving the ability of vehicle platforms to cope with UAV threats, making this system more worthy of widespread adoption. Attached Figure Description

[0055] Figure 1 This is the overall flowchart of the present invention. Detailed Implementation

[0056] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0057] like Figure 1 As shown, this embodiment provides a technical solution: a strategy decision generation method for vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment, comprising the following steps:

[0058] Step S1: Multimodal data acquisition step, in the vehicle driving or parked state, continuously acquire raw perception data of the target airspace through the vehicle-mounted multimodal sensor group; the vehicle-mounted multimodal sensor group includes at least a radio monitoring direction finding unit and an optoelectronic tracking unit;

[0059] Step S2: Feature parameter extraction step, preprocessing the raw sensing data, separating the target signals suspected to be UAVs, and extracting multi-dimensional feature parameters for each target signal; the multi-dimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image;

[0060] Step S3: Target association and matching step. The multidimensional feature parameters extracted in the current period are matched with the locally stored historical trajectory database that is dynamically built and updated in real time as the device runs. Based on the sliding time window, persistent targets that meet the physical kinematic constraints are selected and transient noise interference is filtered out. If the current period is the first time the UAV target is detected and there is no matching historical trajectory data, a new trajectory to be confirmed is created for the target and it enters the continuous observation state.

[0061] Step S4: Dynamic threat assessment step. For persistent targets that have been successfully associated, their multidimensional feature parameters are input into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a lightweight decision tree structure that has been pre-trained and finalized in terms of overall structure and core decision rules. Its node splitting judgment threshold is dynamically adjusted by quantifying the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of photoelectric image through a preset function relationship to suppress false alarms.

[0062] Step S5: Interference strategy invocation step. If the real-time threat level exceeds the preset intervention threshold, the corresponding interference waveform parameters are invoked from the preset interference strategy library according to the target's current master control feature parameters. The master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters.

[0063] Step S6: The interference execution control step is to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate a beam pointing servo control command and a waveform generator excitation signal to drive the interference equipment to perform directional tracking and interference on the target.

[0064] Step S3, the association matching based on the sliding time window, specifically includes:

[0065] Step S31: Construct a sliding window with a length of N scan cycles, and store the multidimensional feature parameters of all detection points within the window as a temporary sequence;

[0066] Step S32: Use the nearest neighbor data association algorithm to perform an association judgment on the tracks in the current cycle and the existing tracks in the historical track library. The association criterion is the weighted Euclidean distance between the feature parameters.

[0067] Step S33: For the new tracks that are not associated with any historical tracks, create a new unconfirmed track for them and enter the continuous observation state. If within consecutive M cycles, the number of tracks of this unconfirmed track exceeds the threshold and its movement track conforms to the preset uniformly accelerated physical model, then activate it and move it into the historical track database; where M is a positive integer less than N.

[0068] By constructing a sliding time window of N scanning cycles to store the temporary sequence of multi-dimensional feature parameters of the detected tracks, using the weighted Euclidean distance as the criterion and the nearest neighbor data association algorithm to achieve the precise association of the current track and the historical track. At the same time, create unconfirmed tracks for the unassociated new tracks and verify them through continuous observation for M cycles. Only activate and move the unconfirmed tracks whose track quantity exceeds the threshold and whose movement tracks conform to the uniformly accelerated physical model into the historical track library. The setting of M less than N takes into account the timeliness of new target recognition. This method not only improves the objectivity and accuracy of the association between tracks and tracks through a quantified association criterion, reduces the problem of track mismatch, but also strictly filters the interference caused by transient noise and false tracks through a hierarchical track verification mechanism, accurately screens out the persistent targets that conform to the physical kinematic constraints, and can also activate new effective target tracks in a timely manner while ensuring the accuracy of target detection, providing a continuous, reliable, and effective target track data basis for subsequent threat assessment and interference decision-making, avoiding the interference of invalid data on the subsequent decision-making process, and improving the accuracy and real-time performance of the overall decision-making.

[0069] The target of the UAV is marked as the target that invades the control area. When the vehicle-mounted platform performs the UAV detection and interference task in the key security authorization control area authorized by law, if the sliding window length N = 8 scanning cycles and M = 4 scanning cycles (M < N) are set, the track quantity threshold of the unconfirmed track is 3, and the allowable range of the acceleration fitting error of the preset uniformly accelerated physical model is ±0.2 m / s²; the signal arrival angle, signal strength, and carrier frequency are set as the core feature parameters for association matching, and their weights are w1 = 0.5, w2 = 0.3, w3 = 0.2 respectively, and the weighted Euclidean distance association threshold is 0.15. If it is less than this threshold, it is determined that the track and the historical track are associated successfully.

[0070] Creation and verification of the track of the new track:

[0071] In the first scanning cycle, a new track P1 (signal arrival angle 25°, signal strength -55 dBm, carrier frequency 5.8 GHz) that invades the control area is detected, and there is no matching historical track. Create an unconfirmed track T1 for it.

[0072] In the second scan cycle, a new trace P2 (signal angle of arrival 26°, signal strength -54dBm, carrier frequency 5.8GHz) was detected intruding into the controlled area and associated with T1. At this time, the number of traces in T1 is 2.

[0073] In the third scan cycle, a new trace P3 (signal angle of arrival 27°, signal strength -53dBm, carrier frequency 5.8GHz) was detected intruding into the controlled area and associated with T1. At this time, the number of traces in T1 is 3, reaching the preset threshold.

[0074] In the fourth scan cycle, a new point P4 (signal angle of arrival 28°, signal strength -52dBm, carrier frequency 5.8GHz) was detected intruding into the control zone. It was associated with T1, and observations were completed for M=4 consecutive cycles. The motion position parameters of P1-P4 were fitted, and a uniform acceleration a=0.6m / s² was obtained. Within the allowable error range, it conforms to the uniform acceleration physical model. At this time, the trajectory T1 to be confirmed was activated and officially moved into the historical trajectory database.

[0075] If a new trace P5 that intrudes into the controlled area is detected only in the first and second cycles, and there is no corresponding trace in the third and fourth cycles, and the number of traces in four consecutive cycles is 2, which does not reach the threshold 3, then the trace is determined to be transient noise interference and is directly removed without creating a valid trajectory.

[0076] Association and matching between current point trace and historical trajectory:

[0077] An activated trajectory T0 exists in the historical trajectory database. Its latest period's feature parameters, after normalization, are x10=0.3, x20=0.4, and x30=0.5.

[0078] The point P that intruded into the controlled area was detected in the previous scan cycle. Its feature parameters, after normalization, are x1=0.32, x2=0.41, and x3=0.51.

[0079] Calculate the association criterion using the weighted Euclidean distance formula:

[0080] ;

[0081] Substitute the values:

[0082] ;

[0083] ;

[0084] ;

[0085] The calculation result 0.0158 < 0.15, which meets the association threshold requirement. It is determined that the point P is successfully associated with the historical trajectory T0. P is added to the feature parameter sequence of T0 to realize continuous trajectory tracking of the target corresponding to T0.

[0086] If the normalized parameters of the current point P6 are x1=0.5, x2=0.6, and x3=0.7, and the calculated value of d≈0.2179>0.15 according to the above formula, then it is determined that there is no associated historical trajectory, and a new trajectory to be confirmed is created for it.

[0087] The method for dynamically adjusting the node splitting threshold in the lightweight threat assessment network in step S4 is as follows:

[0088] Real-time acquisition of the root mean square value of background noise output from the radio monitoring direction finding unit and the signal-to-noise ratio of the image output from the photoelectric tracking unit;

[0089] By substituting the root mean square value of background noise and the image signal-to-noise ratio into a pre-calibrated functional relationship, a threat determination threshold for the current environment is dynamically generated.

[0090] The signal-to-noise ratio (SNR) of the target signal is compared with a dynamically generated threat assessment threshold. If the SNR of the target is higher than the threat assessment threshold, it is determined to be a potential threat and allowed to proceed to the subsequent decision-making process; otherwise, it is classified as environmental noise or a false alarm and filtered out.

[0091] By real-time acquisition of the background noise root mean square value of the radio monitoring direction finding unit and the image signal-to-noise ratio of the photoelectric tracking unit, and substituting them into a pre-calibrated functional relationship, a threat determination threshold is dynamically generated. The target signal signal-to-noise ratio is then compared with this dynamic threshold to complete the threat determination. This method abandons the fixed threshold determination method and allows the threat determination threshold to adaptively adjust with the real-time environmental changes of the vehicle platform. It effectively adapts to the complex environment of background noise fluctuations and image quality changes in vehicle scenarios, accurately distinguishes potential threat targets from environmental noise and false alarm signals, and significantly reduces the false alarm rate. At the same time, only potential threat targets that meet the threshold requirements are allowed to enter the subsequent decision-making process, reducing the amount of invalid data processing and improving the efficiency and accuracy of the overall strategy decision-making. This provides a reliable threat determination basis for the generation of subsequent interference strategies.

[0092] Set baseline threshold dB, noise rise factor α=0.8, image confirmation weighting factor β=0.5, dynamic threat determination threshold formula is: The judgment rule is as follows: when the target signal signal-to-noise ratio (SNR) > τ, it is judged as a potential threat and enters the subsequent decision-making process; when the target signal signal-to-noise ratio (SNR) ≤ τ, it is classified as environmental noise or false alarm and filtered out.

[0093] Add specific control zone restrictions for each scenario, and mark suspected targets as suspected drone targets that have intruded into the control zone. Scenario for judgment in complex electromagnetic environments:

[0094] The vehicle-mounted radio monitoring and direction-finding unit acquires the current root mean square value of background noise η=10. Due to the influence of environmental haze, the photoelectric tracking unit outputs an image with a signal-to-noise ratio ρ=8. Substituting these values ​​into the formula, the dynamic threshold is calculated as follows:

[0095] τ = 15 + 0.8 × 10 - 0.5 × 8;

[0096] τ = 15 + 8 - 4 = 19;

[0097] If a signal with a signal-to-noise ratio (SNR) of 18 is detected as a suspected drone target intruding into the controlled area, since 18 ≤ 19, the signal is determined to be environmental noise or a false alarm and is directly filtered out to avoid false alarms caused by complex electromagnetic noise.

[0098] Judgment scenarios in a clear environment:

[0099] The vehicle-mounted equipment collects the current root mean square value of background noise η=5. Due to the clear environment and sharp target imaging, the photoelectric tracking unit outputs an image with a signal-to-noise ratio ρ=20. Substituting these values ​​into the formula, the dynamic threshold is calculated as follows:

[0100] τ = 15 + 0.8 × 5 - 0.5 × 20;

[0101] τ = 15 + 4 - 10 = 9;

[0102] If a signal with a signal-to-noise ratio (SNR) of 18 is detected as a suspected drone of the same model that has intruded into the controlled area, since 18 > 9, the signal is determined to be a potential threat target, and the process is allowed to proceed to the subsequent jamming strategy decision-making process.

[0103] Scenarios for determining moderate noise and moderate image quality:

[0104] The vehicle-mounted equipment collected a background noise root mean square value η=7 and an image signal-to-noise ratio ρ=12. Substituting these values ​​into the formula, the dynamic threshold was calculated.

[0105] τ = 15 + 0.8 × 7 - 0.5 × 12;

[0106] τ = 15 + 5.6 - 6 = 14.6;

[0107] If a suspected target signal with a signal-to-noise ratio (SNR) of 15 is detected intruding into the controlled area, it is judged as a potential threat because 15 > 14.6; if the target signal with a SNR of 14 is detected, it is judged as environmental noise / false alarm and filtered out because 14 ≤ 14.6.

[0108] The predefined functional relationship is as follows:

[0109] The dynamically generated threat determination threshold is obtained by adding the baseline threshold to the first product term and then subtracting the second product term. The first product term is the product of the noise boosting coefficient and the root mean square value of the background noise, and the second product term is the product of the image confirmation weighting coefficient and the image signal-to-noise ratio. This functional relationship is expressed by the following formula:

[0110] ;

[0111] in, Threat determination thresholds are dynamically generated. The baseline threshold is α, the noise boosting coefficient is η, the root mean square value of background noise is η, the image confirmation weighting coefficient is β, and the image signal-to-noise ratio is ρ.

[0112] By clarifying the specific functional relationship between the product of the baseline threshold and the noise rise coefficient and the root mean square value of the background noise, and then subtracting the product of the image confirmation weighting coefficient and the image signal-to-noise ratio, a standardized and quantifiable calculation method is provided for generating dynamic threat assessment thresholds. This allows threshold adjustment to move away from vague empirical judgments and instead be based on precise mathematical calculations. It enables accurate and reproducible threshold adjustments based on the actual values ​​of background noise and image signal-to-noise ratio. Furthermore, the formula quantifies and couples the positive correlation of noise rise and the negative correlation of image confirmation, allowing dynamic thresholds to better fit the complex electromagnetic and visual environment of vehicles, accurately reflecting the actual environmental conditions for target threat assessment. This further improves the objectivity and accuracy of threat assessment, effectively suppresses false alarms and avoids missing potential threats, providing a concrete and implementable solution for adjusting node splitting thresholds in lightweight threat assessment networks.

[0113] For example, in authorized control zones, a unified baseline threshold should be set. dB, noise rise factor α=0.6, image confirmation weighting factor β=0.4, and the dynamic threat determination threshold is calculated using a fixed standardized formula: The judgment rule is that when the target signal signal-to-noise ratio (SNR) > τ, it is judged as a potential threat and enters the subsequent process; when SNR ≤ τ, it is classified as environmental noise or false alarm and filtered out. Threshold calculation and threat judgment are performed for different vehicle environment scenarios.

[0114] High-noise, low-signal-to-noise ratio scenes on urban roads:

[0115] The vehicle-mounted platform performs tasks within the authorized control area. The equipment collects the root mean square value of background noise η=15 in real time. Due to urban building obstruction and complex lighting, the signal-to-noise ratio of the photoelectric image is ρ=10. Substituting these values ​​into the formula, the dynamic threshold is calculated as follows:

[0116] τ = 20 + 0.6 × 15 − 0.4 × 10;

[0117] τ = 20 + 9 - 4 = 25;

[0118] If a suspected drone target is detected with a signal-to-noise ratio (SNR) of 24, it is determined to be environmental noise / false alarm and filtered out because 24 ≤ 25; if the target's SNR is 26, it is determined to be a potential threat because 26 > 25.

[0119] Low-noise, high image signal-to-noise ratio scene on suburban roads:

[0120] The vehicle-mounted platform performs tasks within the authorized control area. The equipment collects background noise with a root mean square (RMS) of η=4. The suburban environment is open and the imaging is clear, with a photoelectric image signal-to-noise ratio (SNR) of ρ=25. Substituting these values ​​into the formula, the dynamic threshold is calculated as follows:

[0121] τ = 20 + 0.6 × 4 − 0.4 × 25;

[0122] τ = 20 + 2.4 − 10 = 12.4;

[0123] If a suspected target with a signal-to-noise ratio (SNR) of 13 is detected, it is considered a potential threat because 13 > 12.4; if the target with a SNR of 12 is detected, it is considered environmental noise / false alarm because 12 ≤ 12.4.

[0124] Suburban scenes with moderate noise and moderate image signal-to-noise ratio:

[0125] The vehicle-mounted platform performs tasks within the authorized control area. The equipment collects background noise root mean square value η=9 and photoelectric image signal-to-noise ratio ρ=18. Substitute these values ​​into the formula to calculate the dynamic threshold:

[0126] τ = 20 + 0.6 × 9 − 0.4 × 18;

[0127] τ = 20 + 5.4 - 7.2 = 18.2;

[0128] If a suspected target with a signal-to-noise ratio (SNR) of 19 is detected, it is considered a potential threat because 19 > 18.2; if the target with a SNR of 18 is detected, it is considered environmental noise / false alarm because 18 ≤ 18.2.

[0129] Scenes where noise suddenly increases and images become slightly blurry while driving at high speed:

[0130] The vehicle-mounted platform is maneuvering within the authorized control area to perform tasks. The electromagnetic noise of the vehicle body suddenly increases, the root mean square value of the background noise η=20, and vehicle vibration causes slight image blurring. The photoelectric image signal-to-noise ratio ρ=16. Substituting these values ​​into the formula, the dynamic threshold is calculated as follows:

[0131] τ = 20 + 0.6 × 20 − 0.4 × 16;

[0132] τ = 20 + 12 − 6.4 = 25.6;

[0133] If a suspected target is detected with a signal-to-noise ratio (SNR) of 25, it is determined to be environmental noise / false alarm because 25 ≤ 25.6; if the target's SNR is 26, it is determined to be a potential threat because 26 > 25.6.

[0134] After calling the corresponding interference waveform parameters in step S5, a co-frequency interference avoidance step is also included:

[0135] Read the operating frequencies of other communication or detection tasks currently being performed by this vehicle;

[0136] The center frequency of the interference signal to be selected is compared with the operating frequency of other tasks of this vehicle, and the absolute value of the difference between the two is calculated.

[0137] If the absolute value is less than the preset protection bandwidth, the frequency avoidance mechanism is triggered. In the interference strategy library, a backup interference waveform of the same type as the target but with a different frequency is selected, or the interference signal is actively notched by modulation to protect the normal operation of the vehicle's electronic equipment.

[0138] After calling the interference waveform parameters, a co-frequency interference avoidance step is added. By reading the operating frequency of other communication and detection tasks of the vehicle and comparing the absolute value of the difference between the center frequency of the interference signal and that frequency, the frequency avoidance mechanism is triggered when the difference is less than the protection bandwidth. The backup interference waveform or active notch filtering is selected, which can effectively avoid co-frequency or adjacent-frequency interference between the interference signal and the operating frequency of the vehicle's electronic equipment. This prevents the vehicle's communication, detection and other equipment from malfunctioning due to self-interference, ensuring the normal operation of various electronic systems in the vehicle. At the same time, it can still effectively interfere with the UAV target while avoiding interference, taking into account both the interference effect on the UAV and the operational safety of the vehicle platform's own equipment. This makes the operation of the vehicle-mounted UAV interference detection equipment more compatible and reliable, and adapts to the scenario requirements of multiple electronic devices working together on the vehicle platform.

[0139] Authorized control area to perform tasks:

[0140] When the vehicle-mounted platform performs drone detection and interference tasks within the authorized control area, a preset protection bandwidth of 20MHz is uniformly set, and the frequency comparison judgment rule is: if This triggers a frequency avoidance mechanism.

[0141] like It directly uses the selected interference waveform parameters, and presets the center frequency of the primary interference waveform for the same UAV target intruding into the control area from the interference strategy library. The center frequency of the backup interference waveform All of them are valid interference waveforms of the same type.

[0142] No co-frequency interference scenarios:

[0143] The vehicle-mounted equipment determined that the drone target intruding into the controlled area was using the 2.4GHz band for jamming, and selected the center frequency of the primary jamming waveform. The operating frequencies for tasks such as current satellite positioning and vehicle-to-everything communication of this vehicle are as follows: Calculate the absolute value of the frequency difference respectively:

[0144] ;

[0145] ;

[0146] Both calculation results are greater than the 20MHz protection bandwidth, indicating no risk of co-channel interference; therefore, they can be used directly. The main interference waveform parameters are used to interfere with the target.

[0147] Trigger frequency avoidance - select alternative waveform scenarios:

[0148] The vehicle-mounted equipment determines that the same UAV target intruding into the controlled area needs to be jammed using the 5.8GHz band, and selects the center frequency of the primary jamming waveform. Read the operating frequency of the vehicle's current millimeter-wave radar detection mission. Calculate the absolute value of the frequency difference:

[0149] ;

[0150] A protection bandwidth of 10MHz less than 20MHz triggers a frequency avoidance mechanism, retrieving a backup interference waveform of the same type as the target from the interference strategy library and selecting it. Calculate again:

[0151] ;

[0152] No risk of mutual interference, use The backup interference waveform parameters are used to implement interference.

[0153] Trigger frequency avoidance - active notch shot scenario:

[0154] The vehicle-mounted equipment determines that the same UAV target intruding into the controlled area needs to be jammed using the 1.2GHz band, and selects the center frequency of the primary jamming waveform. Read the operating frequency of the current shortwave communication task of this vehicle. Calculate the absolute value of the frequency difference:

[0155] ;

[0156] If the frequency avoidance mechanism is triggered and the target has no suitable backup interference waveform, then... Active notch filtering is performed on the main interference signal. A notch band is set at the 1215MHz frequency point, and the notch bandwidth covers 1210MHz-1220MHz, avoiding the operating frequency of the vehicle's shortwave communication. After processing, the notched interference waveform is used to interfere with the UAV target, which ensures the interference effect while avoiding the impact on the vehicle's communication equipment.

[0157] Following step S6, there is also an online evaluation of the interference effect and a strategy fine-tuning step:

[0158] Step S61: After the interference signal is emitted, continuously monitor the changes in the downlink signal strength of the target UAV and the offset of the motion trajectory in the photoelectric image;

[0159] Step S62: Construct an effectiveness evaluation function. This function calculates the effectiveness value of the current interference strategy by weighted summation of the signal strength attenuation rate and the trajectory deviation angular rate. The specific calculation process is as follows: First, set a weight factor, which is set according to whether the target is mainly image transmission link interference or navigation link interference; then multiply the signal strength attenuation rate by the weight factor to obtain the first weighted value; then multiply the trajectory deviation angular rate by a factor complementary to the weight factor to obtain the second weighted value; finally, add the first weighted value and the second weighted value to obtain the effectiveness value. Specifically:

[0160] ;

[0161] Where Q is the performance value and γ is the weighting factor. The attenuation rate of the signal strength. The angular rate of trajectory deviation;

[0162] Step S63: If the performance value of multiple consecutive evaluation cycles is lower than the preset performance threshold under the current interference parameters, the current interference strategy is determined to be ineffective, and the next candidate strategy of the same type is automatically switched in the interference strategy library until the performance value recovers.

[0163] After the jamming is implemented, an online evaluation and strategy fine-tuning step for the jamming effect is added. By continuously monitoring the downlink signal strength attenuation rate and trajectory deviation angular rate of the UAV, a performance evaluation function with weighted factors is constructed to quantify the jamming effectiveness. When the effectiveness value is continuously lower than the threshold, the candidate strategy is automatically switched. This allows for real-time quantitative judgment of the effectiveness of the current jamming strategy, timely detection of ineffective jamming schemes, and rapid strategy adjustment, avoiding the continuous consumption of vehicle resources by ineffective jamming. At the same time, the weighted factor can be flexibly set according to the type of jamming link, making the performance evaluation more in line with the actual jamming needs, ensuring that the jamming effect on UAV targets is always at an effective level, realizing dynamic optimization of the jamming strategy, and improving the adaptability and jamming success rate of vehicle-mounted jamming equipment to different UAV targets and different environments.

[0164] When the vehicle-mounted platform performs tasks within the authorized control area, the performance evaluation function is uniformly set as follows: Where γ is a weighting factor, set to 0.7 for interference with the image transmission link and 0.3 for interference with the navigation link. The preset performance threshold is 0.6. If the performance value Q < 0.6 for three consecutive evaluation cycles, the current interference strategy is deemed invalid and a candidate strategy is switched. Signal strength attenuation rate. angular rate of trajectory deviation All values ​​are normalized and range from 0 to 1. The larger the value, the better the interference effect.

[0165] For scenarios involving interference evaluation and strategy maintenance of image transmission links:

[0166] Image transmission link interference was implemented against the same UAV target that intruded into the controlled area. With γ=0.7, it was detected during the first evaluation cycle. Substitute into the formula to calculate the performance value:

[0167] Q=0.7×0.8+(1-0.7)×0.5=0.56+0.15=0.71;

[0168] The Q values ​​calculated during the second and third monitoring cycles were 0.73 and 0.70, respectively, both greater than the effectiveness threshold of 0.6. Therefore, the current interference strategy was deemed effective, and the original strategy was maintained to continue the interference.

[0169] Interference effect assessment and strategy maintenance scenarios for navigation links

[0170] Navigation link interference was implemented against the same UAV target that intruded into the controlled area. With γ=0.3, the parameters monitored over three consecutive evaluation periods were as follows: Period 1 Period 2 , cycle 3 .

[0171] Calculation of performance value for cycle 1:

[0172] Q=0.3×0.4+(1-0.3)×0.9=0.12+0.63=0.75;

[0173] Calculation of performance value for cycle 2:

[0174] Q=0.3×0.42+(1-0.3)×0.88=0.126+0.616=0.742;

[0175] Calculation of performance value for cycle 3:

[0176] Q=0.3×0.39+(1-0.3)×0.91=0.117+0.637=0.754;

[0177] The results of all three calculations are greater than 0.6, indicating that the current interference strategy is effective, and the original strategy will continue to be executed.

[0178] Image transmission link interference strategy failure and switching scenarios:

[0179] Image transmission link interference was implemented against the same UAV target that intruded into the controlled area. With γ=0.7, it was detected for three consecutive evaluation cycles. Calculate the performance value sequentially:

[0180] Period 1: Q = 0.7 × 0.4 + (1 - 0.7) × 0.3 = 0.28 + 0.09 = 0.37;

[0181] Period 2: Q = 0.7 × 0.4 + (1 - 0.7) × 0.3 = 0.37;

[0182] Period 3: Q = 0.7 × 0.4 + (1 - 0.7) × 0.3 = 0.37;

[0183] If the effectiveness value of 0.37 is less than the threshold of 0.6 for three consecutive cycles, the current interference strategy is determined to be ineffective. The system automatically switches to the next candidate interference strategy of the same type from the interference strategy library. The first evaluation cycle after the switch detects... Calculate the performance value:

[0184] Q=0.7×0.85+(1-0.7)×0.6=0.595+0.18=0.775>0.6;

[0185] If the new strategy is deemed effective, continue to use the new strategy to interfere.

[0186] Navigation link interference strategy failure and switching scenarios:

[0187] Navigation link interference was implemented against the same UAV target that intruded into the controlled area. With γ=0.3, it was detected for three consecutive evaluation cycles. Calculate the performance value sequentially:

[0188] Period 1: Q = 0.3 × 0.2 + (1 - 0.3) × 0.4 = 0.06 + 0.28 = 0.34;

[0189] Period 2: Q = 0.3 × 0.2 + (1 - 0.3) × 0.4 = 0.34;

[0190] Period 3: Q = 0.3 × 0.2 + (1 - 0.3) × 0.4 = 0.34;

[0191] If the performance value is 0.34 for three consecutive cycles, which is less than the threshold of 0.6, the current strategy is deemed ineffective, and a candidate strategy of the same type is switched. After the switch, it is detected that... Calculate the performance value:

[0192] Q=0.3×0.5+(1-0.3)×0.92=0.15+0.644=0.794>0.6;

[0193] If the new strategy is deemed effective, it will be implemented to disrupt the system.

[0194] In step S63, the system switches to the next candidate policy of the same type, specifically using an ε-greedy policy based on reinforcement learning for exploration and utilization:

[0195] The target signal strength and motion trajectory parameters monitored at the current moment are taken as the current state s;

[0196] Initialize an action-value table for this target model, recording the cumulative performance value under different state-action pairs;

[0197] In the current state s, set an exploration probability ε; select the interference action with the highest value in the current action-value table with probability 1-ε, and randomly select other interference actions with probability ε.

[0198] Based on the performance value feedback after the action is performed, the action-value table is updated online to achieve adaptive evolution of the strategy;

[0199] By employing an ε-greedy strategy based on reinforcement learning for candidate switching of failed jamming strategies, the target signal strength and motion trajectory parameters are used as the current state. An action-value table records the cumulative effectiveness value of different state-action pairs. By combining the exploration of probability ε to balance the utilization and exploration of jamming strategies, the system selects the current optimal jamming action with a high probability of 1-ε to ensure the immediate effectiveness of jamming, while randomly exploring other jamming actions with a low probability of ε to discover better strategies. At the same time, the action-value table is updated online according to the effectiveness value to achieve adaptive evolution of the strategy. This allows the vehicle-mounted jamming equipment to continuously adapt to the dynamic changes of UAV targets, continuously optimize the jamming strategy library, avoid getting trapped in local optimal jamming schemes, improve the jamming adaptability and long-term jamming success rate of different types and states of UAV targets, and make the adjustment of jamming strategies more intelligent and adaptive.

[0200] A certain type of UAV, whose target location is intruding into the controlled area, is uniformly set to an exploration probability ε=0.1, meaning there is a 90% probability of selecting the highest-value interference action from the action-value table (1-ε), and a 10% probability of randomly exploring other interference actions. The action-value table for this type of UAV target is initialized, containing three similar interference actions A1, A2, and A3, all with an initial cumulative effectiveness value of 0. The target signal strength attenuation rate is... angular rate of trajectory deviation As the core parameter of state s, the efficiency value update rule is as follows: (Q is the actual performance value after the current action is executed, and 0.2 is the update coefficient). The preset performance threshold is 0.6. If Q < 0.6 for three consecutive cycles, a strategy switch is triggered. The performance evaluation function is: For interference with the image transmission link, set γ=0.7.

[0201] Initial Phase: Optimal Action Selection and Value Table Update

[0202] State s1: The current action-value table has V(A1)=0, V(A2)=0, and V(A3)=0. Since ε=0.1, action A1 is randomly selected with a 90% probability for execution. After execution, the actual performance value Q(A1) is measured to be 0.35 < 0.6, triggering a strategy switch. The action-value table is then updated according to the rules.

[0203] (A1) = 0 + 0.2 × 0.35 = 0.07, V(A2) and V(A3) are still 0.

[0204] Exploration Phase: Randomly explore new actions and update the value table:

[0205] State s2 is the same as s1. The action-value table has V(A1) = 0.07, V(A2) = 0, and V(A3) = 0. The optimal action is A1. Since ε = 0.1, action A2 is randomly explored and executed with a 10% probability. After execution, the actual effectiveness value Q(A2) is measured to be 0.72 > 0.6, indicating that the strategy is effective. The action-value table is then updated.

[0206] (A2) = 0 + 0.2 × 0.72 = 0.144, V(A1) remains at 0.07, and V(A3) remains at 0.

[0207] Utilization Phase: Selecting the Optimal Action and Updating the Continuous Value Table:

[0208] State s3 is still The current action-value table has V(A1) = 0.07, V(A2) = 0.144, and V(A3) = 0. The optimal action is A2. Since ε = 0.1, A2 is selected with a 90% probability for execution. After execution, Q(A2) = 0.70 > 0.6, so the value table is updated.

[0209] (A2) = 0.144 + 0.2 × 0.70 = 0.284.

[0210] For the next five consecutive cycles, A2 was selected for execution, and the measured Q values ​​were 0.71, 0.69, 0.73, 0.70, and 0.68, respectively. After continuous updates, V(A2) = 0.284 + 0.2 × (0.71 + 0.69 + 0.73 + 0.70 + 0.68) = 0.284 + 0.2 × 3.51 = 0.986. V(A1) and V(A3) had no execution records and remained at their initial values. The action-value table completed adaptive evolution, and A2 became the stable and optimal disturbance action in this state.

[0211] State Change: Strategy Exploration and Utilization in New States

[0212] The status of a certain type of drone that intruded into the controlled area has changed; the new status is s4. The current action-value table has V(A1) = 0.07, V(A2) = 0.986, and V(A3) = 0. The optimal action is A2. A2 is executed with a 90% probability. After execution, Q(A2) = 0.55 < 0.6, triggering a strategy switch. Action A3 is then explored with a 10% probability. After execution, Q(A3) = 0.75 > 0.6, and the value table is updated.

[0213] (A3) = 0 + 0.2 × 0.75 = 0.15. At this point, A3 becomes the optimal action under state s4. Subsequently, A3 will be continuously used and the value table will be updated according to the actual performance value to achieve adaptive adaptation of the strategy to the state changes of a certain type of UAV that intrudes into the control area.

[0214] When the vehicle platform detects multiple incoming targets, the method also includes an interference resource scheduling step based on spatiotemporal conflict detection:

[0215] Extract azimuth, elevation, and threat levels of multiple targets;

[0216] The beam dwell time and mechanical rotation range of the jamming device are modeled as constraints to determine the total available time window within a scheduling cycle, denoted as ΔT.

[0217] By solving a 0-1 integer programming problem, the illumination order and dwell time allocation of the interfering beams are determined within a total time window ΔT, maximizing the sum of the weighted threat levels of the intercepted targets. Specifically, a decision variable xi represents whether target i is illuminated in this round of scheduling, with xi taking the value 0 or 1. An objective function is established to maximize the sum of the threat levels of all illuminated targets, while constraints are set such that the sum of the dwell time required by all illuminated targets and the beam adjustment stabilization time does not exceed the total time window ΔT. The mathematical model of the 0-1 integer programming problem is expressed as follows:

[0218] ;

[0219] ;

[0220] Where xi is a decision variable, indicating whether target i is illuminated in this round of scheduling. When xi=1, it means that target i is selected for illumination, and when xi=0, it means that it is not selected.

[0221] Ti represents the threat level of target i;

[0222] Adjust the stabilization time for beam pointing; The dwell time for interference targeting target i; The total time window within the scheduling period; n is the total number of targets;

[0223] Based on the calculation results of 0-1 integer programming, output the optimal multi-target interference timing command;

[0224] For multi-target attack scenarios, an interference resource scheduling step based on spatiotemporal conflict detection is added. By extracting the azimuth, elevation angle, and threat level of multiple targets, the beam characteristics of the interference equipment are modeled as constraints and a 0-1 integer programming problem is solved to achieve the optimal allocation of the interference beam illumination sequence and dwell time. With the goal of maximizing the sum of the weighted threat levels of the intercepted targets, the interference interception of high-threat-level UAV targets can be prioritized within a limited scheduling time window, avoiding the ineffective allocation of interference resources. At the same time, the interference resources are precisely planned through a quantitative mathematical model, which solves the spatiotemporal conflict problem of interference resources in multi-target scenarios, improves the resource utilization efficiency of vehicle-mounted interference equipment and the overall effect of multi-target interception, and makes interference decisions more scientific and reasonable, adapting to the complex application scenarios of vehicle-mounted platforms dealing with coordinated attacks from multiple UAVs.

[0225] When the vehicle-mounted platform performs tasks within the authorized control area, the total time window ΔT = 200ms is uniformly set within the scheduling cycle, and the beam pointing adjustment stabilization time is also set. This is a fixed value, representing the dwell time of any intrusion into the controlled area. Adapted to threat level, high threat level Medium threat level Low threat level The objective function for 0-1 integer programming is: The constraints are , where Ti is the target threat level (high threat T=3, medium threat T=2, low threat T=1), xi=1 indicates that the target is illuminated, and xi=0 indicates that the target is not illuminated.

[0226] The vehicle-mounted platform detected three drone targets intruding into the controlled area. Target 1 was a high-threat target (T1=3). Target 2 is a medium threat with T2=2. Target 3 is a low-threat target with T3=1. .

[0227] Calculate the total time required to irradiate a single target:

[0228] Objective 1: ;

[0229] Objective 2: ;

[0230] Objective 3: ;

[0231] Enumerate all combinations of objectives and verify the constraints:

[0232] Combination 1: Irradiate only target 1, x1=1, x2=0, x3=0, total time 100ms≤200ms, objective function value=3×1+2×0+1×0=3;

[0233] Combination 2: Irradiate only target 2, x1=0, x2=1, x3=0, total time 70ms≤200ms, objective function value=3×0+2×1+1×0=2;

[0234] Combination 3: Illuminate only target 3, x1=0, x2=0, x3=1, total time 50ms≤200ms, objective function value=3×0+2×0+1×1=1;

[0235] Combination 4: Irradiate target 1 + target 2, x1=1, x2=1, x3=0, total time 100+70=170ms≤200ms, objective function value=3×1+2×1+1×0=5;

[0236] Combination 5: Irradiate target 1 + target 3, x1=1, x2=0, x3=1, total time 100+50=150ms≤200ms, objective function value=3×1+2×0+1×1=4;

[0237] Combination 6: Irradiate target 2 + target 3, x1=0, x2=1, x3=1, total time 70+50=120ms≤200ms, objective function value=3×0+2×1+1×1=3;

[0238] Combination 7: Irradiate target 1 + target 2 + target 3, x1=1, x2=1, x3=1, total time 100+70+50=220ms>200ms, does not meet the constraint condition, this combination is eliminated.

[0239] Determine the optimal scheduling scheme for disruptive resources:

[0240] Comparing the objective function values ​​of all valid combinations, the objective function value of combination 4 is 5, which is the maximum value. Therefore, the optimal solution is x1=1, x2=1, x3=0, which means that in this round of scheduling, high-threat target 1 and medium-threat target 2 are prioritized for illumination, while low-threat target 3 is not illuminated.

[0241] Output interference timing instructions:

[0242] Based on the target azimuth and elevation angles, the planned beam adjustment path is from target 1 to target 2. The beam is first adjusted to the azimuth of target 1, stabilized for 20ms, and then the interference is maintained for 80ms. After that, the beam is adjusted to the azimuth of target 2, stabilized for 20ms, and then the interference is maintained for 50ms. The total time is 20+80+20+50=170ms, which is completed within the total time window of 200ms. The remaining 30ms is the redundancy time to ensure the stability of the interference execution and achieve the optimal allocation of multi-target interference resources within the control area.

[0243] The vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment strategy decision generation system includes:

[0244] A multimodal sensor array, deployed on an onboard platform, includes at least a radio monitoring and direction finding unit and an optoelectronic tracking unit, used to continuously acquire raw perception data of the target airspace while the vehicle is in motion or parked.

[0245] The data preprocessing module, connected to the multimodal sensor group, is used to preprocess the raw sensing data, separate the target signals suspected to be UAVs, and extract multidimensional feature parameters for each target signal; the multidimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image;

[0246] The association matching module, connected to the data preprocessing module, is used to perform association matching between the multi-dimensional feature parameters extracted in the current period and the locally stored historical trajectory database based on a sliding time window, to filter out persistent targets that meet physical kinematic constraints and to filter out transient noise interference.

[0247] The threat assessment module, connected to the association matching module, is used to input the multi-dimensional feature parameters of successfully associated persistent targets into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a decision tree structure, and its node splitting threshold is dynamically adjusted based on the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of the photoelectric image to suppress false alarms.

[0248] The strategy generation module, connected to the threat assessment module, is used to call the corresponding interference waveform parameters from the preset interference strategy library based on the target's current master control feature parameters when the real-time threat level exceeds the preset intervention threshold; where the master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters.

[0249] The interference control module, connected to the strategy generation module, is used to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate servo control commands for beam pointing and excitation signals for the waveform generator, thereby driving the interference equipment to perform directional tracking and interference on the target.

[0250] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0251] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0252] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for generating strategy decisions for vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment, characterized in that, Includes the following steps: Step S1: Multimodal data acquisition step, in the vehicle driving or parked state, continuously acquire raw perception data of the target airspace through the on-board multimodal sensor group; The vehicle-mounted multimodal sensor array includes at least a radio monitoring and direction-finding unit and an optoelectronic tracking unit; Step S2 The feature parameter extraction step involves preprocessing the raw sensing data, separating the target signals suspected to be drones, and extracting multi-dimensional feature parameters for each target signal. Multidimensional The characteristic parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image; Step S3: Target association and matching step. The multidimensional feature parameters extracted in the current period are matched with the locally stored historical trajectory database that is dynamically built and updated in real time as the device runs. Based on the sliding time window, persistent targets that meet the physical kinematic constraints are selected and transient noise interference is filtered out. If the current period is the first time the UAV target is detected and there is no matching historical trajectory data, a new trajectory to be confirmed is created for the target and it enters the continuous observation state. Step S4: Dynamic threat assessment step. For persistent targets that have been successfully associated, their multidimensional feature parameters are input into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a lightweight decision tree structure that has been pre-trained and finalized in terms of overall structure and core decision rules. Its node splitting judgment threshold is dynamically adjusted by quantifying the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of photoelectric image through a preset function relationship to suppress false alarms. Step S5: Interference strategy invocation step. If the real-time threat level exceeds the preset intervention threshold, the corresponding interference waveform parameters are invoked from the preset interference strategy library according to the target's current master control feature parameters. The master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters. Step S6: The interference execution control step is to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate a beam pointing servo control command and a waveform generator excitation signal to drive the interference equipment to perform directional tracking and interference on the target.

2. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 1, characterized in that: Step S3, the association matching based on the sliding time window, specifically includes: Step S31: Construct a sliding window with a length of N scan cycles, and store the multidimensional feature parameters of all detection points within the window as a temporary sequence; Step S32: Use the nearest neighbor data association algorithm to associate the current period's points with existing trajectories in the historical trajectory database. The association criterion is the weighted Euclidean distance between feature parameters. Step S33: For a new point that is not associated with any historical trajectory, create a new trajectory to be confirmed and enter the continuous observation state; if the number of points of the trajectory to be confirmed exceeds the threshold within M consecutive cycles, and its motion trajectory conforms to the preset uniform acceleration physical model, then activate it and move it into the historical trajectory database; where M is a positive integer less than N.

3. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 2, characterized in that: The method for dynamically adjusting the node splitting threshold in the lightweight threat assessment network in step S4 is as follows: Real-time acquisition of the root mean square value of background noise output from the radio monitoring direction finding unit and the signal-to-noise ratio of the image output from the photoelectric tracking unit; By substituting the root mean square value of background noise and the image signal-to-noise ratio into a pre-calibrated functional relationship, a threat determination threshold for the current environment is dynamically generated. The signal-to-noise ratio (SNR) of the target signal is compared with a dynamically generated threat assessment threshold. If the SNR of the target is higher than the threat assessment threshold, it is determined to be a potential threat and allowed to proceed to the subsequent decision-making process; otherwise, it is classified as environmental noise or a false alarm and filtered out.

4. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 3, characterized in that: The predefined functional relationship is as follows: The dynamic threat determination threshold is obtained by adding the baseline threshold to the first product term and then subtracting the second product term; wherein, the first product term is the product of the noise rise coefficient and the root mean square value of the background noise, and the second product term is the product of the image confirmation weighting coefficient and the image signal-to-noise ratio.

5. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 4, characterized in that: After calling the corresponding interference waveform parameters in step S5, a co-frequency interference avoidance step is also included: Read the operating frequencies of other communication or detection tasks currently being performed by this vehicle; The center frequency of the interference signal to be selected is compared with the operating frequency of other tasks of this vehicle, and the absolute value of the difference between the two is calculated. If the absolute value is less than the preset protection bandwidth, the frequency avoidance mechanism is triggered. In the interference strategy library, a backup interference waveform of the same type as the target but with a different frequency is selected, or the interference signal is actively notched through modulation to protect the normal operation of the vehicle's electronic equipment.

6. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 5, characterized in that: Following step S6, there is also an online evaluation of the interference effect and a strategy fine-tuning step: Step S61: After the interference signal is emitted, continuously monitor the changes in the downlink signal strength of the target UAV and the offset of the motion trajectory in the photoelectric image; Step S62: Construct an effectiveness evaluation function that calculates the effectiveness value of the current interference strategy by weighted summation of the signal strength attenuation rate and the trajectory offset angular rate; The specific calculation process is as follows: First, a weighting factor is set, which is set according to whether the target is mainly interference from the image transmission link or the navigation link; then, the signal strength attenuation rate is multiplied by the weighting factor to obtain the first weighted value; then, the trajectory deviation angular rate is multiplied by a factor complementary to the weighting factor to obtain the second weighted value; finally, the first weighted value and the second weighted value are added together to obtain the performance value. Step S63: If the performance value of multiple consecutive evaluation cycles is lower than the preset performance threshold under the current interference parameters, the current interference strategy is determined to be ineffective, and the system will automatically switch to the next candidate strategy of the same type in the interference strategy library until the performance value recovers.

7. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 6, characterized in that: In step S63, the system switches to the next candidate policy of the same type, specifically using an ε-greedy policy based on reinforcement learning for exploration and utilization: The target signal strength and motion trajectory parameters monitored at the current moment are taken as the current state s; Initialize an action-value table for this target model, recording the cumulative performance value under different state-action pairs; Given the current state s, set an exploration probability ε; Select the interference action with the highest value in the current action-value table with probability 1-ε, and randomly select other interference actions with probability ε. Based on the performance values ​​fed back after the actions are performed, the action-value table is updated online to achieve adaptive evolution of the strategy.

8. The strategy decision generation method for vehicle-mounted UAV interference detection equipment according to claim 7, characterized in that: When the vehicle platform detects multiple incoming targets, the method also includes an interference resource scheduling step based on spatiotemporal conflict detection: Extract azimuth, elevation, and threat levels of multiple targets; The beam dwell time and mechanical rotation range of the jamming device are modeled as constraints to determine the total available time window within a scheduling cycle, denoted as ΔT. By solving a 0-1 integer programming problem, the illumination order and dwell time allocation of the jamming beams are determined within the total time window ΔT, so as to maximize the sum of the weighted threat levels of the intercepted targets. Specifically, the decision variable xi is set to represent whether target i is illuminated in this round of scheduling, and xi takes the value of 0 or 1. An objective function is established to maximize the sum of the threat levels of all illuminated targets, while constraints are established to require that the sum of the dwell time required by all illuminated targets and the beam adjustment stabilization time does not exceed the total time window ΔT. Based on the calculation results of 0-1 integer programming, the optimal multi-target interference timing instruction is output.

9. A strategy decision generation system for vehicle-mounted unmanned aerial vehicle (UAV) interference detection equipment, wherein the system is applied in any one of the methods described in claims 1-8, characterized in that: the system comprises: A multimodal sensor array, deployed on an onboard platform, includes at least a radio monitoring and direction finding unit and an optoelectronic tracking unit, used to continuously acquire raw perception data of the target airspace while the vehicle is in motion or parked. The data preprocessing module, connected to the multimodal sensor group, is used to preprocess the raw sensing data, separate the target signals suspected to be UAVs, and extract multidimensional feature parameters for each target signal; the multidimensional feature parameters include at least the signal angle of arrival, signal strength, signal-to-noise ratio, carrier frequency, bandwidth, pulse repetition interval, and target size in the photoelectric image; The association matching module, connected to the data preprocessing module, is used to perform association matching between the multi-dimensional feature parameters extracted in the current period and the locally stored historical trajectory database based on a sliding time window, to filter out persistent targets that meet physical kinematic constraints and to filter out transient noise interference. The threat assessment module, connected to the association matching module, is used to input the multi-dimensional feature parameters of successfully associated persistent targets into a lightweight threat assessment network to calculate the real-time threat level of the target. The lightweight threat assessment network adopts a decision tree structure, and its node splitting threshold is dynamically adjusted based on the root mean square value of background noise measured in real time by the vehicle platform and the signal-to-noise ratio of the photoelectric image to suppress false alarms. The strategy generation module, connected to the threat assessment module, is used to call the corresponding interference waveform parameters from the preset interference strategy library based on the target's current master control feature parameters when the real-time threat level exceeds the preset intervention threshold; where the master control feature parameters are the feature dimensions with the highest contribution rate after principal component analysis of the multidimensional feature parameters. The interference control module, connected to the strategy generation module, is used to jointly calculate the interference waveform parameters with the vehicle's current heading and attitude information to generate servo control commands for beam pointing and excitation signals for the waveform generator, thereby driving the interference equipment to perform directional tracking and interference on the target.

Images