Data processing method and device based on anti-unmanned aerial vehicle scene, equipment and medium

Abstract

The invention discloses a data processing method, device and equipment based on an anti-unmanned aerial vehicle scene and a medium, and the method comprises the steps: collecting unmanned aerial vehicle data through a data collection device configured by an anti-unmanned aerial vehicle system, and generating a first data set and a second data set based on the unmanned aerial vehicle data; inputting the first data set into a main model, and outputting a corresponding first prediction result by the main model; and inputting the second data set into a second model, and outputting a corresponding second prediction result by the second model, comparing and analyzing the first prediction result and the second prediction result to obtain a consistency index; when the result is consistent, feeding back a consistency index to the anti-unmanned aerial vehicle system, so that the anti-unmanned aerial vehicle system generates an anti-unmanned aerial vehicle control strategy; and when the results are inconsistent, performing parameter updating on the main model or the second model, and feeding back to the anti-unmanned aerial vehicle system. According to the invention, the data processing effect based on the anti-unmanned aerial vehicle scene can be improved, and the control effect of the anti-unmanned aerial vehicle system is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data processing method, device, equipment and medium based on an anti-UAV scenario. Background Art

[0002] The Counter-Unmanned Aircraft Systems-command and control (CUAS-C2) system relies on multiple sensors to provide environmental awareness and situational analysis of target drones. For example, it collects data through devices such as radar, time difference of arrival (TDOA), global positioning system (GPS), and electro-optical sensors, and then performs environmental awareness and situational analysis based on the collected data.

[0003] However, due to the limited computing power of the system's various components, there are deficiencies in processing the heterogeneity of multi-sensor data, noise interference, outliers, and adaptability to dynamic environments. This makes it difficult for detection accuracy and real-time detection to meet standard requirements. In addition, existing detection solutions lack a comprehensive processing mechanism for sensor data classification, effective data fusion, and model adaptive optimization, resulting in an inability to fully meet the high reliability and rapid response requirements of the CUAS-C2 system. Therefore, how to improve the data processing effect based on anti-UAV scenarios and thereby enhance the control effect of the CUAS-C2 system is a problem that technicians in this field need to solve. Summary of the Invention

[0004] Embodiments of the present invention provide a data processing method, apparatus, computer equipment, and storage medium based on anti-UAV scenarios, aiming to improve the data processing effect based on anti-UAV scenarios and thereby enhance the control effect of the anti-UAV system.

[0005] In a first aspect, an embodiment of the present invention provides a data processing method based on an anti-UAV scenario, comprising:

[0006] Collecting drone data through a data acquisition device configured for the anti-drone system, and classifying and fusing the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions;

[0007] Inputting the first data set into a primary model, and having the primary model output a corresponding first prediction result; and inputting the second data set into a secondary model, and having the second model output a corresponding second prediction result;

[0008] Comparing and analyzing the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent;

[0009] When the consistency indicator shows a consistent result, the consistency indicator is fed back to the anti-UAV system, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set;

[0010] When the consistency indicator shows inconsistent results, the parameters of the main model or the second model are updated based on the consistency indicator and fed back to the anti-UAV system.

[0011] In a second aspect, an embodiment of the present invention provides a data processing device based on an anti-UAV scenario, comprising:

[0012] a data acquisition unit, configured to collect drone data through a data acquisition device configured in the anti-drone system, and classify and fuse the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions;

[0013] a model prediction unit, configured to input the first data set into a main model, and have the main model output a corresponding first prediction result; and input the second data set into a second model, and have the second model output a corresponding second prediction result;

[0014] A result analysis unit is used to compare and analyze the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent;

[0015] a first feedback unit, configured to feed back the consistency indicator to the anti-UAV system when the consistency indicator shows a consistent result, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set;

[0016] The second feedback unit is used to update the parameters of the main model or the second model based on the consistency index when the consistency index shows an inconsistent result, and feed back the updated parameters to the anti-UAV system.

[0017] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the data processing method based on the anti-UAV scenario as described in the first aspect is implemented.

[0018] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the data processing method based on the anti-UAV scenario as described in the first aspect is implemented.

[0019] Embodiments of the present invention provide a data processing method, apparatus, computer equipment, and storage medium based on an anti-UAV scenario. The method can classify, fuse, and adaptively optimize the collected data, thereby improving the reliability and response speed of the anti-UAV system, and enhancing the detection accuracy and real-time performance of the anti-UAV system for UAVs, thereby improving the efficiency of UAV countermeasures and enhancing the control effect of the anti-UAV. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0021] Figure 1 A flowchart of a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0022] Figure 2 A schematic diagram of a sub-flow chart of step S101 in a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0023] Figure 3 A schematic diagram of a sub-flow chart of step S103 in a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0024] Figure 4 A schematic diagram of a sub-flow chart of step S301 in a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0025] Figure 5 A schematic diagram of a sub-flow chart of step S104 in a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0026] Figure 6 Another flowchart of a data processing method based on an anti-UAV scenario provided by an embodiment of the present invention;

[0027] Figure 7 A schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention;

[0028] Figure 8 A first sub-schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention;

[0029] Figure 9 A second sub-schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention;

[0030] Figure 10 A third sub-schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention;

[0031] Figure 11 A fourth sub-schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention;

[0032] Figure 12 Another schematic block diagram of a data processing device based on an anti-UAV scenario provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0034] It will be understood that when used in this specification and the appended claims, the terms “comprises” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0035] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the present invention. As used in the specification and appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0036] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0037] See below Figure 1 , an embodiment of the present invention provides a data processing method based on an anti-UAV scenario, specifically comprising: steps S101 to S105.

[0038] Step S101: collecting drone data using a data collection device configured in the anti-drone system, and classifying and fusing the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions;

[0039] Step S102: input the first data set into a main model, and the main model outputs a corresponding first prediction result; and input the second data set into a second model, and the second model outputs a corresponding second prediction result;

[0040] Step S103: Compare and analyze the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent;

[0041] Step S104: When the consistency indicator shows a consistent result, the consistency indicator is fed back to the anti-UAV system, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set;

[0042] Step S105: When the consistency indicator indicates inconsistent results, the parameters of the main model or the second model are updated based on the consistency indicator and fed back to the anti-UAV system.

[0043] In this embodiment, drone data is first collected using a data acquisition device to generate a first dataset and a second dataset, which are used for analysis in the primary model and the secondary model, respectively. This ensures data consistency and reliability, thereby addressing issues such as sensor noise and heterogeneity. Next, a primary model analysis is performed, where the first dataset is input into the primary model. The primary model can be a prediction or decision-making algorithm adapted to the real-time requirements of drones (such as a long short-term memory network (LSTM) or the object detection algorithm YOLO). The primary model generates a first prediction result, such as target position prediction, target count prediction, or threat assessment. Simultaneously, a secondary model analysis is performed, where the second dataset is input into a second model. The second model is an independent model with a different algorithm or parameter configuration than the primary model. It can generate a second prediction result, such as trajectory prediction, that differs from the first. The second model provides independent results that can be used to subsequently verify the reliability of the primary model. Subsequently, a result verification and consistency comparison is performed, where the first prediction result obtained from the primary model analysis is compared with the second prediction result obtained from the secondary model analysis. The consistency between the first and second prediction results is evaluated using a consistency index to determine whether the first and second prediction results are consistent. This allows the reliability of the model output to be determined, helping to reduce the risk of incorrect decisions.

[0044] When the first and second predictions are consistent, command actions (such as path planning and target lock) or decisions (such as continued monitoring and mode switching) can be executed based on the comparison results. The consistency results are also fed back to the system to optimize the command strategy. When the first and second predictions are inconsistent, the cause of the inconsistency (such as sensor failure and model overfitting) is analyzed, and the primary or secondary model is updated through online learning or parameter adjustment to achieve system adaptability and fault tolerance.

[0045] Here, when updating model parameters, the analysis results of the inconsistency reasons can be used to determine whether to update the main model or the secondary model. For example, when the sensor corresponding to the main model is faulty, resulting in unreliable collected data, which in turn causes prediction deviations in the main model, the main model parameters can be updated (correcting the faulty sensor and using the data collected by the corrected sensor to train and update the main model). For another example, when the second model is overfitting, the second model can be updated. In other words, the specific model to be updated can be based on the respective prediction results. That is, when a prediction result deviates, the corresponding model is updated.

[0046] Finally, record the verification results, execution actions, and task feedback (such as prediction error and task success rate). When recording feedback, if the consistency indicator indicates consistent results, record that consistency indicator directly. If the consistency indicator indicates inconsistent results, record the parameter update action and results, or record the consistency indicator for inconsistent results together. The specific setting can be set based on actual needs. Then, based on the feedback data, optimize the data processing process, model performance, or sensor configuration. This allows for continuous improvement of system performance through long-term feedback.

[0047] This embodiment can classify, fuse and adaptively optimize the collected data, thereby improving the reliability and response speed of the anti-UAV system, improving the detection accuracy and real-time performance of the anti-UAV system for UAVs, thereby improving the efficiency of UAV countermeasures and enhancing the control effect of anti-UAVs.

[0048] In actual application scenarios, machine learning can be applied to extract the motion and trajectory characteristics of drones from the multi-sensor heterogeneous big data within the distributed CUAS-C2 system's data architecture. Simulation analysis can then be used to develop probabilistic proxy models by coupling numerical simulation with machine learning. Furthermore, raw data can be acquired from multiple sensors on low-altitude drones, and a unified dataset can be generated through denoising, time synchronization, and format standardization. Data fusion algorithms (such as Kalman filtering or deep learning fusion) can then be used to generate the first and second datasets. Here, the first and second datasets can be data acquired from different locally deployed sensors, locally stored datasets of the same type of sensor but at different times, or acquired from heterogeneous sensors in different locations.

[0049] In one embodiment, if Figure 2 As shown, the step S101 includes steps S1011 to S1013.

[0050] Step S1011: classify the drone data and assign metadata tags to the drone data;

[0051] Step S1012: pre-processing the classified drone data; wherein the pre-processing includes data cleaning and data time series normalization;

[0052] Step S1013: performing data fusion on the preprocessed UAV data to obtain the first data set and the second data set.

[0053] In this embodiment, when generating a dataset, the collected drone data is first classified. For example, classification can be performed based on the physical type of the sensor. For data acquisition devices such as radar, TDOA, CPS, and optoelectronics, classification can be based on the sensor's perception principle or data output dimension. Metadata tags are then assigned to each data stream to ensure compatibility with specific sensor algorithms and facilitate subsequent processing. Next, multi-sensor data fusion and preprocessing are performed on the data from different data acquisition devices to integrate measurements from different sensor types and generate a unified, time-aligned dataset.

[0054] Preprocessing involves data cleaning to remove unnecessary data, such as noise, outliers, or other interfering signals. This cleaning process can be implemented using filtering techniques, such as Kalman filtering or low-pass filtering, to improve data quality. Preprocessing also involves standardizing or formatting time series data, such as drone image sequences, sound signals, or spatial coordinates. Multi-sensor data fusion is then performed on the classified data to generate a first dataset and a second dataset, which are then used to meet the input requirements of the subsequent analysis modules.

[0055] In one embodiment, if Figure 3 As shown, the step S102 includes steps S1021 to S1024.

[0056] Step S1021: Utilize the master model to perform multi-dimensional observation on the first data set to obtain corresponding observation results;

[0057] Step S1022: performing a performance evaluation on the observation result to obtain a corresponding performance evaluation result; wherein the performance evaluation result is whether the performance meets expectations or whether the performance does not meet expectations;

[0058] Step S1023: When the performance evaluation result shows that the performance meets expectations, output the performance evaluation result as the first prediction result;

[0059] Step S1024: When the performance evaluation result shows that the performance does not meet expectations, the main model is adaptively optimized.

[0060] In this embodiment, when processing the input data set through the model, the data in the data set is first observed in multiple dimensions through an observation module with noise, and then a comparison algorithm is used to perform performance evaluation. Here, the consistency of the observed output with the expected output of the model can be judged based on preset performance indicators (such as accuracy, error range, etc.). If the observed output matches the model predicted output, that is, the performance evaluation result meets expectations, the corresponding data can be used for subsequent prediction analysis or output to the decision module of the anti-UAV system, and the performance evaluation result is output as the first prediction result; if the observed output does not match the expected output of the model, or it is confirmed through the judgment box that there is a significant deviation, that is, the performance evaluation result does not meet expectations, then the adaptive optimization process is triggered. When performing adaptive optimization, statistical methods (such as least squares method, Bayesian inference) or machine learning techniques (such as neural networks, reinforcement learning) can be used to update model parameters. For example, when using the least squares method for target velocity and acceleration fitting, the specific steps may include: (1) assuming that the output y has a linear relationship with the input feature x; (2) collecting a batch of the latest observation data and calculating the prediction error; (3) minimizing the square error loss function; (4) updating the parameter w.

[0061] Furthermore, the updated model parameters can be fed back to the observation module, and the input data can be reprocessed until the output meets the requirements of the performance evaluation module. This optimization process is iterative, ensuring that the model adapts to dynamic environments or new sensor data characteristics. It should also be noted that while the analysis here focuses on the first dataset and the primary model, in practice, the above process can also be applied to analyze a second dataset and a second model, and can even be extended to other sensor types or application scenarios.

[0062] Specifically, such as Figure 4 As shown, the step S1021 includes: steps S10211 to S10214.

[0063] Step S10211: performing image observation on the drone data in the first data set to extract the appearance features or behavior patterns of the drone;

[0064] Step S10212: performing motion observation on the drone data in the first data set to extract motion parameters;

[0065] Step S10213: performing trajectory observation on the drone data in the first data set to track and obtain the spatial position and motion trajectory of the drone;

[0066] Step S10214: Summarize the appearance features or behavior patterns, motion parameters, spatial position and motion trajectory of the drone as the observation results.

[0067] In this embodiment, the multi-dimensional observation may specifically include image observation, motion observation, and trajectory observation. Image observation refers to processing image or video data to extract the appearance features or behavior patterns of the drone. Motion observation refers to analyzing the speed, acceleration, or other motion-related parameters of the drone to extract important information from the dynamic behavior of the drone for subsequent tasks such as target identification, threat assessment, and model verification. For example, rapid acceleration or circling behavior may indicate evasion, reconnaissance, or attack intentions. Through the speed and acceleration feature vectors, established behavior templates can be matched, such as passage, abnormal circling, rapid escape, intrusion into the target area, etc. Trajectory observation refers to tracking and obtaining the spatial position and motion trajectory of the drone. Here, each dimension of observation will generate a specific type of output based on the input data, such as a feature vector, motion parameter, or trajectory prediction. The corresponding observation results will be input into the subsequent performance evaluation module to evaluate whether the output meets the predefined model expected output.

[0068] In one embodiment, if Figure 5 As shown, the step S103 includes: steps S1031 to S1033.

[0069] Step S1031: performing comparative analysis on the first prediction result and the second prediction result using an index evaluation method; wherein the index evaluation method includes any one of Euclidean distance calculation, KL divergence calculation, or confidence interval calculation;

[0070] Step S1032: When the result of the comparative analysis meets the preset result threshold, the consistency indicator is set to result consistency;

[0071] Step S1033: When the result of the comparative analysis does not meet the preset result threshold, the consistency indicator is set to inconsistent results.

[0072] When evaluating the consistency of the first and second prediction results, a consistency metric (such as Euclidean distance, KL divergence, or confidence interval) can be used. If the results meet a preset threshold, they are considered consistent; otherwise, they are considered inconsistent. This step ensures the reliability of the model output and reduces the risk of incorrect decisions.

[0073] In one embodiment, if Figure 6 As shown, the data processing method based on the anti-UAV scenario also includes: steps S201 to S205.

[0074] Step S201: using a state variable estimator to estimate the pre-processed UAV data to obtain corresponding state variables;

[0075] Step S202: Adopting an adaptive multi-sensor data association filter neural network to dynamically adjust the association threshold of the drone data;

[0076] Step S203: Based on the adjusted threshold, calculate the association probability that the drone data collected by different data collection devices belong to the same drone target;

[0077] Step S204: Based on the association probability, update the state probability distribution of the UAV target through a filtering method;

[0078] Step S205: Generate a current optimal state estimation result based on the updated state probability distribution, and generate an anti-UAV control strategy based on the current optimal estimation result.

[0079] In this embodiment, a dynamic correlation filtering process based on adaptive multi-sensor data is used to further optimize data processing and evaluation, thereby improving the efficiency of drone countermeasures and further enhancing the control effectiveness of the anti-drone system. Specifically, the collected drone data is first pre-processed through denoising, filtering, and normalization. The noise characteristics of each sensor data are quantified to improve data quality. For example, Gaussian filtering is performed on images collected by the camera to remove noise, and Kalman filtering is performed on data collected by the IMU sensor to reduce jitter.

[0080] Next, the state variable estimator is used to process the preprocessed data and estimate the state variables of the drone target (such as position, velocity, heading, etc.). For example, by fusing radar and IMU data, the three-dimensional coordinates and velocity of the target can be estimated.

[0081] Then, the AMSDAF (Adaptive Multi-Sensor Data Association Filter) neural network is used to dynamically adjust the data association threshold according to the environment and data characteristics to improve the accuracy of target matching. For example, when the targets are dense, the threshold is automatically increased to reduce false associations, while when the targets are sparse, the threshold is lowered to avoid missed detections. In practical applications, threshold association can be achieved according to the following steps: (1) Set the target state and observation model; (2) Calculate the distance / residual between each pair of target-observation; (3) Eliminate impossible associations based on the threshold; (4) Calculate the association probability (JPDA or Softmax). Based on the threshold adjusted by AMSDAF, the system can calculate the residual (such as Mahalanobis distance) between each observation and the target, and then normalize it into an association probability to determine which observation data belongs to which target, thereby achieving highly robust multi-target tracking.

[0082] Then, using methods such as Bayesian filtering and Kalman filtering, combined with the new observation data, the state probability distribution of the drone target is updated. For example, the Kalman filter can be used to correct the drone's speed and position estimates based on the latest radar and IMU data.

[0083] It then outputs the current optimal state estimation result and estimates relevant parameters (such as target type, threat level, etc.). For example, the system outputs the target's precise location and speed, and determines whether it is a "suspicious drone" or a "friendly drone."

[0084] Finally, based on the state estimation results, corresponding decision suggestions are generated, such as interception, avoidance, tracking, etc. For example, the system recommends activating a jammer to perform electronic interference on a suspicious drone.

[0085] It's important to note that when performing state estimation on a drone target, the final state estimation result typically contains a variety of important information. For example, the drone target's current location can be represented by its position, typically in a geographic coordinate system (e.g., longitude and latitude, or a local coordinate system). It can also include state information such as velocity estimate, acceleration estimate, and heading angle. Based on this estimated state information, the system can generate a series of decision recommendations to support automated responses by the drone or command system. Based on the target's speed and position, it can determine whether to employ interceptors, drone tracking, or other disruptive measures. For example, if the drone target is moving along a predetermined flight path, the system can generate an evasive route based on the target's position and speed to ensure flight safety. For another example, if the target state estimation indicates that the drone target is rapidly approaching or changing direction, the system may automatically increase its threat level and initiate stronger countermeasures. For example, if the state estimation confirms that the drone target is an enemy target and is within attack range, the system may recommend activating the automatic target lock system and prepare for attack operations.

[0086] In one embodiment, the data processing method based on the anti-UAV scenario further includes:

[0087] Acquiring performance data of the anti-UAV system, and evaluating the anti-UAV system based on the performance data to obtain a system evaluation result of the anti-UAV system;

[0088] When the system evaluation result of the anti-UAV system is lower than a preset evaluation threshold, a system optimization mechanism is triggered, and the anti-UAV system is dynamically optimized using the system optimization mechanism.

[0089] This embodiment can also evaluate the degree of performance degradation of the anti-UAV system, update the intervention strategy of the anti-UAV system accordingly, and dynamically optimize the system performance. Specifically, first evaluate the system performance (such as detection accuracy, tracking error, etc.) in real time to determine whether performance degradation occurs. For example, if it is found that the tracking error increases and the detection accuracy decreases, it indicates that the system performance has degraded. The degree of system performance degradation is evaluated, and according to the performance degradation evaluation results (i.e., the system evaluation results), the intervention strategy of the anti-UAV (CUAS) system is dynamically adjusted to improve the defense effect. For example, according to the target speed and trajectory, the flight path or interference frequency of the intercepted UAV is adjusted. During dynamic optimization, if the system performance is lower than the preset evaluation threshold, the optimization mechanism is automatically triggered. Specifically, the system performance can be restored by adjusting the filter parameters, retraining the neural network, etc. For example, if it is detected that the tracking error continues to exceed the standard, the system automatically optimizes the filter parameters or switches to the backup algorithm.

[0090] In practical applications, when the system evaluation results show excellent (i.e., the system is operating normally with high accuracy and small error, which can be understood as the system evaluation results of the anti-UAV system are higher than the preset evaluation threshold), it indicates that the current parameter configuration, filtering algorithm and sensor status are performing well, and the system can accurately perform the task. When the performance evaluation results are poor (which can be understood as the system evaluation results of the anti-UAV system are lower than the preset evaluation threshold), the system can take the following measures to adjust: (1) Considering that the decline in system performance may be due to inaccurate models or changes in noise characteristics, the system performance can be improved by adjusting the covariance matrix and re-estimating the system model. (2) If neural networks and machine learning models are used, new samples can be added for training to improve the generalization ability of the network. Incremental learning methods can also be used to gradually adjust the network model instead of retraining the entire network. (3) If there is a problem with the sensor, it can be calibrated and reconfigured, such as by detecting the sensor's status indicators (such as the sensor's noise level, accuracy, etc.) and making necessary adjustments.

[0091] Figure 7A schematic block diagram of a data processing device 300 based on an anti-UAV scenario provided by an embodiment of the present invention, the device 300 includes:

[0092] A data acquisition unit 301 is configured to collect drone data using a data acquisition device configured in the anti-drone system, and to classify and fuse the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions;

[0093] The model prediction unit 302 is configured to input the first data set into a main model, and have the main model output a corresponding first prediction result; and input the second data set into a second model, and have the second model output a corresponding second prediction result;

[0094] A result analysis unit 303 is configured to compare and analyze the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent;

[0095] A first feedback unit 304 is configured to feed back the consistency indicator to the anti-UAV system when the consistency indicator shows a consistent result, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set;

[0096] The second feedback unit 305 is configured to update parameters of the main model or the second model based on the consistency indicator when the consistency indicator indicates inconsistent results, and provide feedback to the anti-UAV system.

[0097] In one embodiment, if Figure 8 As shown, the data acquisition unit 301 includes:

[0098] a data classification unit 3011, configured to classify the drone data and assign metadata tags to the drone data;

[0099] The data preprocessing unit 3012 is used to preprocess the classified drone data; wherein the preprocessing includes data cleaning and data time series normalization;

[0100] The data fusion unit 3013 is used to perform data fusion on the pre-processed UAV data to obtain the first data set and the second data set.

[0101] In one embodiment, if Figure 9 As shown, the model prediction unit 302 includes:

[0102] A data observation unit 3021 is configured to perform multi-dimensional observation on the first data set using the main model to obtain corresponding observation results;

[0103] The performance evaluation unit 3022 is configured to perform a performance evaluation on the observation result to obtain a corresponding performance evaluation result; wherein the performance evaluation result is whether the performance meets expectations or whether the performance does not meet expectations;

[0104] A result output unit 3023 is configured to output the performance evaluation result as the first prediction result when the performance evaluation result indicates that the performance meets expectations;

[0105] The adaptive optimization unit 3024 is configured to perform adaptive optimization processing on the main model when the performance evaluation result indicates that the performance does not meet expectations.

[0106] In one embodiment, if Figure 10 As shown, the data observation unit 3021 includes:

[0107] An image observation unit 30211 is configured to perform image observation on the drone data in the first data set to extract appearance features or behavior patterns of the drone;

[0108] a motion observation unit 30212, configured to perform motion observation on the drone data in the first data set to extract motion parameters;

[0109] a trajectory observation unit 30213, configured to perform trajectory observation on the UAV data in the first data set to track the spatial position and motion trajectory of the UAV;

[0110] The observation aggregation unit 30214 is used to aggregate the appearance features or behavior patterns, motion parameters, and track and obtain the spatial position and motion trajectory of the drone into the observation results.

[0111] In one embodiment, if Figure 11 As shown, the result analysis unit 303 includes:

[0112] An index analysis unit 3031 is configured to perform comparative analysis on the first prediction result and the second prediction result using an index evaluation method; wherein the index evaluation method includes any one of Euclidean distance calculation, KL divergence calculation, or confidence interval calculation;

[0113] The first setting unit 3032 is configured to set the consistency indicator to result consistency when the result of the comparative analysis meets a preset result threshold;

[0114] The second setting unit 3033 is configured to set the consistency indicator to inconsistent results when the result of the comparative analysis does not meet a preset result threshold.

[0115] In one embodiment, if Figure 12As shown, the data processing device 300 based on the anti-UAV scenario also includes:

[0116] The state estimation unit 310 is used to estimate the pre-processed UAV data using a state variable estimator to obtain corresponding state variables;

[0117] A threshold adjustment unit 320 is used to dynamically adjust the association threshold of the drone data using an adaptive multi-sensor data association filter neural network;

[0118] The association calculation unit 330 is used to calculate the association probability that the drone data collected by different data collection devices belong to the same drone target based on the adjusted threshold;

[0119] a distribution updating unit 340 for updating the state probability distribution of the UAV target by a filtering method based on the association probability;

[0120] The state generation unit 350 is used to generate a current optimal state estimation result based on the updated state probability distribution, and generate an anti-UAV control strategy according to the current optimal estimation result.

[0121] In one embodiment, the data processing device 700 based on the anti-UAV scenario further includes:

[0122] a system evaluation unit, configured to obtain performance data of the anti-UAV system, and evaluate the anti-UAV system based on the performance data to obtain a system evaluation result of the anti-UAV system;

[0123] A dynamic optimization unit is used to trigger a system optimization mechanism when the system evaluation result of the anti-UAV system is lower than a preset evaluation threshold, and use the system optimization mechanism to dynamically optimize the anti-UAV system.

[0124] Since the embodiments of the apparatus part correspond to the embodiments of the method part, please refer to the description of the embodiments of the method part for the embodiments of the apparatus part, and will not be repeated here.

[0125] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When executed, the computer program can implement the steps provided in the above embodiments. The storage medium can include a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, among other media capable of storing program code.

[0126] The present invention also provides a computer device that may include a memory and a processor. The memory stores a computer program, and when the processor calls the computer program in the memory, the steps provided in the above embodiment can be implemented. Of course, the computer device may also include various network interfaces, a power supply, and other components.

[0127] The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of this application, several improvements and modifications can be made to this application, and these improvements and modifications also fall within the scope of protection of the claims of this application.

[0128] It should also be noted that, in this specification, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus comprising the element.

Claims

1. A data processing method based on anti-UAV scenario, characterized in that: include: Collecting drone data using a data acquisition device configured for the anti-drone system, and classifying and fusing the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions; Inputting the first data set into a primary model, and having the primary model output a corresponding first prediction result; and inputting the second data set into a secondary model, and having the second model output a corresponding second prediction result; Comparing and analyzing the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent; When the consistency indicator shows a consistent result, the consistency indicator is fed back to the anti-UAV system, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set; When the consistency indicator shows inconsistent results, the parameters of the main model or the second model are updated based on the consistency indicator and fed back to the anti-UAV system.

2. The data processing method based on the anti-UAV scenario according to claim 1 is characterized in that: The data collection device configured by the anti-UAV system collects UAV data, and classifies and fuses the UAV data to generate a first data set and a second data set, including: classifying the drone data and assigning metadata tags to the drone data; Preprocess the classified drone data; preprocessing includes data cleaning and data time series standardization; The preprocessed UAV data is fused to obtain the first data set and the second data set.

3. The data processing method based on the anti-UAV scenario according to claim 1 is characterized in that: Inputting the first data set into the main model, and having the main model output a corresponding first prediction result, includes: Using the main model to perform multi-dimensional observation on the first data set to obtain corresponding observation results; Performing a performance evaluation on the observation result to obtain a corresponding performance evaluation result; wherein the performance evaluation result is whether the performance meets expectations or whether the performance does not meet expectations; When the performance evaluation result shows that the performance meets expectations, outputting the performance evaluation result as the first prediction result; When the performance evaluation result shows that the performance does not meet expectations, the main model is adaptively optimized.

4. The data processing method based on the anti-UAV scenario according to claim 3 is characterized in that: The using the main model to perform multi-dimensional observation on the first data set to obtain corresponding observation results includes: Performing image observation on the drone data in the first data set to extract appearance features or behavior patterns of the drone; Performing motion observation on the drone data in the first data set to extract motion parameters; Performing trajectory observation on the drone data in the first data set to track and obtain the spatial position and motion trajectory of the drone; The appearance features or behavior patterns, motion parameters, spatial position and motion trajectory of the UAV are summarized as the observation results.

5. The data processing method based on the anti-UAV scenario according to claim 1 is characterized in that: The comparison and analysis of the first prediction result and the second prediction result to obtain a consistency index includes: Performing a comparative analysis on the first prediction result and the second prediction result using an indicator evaluation method; wherein the indicator evaluation method includes any one of Euclidean distance calculation, KL divergence calculation, or confidence interval calculation; When the result of the comparative analysis meets the preset result threshold, the consistency indicator is set to the result consistency; When the result of the comparative analysis does not meet the preset result threshold, the consistency indicator is set to inconsistent results.

6. The data processing method based on the anti-UAV scenario according to claim 1 is characterized in that: Also includes: The state variable estimator is used to estimate the preprocessed UAV data to obtain the corresponding state variables; Adopting adaptive multi-sensor data association filter neural network to dynamically adjust the association threshold of UAV data; Based on the adjusted threshold, the association probability that the drone data collected by different data collection devices belong to the same drone target is calculated; Based on the association probability, the state probability distribution of the UAV target is updated through a filtering method; Based on the updated state probability distribution, the current optimal state estimation result is generated, and the anti-UAV control strategy is generated according to the current optimal estimation result.

7. The data processing method based on the anti-UAV scenario according to any one of claims 1 to 6, characterized in that: Also includes: Acquiring performance data of the anti-UAV system, and evaluating the anti-UAV system based on the performance data to obtain a system evaluation result of the anti-UAV system; When the system evaluation result of the anti-UAV system is lower than a preset evaluation threshold, a system optimization mechanism is triggered, and the anti-UAV system is dynamically optimized using the system optimization mechanism.

8. A data processing device based on anti-UAV scenario, characterized in that: include: a data acquisition unit, configured to collect drone data through a data acquisition device configured in the anti-drone system, and classify and fuse the drone data to generate a first data set and a second data set; wherein the first data set and the second data set have different data compositions; a model prediction unit, configured to input the first data set into a main model, and have the main model output a corresponding first prediction result; and input the second data set into a second model, and have the second model output a corresponding second prediction result; A result analysis unit is used to compare and analyze the first prediction result and the second prediction result to obtain a consistency index; wherein the consistency index is whether the results are consistent or inconsistent; a first feedback unit, configured to feed back the consistency indicator to the anti-UAV system when the consistency indicator shows a consistent result, so that the anti-UAV system generates an anti-UAV control strategy based on the first data set and the second data set; The second feedback unit is used to update the parameters of the main model or the second model based on the consistency index when the consistency index shows an inconsistent result, and feed back the updated parameters to the anti-UAV system.

9. A computer device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the data processing method based on the anti-UAV scenario as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the data processing method based on the anti-UAV scenario according to any one of claims 1 to 7 is implemented.

Images