Low-altitude flight target identification method and system based on three-dimensional digital earth

Abstract

The invention relates to the technical field of flight target identification, in particular to a low-altitude flight target identification method and system based on a three-dimensional digital earth. The method comprises the following steps: performing towering surface element marking on low-altitude monitoring range data to obtain towering surface element marking data; performing spatial relative distance calculation on the towering earth surface element mark data to obtain flying target-earth surface element spatial relative distance data; performing flight out-of-control probability estimation based on the flight target-surface element space relative distance data to obtain flight out-of-control probability data; performing flight target state recognition model construction on the flight out-of-control probability data to obtain a flight target state recognition model; and performing risk flight intention prediction on the low-altitude flight target according to the flight target state recognition model to obtain target risk flight intention data. According to the invention, the flight target identification technology is optimized, so that the flight target identification technology is more perfect.

Description

Technical Field

[0001] The present invention relates to the technical field of flying target recognition, and in particular to a method and system for recognizing low-altitude flying targets based on a three-dimensional digital earth. Background Art

[0002] With the widespread application of low-altitude flying targets such as drones and low-altitude aircraft in military, commercial and civilian fields, efficient and accurate identification of low-altitude flying targets has become an important topic. Traditional monitoring methods such as radar and optoelectronics are easily interfered by factors such as terrain occlusion and meteorological conditions under complex terrain conditions, making it difficult to accurately identify low-altitude flying targets. The low-altitude flying target recognition method based on three-dimensional digital earth technology can make full use of the advantages of global positioning, real-time monitoring, dynamic analysis, etc., and realize comprehensive perception and intelligent recognition of targets by building high-precision geographic information and environmental models. Three-dimensional digital earth technology integrates multiple spatial information technologies such as geographic information system (GIS), remote sensing (RS), and global navigation satellite system (GNSS), and can collect and analyze multi-dimensional data such as trajectory, attitude, and speed of low-altitude flying targets in real time. On this basis, by integrating the appearance features, motion features, and behavior patterns of the target, a more sophisticated recognition model is constructed to realize the automated and intelligent monitoring and analysis of low-altitude flying targets. This method can not only improve the recognition accuracy of low-altitude targets, but also reduce the false alarm rate and missed alarm rate, especially in complex environments and severe weather conditions. Its application scenarios cover many fields such as urban security, border monitoring, aviation supervision, and drone swarm management. It helps to improve airspace safety management capabilities and provide technical support for the safe operation and standardized management of low-altitude aircraft. With the continuous development of three-dimensional digital earth technology, this method will play a more important role in future airspace management and target identification. However, a traditional low-altitude flight target identification method based on three-dimensional digital earth has the problem of inaccurate analysis of the probability of loss of control of low-altitude flight targets and inaccurate judgment of their flight risk intentions. Summary of the invention

[0003] Based on this, it is necessary to provide a low-altitude flying target recognition method and system based on a three-dimensional digital earth to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for identifying low-altitude flying targets based on a three-dimensional digital earth is provided, the method comprising the following steps:

[0005] Step S1: acquiring low-altitude monitoring range data; marking the high-altitude surface elements on the low-altitude monitoring range data to obtain high-altitude surface element marking data;

[0006] Step S2: Calculate the spatial relative distance of the towering surface element marker data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; estimate the flight out-of-control probability based on the near-surface airflow disturbance data to obtain the flight out-of-control probability data;

[0007] Step S3: constructing a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model;

[0008] Step S4: predict the risky flight intention of the low-altitude flying target according to the flight target state recognition model, obtain the target risky flight intention data, and feed back the target risky flight intention data to the terminal.

[0009] The present invention can accurately identify and record the towering objects on the surface that affect the aircraft by acquiring low-altitude monitoring range data and marking the towering surface elements. This process helps to build a detailed map of the flight environment and provide important basic data for subsequent flight safety analysis. The towering surface element marking data can help identify potential obstacles in the flight path, thereby avoiding these obstacles in flight planning and operation and reducing the occurrence of flight accidents. By calculating the relative spatial distance between the flight target and the towering surface elements, the actual spatial relationship between the aircraft and the surface obstacles can be understood in detail. This process makes it possible to evaluate the relative position of the aircraft with the obstacle when flying at low altitude, and further analyze the impact of near-surface airflow disturbances. By conducting in-depth analysis of these disturbances, potential airflow interference areas can be identified, and the impact of airflow on the stability of the aircraft can be predicted, so that corresponding control measures can be taken during the flight. The estimation of the probability of flight out of control based on near-surface airflow disturbance data is of great significance. By analyzing the impact of airflow disturbances on the aircraft, the risk probability of aircraft out of control can be quantified, thereby providing a scientific basis for flight operations. This estimation helps pilots or automatic flight systems to better deal with airflow disturbances during flight, improve flight safety, and reduce the incidence of flight accidents. The Q-Learning algorithm is used to build a flight target state recognition model, which can effectively process flight out-of-control probability data and learn risk characteristics under different flight states. This model is continuously optimized through reinforcement learning methods, enabling it to accurately identify and predict the state of the aircraft under various environmental conditions. This model can not only help understand the potential impact of different flight states on safety, but also provide information on how to adjust flight strategies under specific flight states, thereby enhancing the adaptability and safety of aircraft in complex environments. Through a dynamically updated learning mechanism, the model can continuously improve recognition accuracy and provide personalized safety measures for different flight environments. The risk flight intention prediction based on the flight target state recognition model can provide real-time warnings of potential risks for aircraft. By predicting the risk flight intention of the target, the flight system can actively adjust the flight strategy or remind the pilot to prevent flight hazards. This prediction information is transmitted through the terminal feedback mechanism to ensure that the pilot or automatic flight system can obtain the latest risk assessment data in a timely manner and make necessary safety adjustments. This can not only improve the safety of flight operations, but also enhance the ability to respond to emergencies and significantly reduce flight risks. Through real-time feedback and adjustments, a high level of safety can be maintained during the flight.Therefore, the present invention is an optimization processing of a traditional low-altitude flight target recognition method based on three-dimensional digital earth, which solves the problems of inaccurate analysis of the probability of low-altitude flight target loss of control and inaccurate judgment of its flight risk intention in the traditional low-altitude flight target recognition method based on three-dimensional digital earth, improves the accuracy of the analysis of the probability of low-altitude flight target loss of control and improves the accuracy of the judgment of flight risk intention.

[0010] Preferably, step S1 comprises the following steps:

[0011] Step S11: Acquire low-altitude monitoring range data;

[0012] Step S12: collecting monitoring range environment data for the low-altitude monitoring range data to obtain monitoring range environment data;

[0013] Step S13: marking the high-rise surface elements on the low-altitude monitoring range data according to the monitoring range environment data to obtain high-rise surface element marking data.

[0014] The present invention obtains low-altitude monitoring range data to lay the foundation for subsequent analysis. This step forms a monitoring framework for low-altitude flight by collecting basic geographic data and environmental information of the area where the aircraft is located. By accurately obtaining data in this range, the space environment where the aircraft is located can be fully understood, ensuring that all areas that affect the flight can be covered in the subsequent analysis. The acquisition of such data is crucial to the formation of a comprehensive monitoring system, and provides raw data support for subsequent high-rise surface element marking and airflow analysis. By collecting environmental data for low-altitude monitoring range data, the environmental characteristics and surface conditions within the monitoring range can be understood in detail. This step provides necessary background data for high-rise surface element marking by collecting information on environmental factors such as climate, topography, and buildings. The collection of these environmental data helps to accurately identify and mark the high-rise surface elements that affect the aircraft, ensure the comprehensiveness and accuracy of the analysis, and thus improve the reliability of subsequent safety assessments. By marking the high-rise surface elements of the low-altitude monitoring range data according to the monitoring range environmental data, surface obstacles that pose a threat to the aircraft can be systematically identified and recorded. This marking process not only helps to form a detailed obstacle map, but also provides key information for the path planning and flight safety analysis of the aircraft. By accurately marking towering surface elements, the risk of aircraft colliding with these obstacles during low-altitude flight can be effectively reduced, flight safety can be improved, and accurate basic data can be provided for subsequent airflow disturbance analysis and estimation of the probability of flight loss of control.

[0015] Preferably, step S2 comprises the following steps:

[0016] Step S21: Real-time monitoring of low-altitude flying targets in the low-altitude monitoring range data is performed by radar to obtain real-time monitoring data of low-altitude flying targets;

[0017] Step S22: Calculate the spatial relative distance of the high-rise ground element marking data according to the real-time monitoring data of the low-altitude flying target to obtain the spatial relative distance data between the flying target and the ground element;

[0018] Step S23: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data;

[0019] Step S24: Estimate the flight out-of-control probability based on the near-surface airflow disturbance data and the real-time monitoring data of the low-altitude flying target to obtain flight out-of-control probability data.

[0020] The present invention uses radar to monitor the flying targets within the low-altitude monitoring range in real time, and can accurately capture the dynamic state and position of the aircraft. This process can provide continuous and real-time data updates, so that the movement trajectory, speed and spatial position of the aircraft can be grasped in time. This real-time monitoring is crucial to ensuring flight safety, because it can track the changes of the flying targets in real time, help detect and respond to abnormal situations, enhance the comprehensive understanding of the aircraft status, and provide key data support for subsequent spatial distance calculation and airflow analysis. According to the real-time monitoring data of low-altitude flying targets, the spatial relative distance calculation of the towering surface element marking data can be performed, and the relative position between the flying target and the towering surface obstacles can be accurately evaluated. This calculation process can reveal the actual distance between the aircraft and the ground obstacles, thereby identifying the existing spatial conflicts. This data is crucial for the path planning and safe obstacle avoidance of the aircraft, because it can provide information about the distance between the aircraft and the obstacles in a specific environment, help to make timely adjustments and avoidances in flight, and reduce the risk of accidents. Based on the flight target-surface element spatial relative distance data, the near-surface airflow disturbance analysis of the low-altitude flying target can deeply understand the impact of airflow disturbance on the aircraft. This analysis process identifies the airflow disturbance area and intensity by evaluating the airflow disturbance of the flying target at different spatial distances. These near-surface airflow disturbance data provide important environmental information for the aircraft, which helps to understand and predict the impact of airflow changes on flight stability, so that corresponding flight strategies can be adopted to deal with airflow disturbances and ensure flight safety. The probability of flight loss of control of low-altitude flying targets can be estimated based on near-surface airflow disturbance data, and the risk of loss of control of the aircraft in the current airflow environment can be quantified. This estimate provides a scientific assessment of the possibility of aircraft loss of control by comprehensively considering the potential impact of airflow disturbances on the aircraft. This risk assessment is of great guiding significance to pilots or automatic flight systems because it can provide early warning information during flight, enabling pilots to adjust flight strategies in a timely manner, reduce the probability of loss of control events, improve flight safety, and ensure the smooth completion of flight missions.

[0021] Preferably, step S23 includes the following steps:

[0022] Step S231: extracting the flight speed of the low-altitude flying target real-time monitoring data to obtain the target flight speed data;

[0023] Step S232: Decomposing the multi-directional airflow interference vector in the flight path based on the target flight speed data and the flight target-surface element space relative distance data to obtain the multi-directional airflow interference vector data;

[0024] Step S233: performing interference airflow manifold feature recognition on the multi-directional airflow interference vector data to obtain interference airflow manifold feature data;

[0025] Step S234: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target according to the interfering airflow manifold characteristic data to obtain near-surface airflow disturbance data.

[0026] The present invention can provide the speed information of the aircraft at a specific moment by extracting the flight speed from the real-time monitoring data of low-altitude flight targets. This speed data is crucial for accurately analyzing the behavior of the aircraft in a low-altitude environment, because it directly affects the degree of influence of airflow disturbance on the aircraft. By obtaining the target flight speed data, the dynamic state of the aircraft can be accurately understood, and basic data can be provided for the subsequent airflow interference analysis, so that the performance of the aircraft in a complex airflow environment can be evaluated more carefully and reliably. Based on the target flight speed data and the flight target-surface element space relative distance data, the multi-directional airflow interference vector decomposition during the flight path can identify the direction and intensity of the airflow interference suffered by the aircraft during the path process. This process can reveal the influence of airflow on the aircraft in different directions by decomposing the multi-directional airflow interference vector, and provide important data for accurately predicting the performance of the aircraft in a complex airflow environment. This decomposition capability helps to better understand the characteristics of airflow disturbances and provide a scientific basis for the aircraft to adjust the flight strategy, thereby improving flight stability and safety. The multi-directional airflow interference vector data is used to identify the manifold characteristics of the interference airflow, and the manifold characteristics of the airflow disturbance can be deeply analyzed. This process can identify the pattern, intensity and change law of airflow disturbance, so as to have a comprehensive understanding of the structural characteristics of airflow disturbance. By obtaining the manifold characteristic data of the disturbing airflow, the impact of airflow disturbance on the aircraft can be described more accurately, helping pilots or automatic flight systems to identify potential airflow disturbance areas and formulate appropriate flight adjustment strategies to ensure flight safety. The near-surface airflow disturbance analysis of the real-time monitoring data of low-altitude flight targets based on the manifold characteristic data of the disturbing airflow can comprehensively consider the impact of the airflow manifold characteristics on the aircraft. This analysis process provides a comprehensive assessment of near-surface airflow disturbances by combining the manifold characteristic data with the real-time monitoring data of the aircraft. This assessment can reveal the actual impact of airflow disturbances, predict the airflow problems encountered by the aircraft, and provide timely adjustment suggestions for pilots or automatic flight systems, thereby improving the stability and safety of the aircraft in low-altitude flight.

[0027] Preferably, step S24 comprises the following steps:

[0028] Step S241: identifying the wind cut behavior during the flight path of the low-altitude flight target real-time monitoring data according to the near-surface airflow disturbance data to obtain wind cut behavior identification data;

[0029] Step S242: performing flight airflow load simulation calculation on the low-altitude flight target real-time monitoring data according to the wind cut behavior recognition data to obtain the target flight airflow load data;

[0030] Step S243: performing load intensity distribution analysis on the target flight airflow load data to obtain load intensity distribution data;

[0031] Step S244: performing out-of-control behavior pre-identification of the low-altitude flying target according to the load intensity distribution data and the wind cut behavior identification data to obtain out-of-control behavior pre-identification data;

[0032] Step S245: Estimate the flight out-of-control probability of the low-altitude flight target real-time monitoring data according to the out-of-control behavior pre-identification data, the load intensity distribution data, and the wind cut behavior identification data to obtain the flight out-of-control probability data.

[0033] The present invention identifies the wind cut behavior of the real-time monitoring data of low-altitude flight targets based on the near-surface airflow disturbance data, and can accurately identify the wind cut phenomenon encountered by the aircraft in the flight path. This process can find out whether there is a significant wind speed change or wind cut effect in the area where the aircraft passes by analyzing the airflow disturbance data. Wind cut behavior identification can reveal the impact of the sharp change of airflow on the aircraft, thereby providing key data for subsequent load calculation and loss of control prediction, which helps to respond to the flight risks brought by airflow changes in a timely manner and ensure the stability and safety of the aircraft. According to the wind cut behavior identification data, the real-time monitoring data of low-altitude flight targets is simulated and calculated, and the specific airflow load of wind cuts and other airflow disturbances on the aircraft can be quantified. This simulation calculation process calculates the load borne by the aircraft under different airflow conditions by combining the actual wind cut identification data, and provides basic data for the subsequent load intensity analysis. The accurate simulation of airflow load data can help predict the performance of the aircraft in different airflow environments, and provide valuable reference for aircraft design and control. The load intensity distribution analysis of the target flight airflow load data can provide a detailed understanding of the intensity distribution of the airflow load on the aircraft. This analysis can reveal the load intensity and distribution law of each part of the aircraft by mapping and evaluating the spatial distribution characteristics of airflow loads. The acquisition of load intensity distribution data helps to identify the stress of the aircraft under different flight phases and airflow conditions, and provides important information for evaluating the load-bearing capacity and safety of the aircraft, so as to make corresponding improvements in design and maintenance. Pre-identification of the loss of control behavior of low-altitude flying targets based on load intensity distribution data and wind cut behavior recognition data can predict the risk of aircraft loss of control. This pre-identification process evaluates the probability of loss of control of the aircraft under airflow disturbance by integrating load intensity and wind cut behavior information. Pre-identification of loss of control behavior can detect potential loss of control risks in advance and provide early warnings for pilots or automatic flight systems, so that appropriate preventive measures can be taken during flight to reduce the occurrence of loss of control events and ensure flight safety. The flight loss of control probability is estimated based on the pre-identification data of loss of control behavior, load intensity distribution data, and wind cut behavior recognition data of real-time monitoring data of low-altitude flying targets, which can comprehensively evaluate the actual probability of aircraft loss of control. This estimation process integrates information from multiple data sources to provide a comprehensive loss of control risk assessment. By accurately calculating the probability of flight loss of control, strong data support can be provided to pilots or automatic control systems, enabling them to adjust and respond based on actual risks, optimize flight strategies, and significantly improve the safety and stability of flight missions.

[0034] Preferably, pre-identifying the out-of-control behavior of a low-altitude flying target based on the load intensity distribution data and the wind cut behavior identification data comprises the following steps:

[0035] Perform airflow directionality analysis on the wind cutout behavior recognition data to obtain wind cutout airflow directionality data;

[0036] According to the wind cutout airflow directionality data, the air pressure intensity of the three-dimensional cross section of the wind cutout behavior identification data is calculated to obtain the air pressure intensity data of the three-dimensional cross section of the space;

[0037] Based on the air pressure intensity data of the three-dimensional space section and the airflow directionality data of the wind cut, the maximum flight elevation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum flight elevation angle data of the airflow impact;

[0038] Based on the air pressure intensity data of the three-dimensional cross section and the airflow directionality data of the wind cut, the maximum lateral / longitudinal deviation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum lateral / longitudinal deviation angle data of the airflow impact;

[0039] According to the maximum flight elevation angle data of airflow impact and the maximum lateral / longitudinal deviation angle data of airflow impact, the out-of-control behavior pre-identification of the low-altitude flying target is performed to obtain the out-of-control behavior pre-identification data.

[0040] The present invention can clarify the flow direction and change law of the airflow in the wind cut area by performing directional analysis on the wind cut behavior. This analysis process can reveal the specific direction of the airflow in the wind cut area and its impact on the aircraft. Accurate wind cut airflow directional data provides a key basis for further calculation of air pressure intensity and evaluation of aircraft airflow impact. This helps to understand the direction of the airflow on the aircraft, thereby providing data support for subsequent prediction of out-of-control behavior and formulation of adjustment strategies, and improving flight safety. By calculating the air pressure intensity of the three-dimensional cross-section of space, the pressure distribution of the airflow in the wind cut area can be understood in detail. This process involves accurately calculating the pressure intensity of the airflow on different spatial cross-sections, thereby revealing the actual pressure effect of the airflow on the aircraft. Accurate air pressure intensity data helps to evaluate the specific effect of the airflow on the aircraft, and provides a scientific basis for further analysis of airflow impact and aircraft out-of-control risks, enhances the understanding of the aircraft airflow load, and thus provides a strong guarantee for flight safety. By calculating the maximum flight elevation angle of the airflow impact, the maximum elevation angle that the aircraft can withstand in the wind cut area can be predicted. This data reflects the maximum acceptable elevation angle of the aircraft under specific airflow conditions, thereby evaluating the control limit of the aircraft under airflow impact. Knowing the maximum flight elevation angle helps the pilot or automatic control system adjust the flight attitude, avoid the risk of flight loss of control caused by airflow impact, and improve the stability and safety of the aircraft in complex airflow environments. By calculating the maximum lateral and longitudinal offset angles of airflow impact, the lateral and longitudinal stability of the aircraft under airflow disturbance can be evaluated. This data can reveal the maximum lateral and longitudinal offset of the aircraft under specific airflow conditions, thereby predicting the impact of airflow on the flight path. Accurate offset angle data is crucial for the trajectory control and safe operation of the aircraft, helping to maintain flight stability in complex airflow environments and prevent deviation and loss of control due to airflow impact. By integrating the maximum flight elevation angle and maximum lateral / longitudinal offset angle data of airflow impact, the loss of control behavior of the aircraft can be comprehensively evaluated. This pre-identification process can combine multiple airflow impact factors to predict the risk of loss of control of the aircraft under complex airflow conditions. Accurate pre-identification data of loss of control behavior provides timely warnings for pilots or automatic flight systems, helping to formulate effective flight strategies and adjustment measures, thereby effectively reducing the risk of loss of control and improving the safety and reliability of flight missions.

[0041] Preferably, step S3 comprises the following steps:

[0042] Step S31: Predicting the flight target deviation trajectory based on the near-surface airflow disturbance data and the flight loss of control probability data to obtain flight target deviation trajectory prediction data;

[0043] Step S32: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data to obtain deviation trajectory smoothing data;

[0044] Step S33: constructing a flight target state recognition model based on the flight out of control probability data and the deviation trajectory smoothing data based on the Q-Learning algorithm to obtain a flight target state recognition model.

[0045] The present invention estimates the flight target deviation trajectory of the flight out of control probability data based on the near-surface airflow disturbance data, and can accurately predict the trajectory of the flight target in a complex airflow environment. This step generates an estimate of the future flight path by analyzing the impact of airflow disturbance on the aircraft, thereby revealing the movement trend of the aircraft in the out-of-control situation. Such estimated data is of great significance for formulating response strategies and optimizing flight plans, and helps to identify potential deviation risks in advance, ensuring that the aircraft can make timely adjustments during the actual flight process to maintain flight stability. Performing trajectory smoothing analysis on the flight target deviation trajectory estimation data helps to reduce trajectory fluctuations caused by airflow disturbances. Trajectory smoothing analysis processes the estimated trajectory data by applying smoothing technology to eliminate or reduce the impact caused by noise and irregular disturbances. This process can generate more coherent and stable deviation trajectory data, thereby providing more accurate trajectory prediction, helping the flight control system to make more effective decisions, and improving the trajectory tracking capability and flight safety of the aircraft. Based on the Q-Learning algorithm, a flight target state recognition model is constructed for the flight out of control probability data and the deviation trajectory smoothing data, and an intelligent flight target state recognition system can be generated. Through continuous learning and optimization, the Q-Learning algorithm can build an efficient state recognition model based on the probability of flight loss of control and trajectory data. This model can identify the state of the flight target in real time, predict the behavior of the aircraft, and provide decision support. The application of this model can effectively improve the ability to accurately assess the state of the aircraft, optimize the flight control strategy, reduce the risk of flight loss of control, and enhance overall flight safety.

[0046] Preferably, step S32 includes the following steps:

[0047] Step S321: performing trajectory offset time series difference calculation on the flight target offset trajectory estimation data to obtain trajectory offset time series difference data;

[0048] Step S322: performing a displacement discontinuity analysis on the flight target displacement trajectory estimation data according to the trajectory displacement time series difference data to obtain trajectory displacement discontinuity data;

[0049] Step S323: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data according to the trajectory deviation discontinuity data to obtain deviation trajectory smoothing data.

[0050] The present invention performs trajectory offset time series difference calculation on flight target offset trajectory prediction data, and can quantify the change amplitude of the trajectory at different time points. This process can reveal the rate and amplitude of trajectory change by calculating the trajectory offset difference between each time point, thereby providing information about the dynamic change of the flight target in time. Understanding these difference data is crucial for identifying and analyzing abnormal fluctuations and sharp changes in the flight trajectory. Through these data, the flight control system can more accurately predict future trajectory changes, make targeted adjustments, and improve the prediction accuracy and stability of the flight path. According to the trajectory offset time series difference data, the flight target offset trajectory prediction data is analyzed for offset discontinuity, and the consistency and stability of the trajectory change can be evaluated. The offset discontinuity analysis helps identify potential abnormal situations or unstable factors by detecting whether there is discontinuous or sudden offset in the trajectory change. This analysis can provide information about the continuity and stability of the trajectory, and reveal the fractures and irregular fluctuations of the flight target during flight. It is of great value for formulating smooth and stable flight control strategies, helps to prevent the risk of flight loss of control caused by intermittent offsets, and ensures the consistency of the flight path and flight safety. Trajectory smoothing analysis of the flight target deviation trajectory estimation data based on trajectory deviation discontinuity data can optimize the smoothness of trajectory data. Trajectory smoothing analysis makes the trajectory of the flight target smoother and more continuous by processing discontinuity data. This process helps to eliminate trajectory fluctuations caused by noise and discontinuity changes, making the trajectory data more stable and reliable. The smoothed trajectory data can more accurately reflect the actual flight path of the flight target and provide clearer reference information to the flight control system. By reducing irregularities and mutations in the trajectory, the aircraft can be controlled on the basis of a more stable trajectory, further improving flight safety and control accuracy.

[0051] Preferably, step S33 includes the following steps:

[0052] Step S331: Calculate the local risk flight curvature on the flight out of control probability data and the deviation trajectory smoothing data to obtain local risk flight curvature data;

[0053] Step S332: dividing the local risk flight curvature data into a training set and a test set to obtain a local risk flight curvature training set and a local risk flight curvature test set;

[0054] Step S333: constructing an initial recognition model of the flight target state for the local risk flight curvature training set based on the Q-Learning algorithm to obtain an initial recognition model of the flight target state;

[0055] Step S334: Perform model optimization test on the initial recognition model of the flight target state through the local risk flight curvature test set to obtain the flight target state recognition model.

[0056] The present invention can identify and evaluate local risk points in a flight trajectory by calculating the local risk flight curvature of the flight out of control probability data and the offset trajectory smoothing data. The local risk flight curvature data can reveal the curvature and change rate of the flight trajectory, thereby helping to identify the flight risk area. Through these data, the risk situation of the flight target in a specific area can be analyzed more accurately, providing a valuable reference for the subsequent flight target state identification. This helps to warn potential risks in advance during the flight process, thereby improving flight safety and stability. Dividing the local risk flight curvature data into a training set and a test set can provide effective data support for model training and evaluation. Through this division, the training set can be used to construct and optimize the initial recognition model of the flight target state, while the test set is used to evaluate the performance and generalization ability of the model. This data division method ensures that the model can learn comprehensive features during the training process, and can also verify its actual effect and prediction ability through the test set, thereby improving the reliability and accuracy of the model. Based on the Q-Learning algorithm, the initial recognition model of the flight target state is constructed for the local risk flight curvature training set, and an intelligent state recognition system can be established. The Q-Learning algorithm can gradually optimize the recognition ability of the model by continuously learning the local risk flight curvature data, thereby constructing an initial model that can accurately identify the flight target state. This process helps to achieve efficient recognition of the flight state, provide a scientific basis for flight control and decision-making, and improve the intelligent level of flight safety management. The performance of the model can be evaluated and improved by performing model optimization tests on the initial recognition model of the flight target state through the local risk flight curvature test set. By testing the prediction ability of the model on an independent test data set, the optimization test can discover the shortcomings of the model in practical applications and make corresponding adjustments and improvements. This process ensures that the final flight target state recognition model has good generalization ability and accuracy, so that it can more effectively perform real-time recognition and risk assessment of flight status, and improve overall flight safety.

[0057] Preferably, the present invention further provides a low-altitude flying target recognition system based on a three-dimensional digital earth, which is used to execute the low-altitude flying target recognition method based on a three-dimensional digital earth as described above, and the low-altitude flying target recognition system based on a three-dimensional digital earth comprises:

[0058] The surface element marking module is used to obtain low-altitude monitoring range data; the low-altitude monitoring range data is marked with high-altitude surface elements to obtain high-altitude surface element marking data;

[0059] The flight out-of-control probability estimation module is used to calculate the spatial relative distance of the towering surface element marking data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; and perform flight out-of-control probability estimation based on the near-surface airflow disturbance data to obtain flight out-of-control probability data;

[0060] A flight target state recognition model building module is used to build a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model;

[0061] The risk flight intention prediction module is used to predict the risk flight intention of low-altitude flying targets according to the flight target state recognition model, obtain the target risk flight intention data, and feed back the target risk flight intention data to the terminal.

[0062] The beneficial effect of the present invention is that by acquiring low-altitude monitoring range data and marking tall surface elements, tall objects on the surface that affect the aircraft can be accurately identified and recorded. This process helps to build a detailed map of the flight environment and provide important basic data for subsequent flight safety analysis. The tall surface element marking data can help identify potential obstacles in the flight path, thereby avoiding these obstacles in flight planning and operation and reducing the occurrence of flight accidents. By calculating the relative spatial distance between the flight target and the tall surface elements, the actual spatial relationship between the aircraft and the surface obstacles can be understood in detail. This process makes it possible to evaluate the relative position of the aircraft with obstacles when flying at low altitudes, and further analyze the impact of near-surface airflow disturbances. By conducting in-depth analysis of these disturbances, potential airflow interference areas can be identified, and the impact of airflow on aircraft stability can be predicted, so that corresponding control measures can be taken during flight. The probability estimation of flight out-of-control based on near-surface airflow disturbance data is of great significance. By analyzing the impact of airflow disturbances on aircraft, the risk probability of aircraft out-of-control can be quantified, thereby providing a scientific basis for flight operations. This estimation helps pilots or automatic flight systems to better deal with airflow disturbances during flight, improve flight safety, and reduce the incidence of flight accidents. The Q-Learning algorithm is used to build a flight target state recognition model, which can effectively process flight out-of-control probability data and learn risk characteristics under different flight states. This model is continuously optimized through reinforcement learning methods, enabling it to accurately identify and predict the state of the aircraft under various environmental conditions. This model can not only help understand the potential impact of different flight states on safety, but also provide information on how to adjust flight strategies under specific flight states, thereby enhancing the adaptability and safety of aircraft in complex environments. Through a dynamically updated learning mechanism, the model can continuously improve recognition accuracy and provide personalized safety measures for different flight environments. The risk flight intention prediction based on the flight target state recognition model can provide real-time warnings of potential risks for aircraft. By predicting the risk flight intention of the target, the flight system can actively adjust the flight strategy or remind the pilot to prevent flight hazards. This prediction information is transmitted through the terminal feedback mechanism to ensure that the pilot or automatic flight system can obtain the latest risk assessment data in a timely manner and make necessary safety adjustments. This can not only improve the safety of flight operations, but also enhance the ability to respond to emergencies and significantly reduce flight risks. Through real-time feedback and adjustments, a high level of safety can be maintained during the flight.Therefore, the present invention is an optimization processing of a traditional low-altitude flight target recognition method based on three-dimensional digital earth, which solves the problems of inaccurate analysis of the probability of low-altitude flight target loss of control and inaccurate judgment of its flight risk intention in the traditional low-altitude flight target recognition method based on three-dimensional digital earth, improves the accuracy of the analysis of the probability of low-altitude flight target loss of control and improves the accuracy of the judgment of flight risk intention. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 A schematic diagram of the steps of a low-altitude flying target recognition method based on a three-dimensional digital earth;

[0064] Figure 2 for Figure 1 Detailed implementation steps of step S2 in FIG.

[0065] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0066] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0067] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0068] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0069] To achieve this, please refer to Figure 1 to Figure 2, a low-altitude flying target recognition method based on three-dimensional digital earth, the method comprising the following steps:

[0070] Step S1: acquiring low-altitude monitoring range data; marking the high-altitude surface elements on the low-altitude monitoring range data to obtain high-altitude surface element marking data;

[0071] Step S2: Calculate the spatial relative distance of the towering surface element marker data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; estimate the flight out-of-control probability based on the near-surface airflow disturbance data to obtain the flight out-of-control probability data;

[0072] Step S3: constructing a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model;

[0073] Step S4: predict the risky flight intention of the low-altitude flying target according to the flight target state recognition model, obtain the target risky flight intention data, and feed back the target risky flight intention data to the terminal.

[0074] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for identifying low-altitude flying targets based on a three-dimensional digital globe according to the present invention. In this example, the method for identifying low-altitude flying targets based on a three-dimensional digital globe includes the following steps:

[0075] Step S1: acquiring low-altitude monitoring range data; marking the high-altitude surface elements on the low-altitude monitoring range data to obtain high-altitude surface element marking data;

[0076] In the embodiment of the present invention, a laser radar (LiDAR) and a synthetic aperture radar (SAR) system are used to perform detailed surface elevation measurements. These measurement data cover the height changes in low-altitude areas and form a three-dimensional digital earth model. Then, the obtained three-dimensional terrain data is analyzed by data processing software to identify all tall surface elements on the surface, such as towers, buildings or mountains. The outlines and positions of these tall surface elements are marked as tall surface element marking data for further processing.

[0077] Step S2: Calculate the spatial relative distance of the towering surface element marker data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; estimate the flight out-of-control probability based on the near-surface airflow disturbance data to obtain the flight out-of-control probability data;

[0078] In an embodiment of the present invention, the spatial relative distance calculation is performed on the marked towering surface element data. First, the current position of the flying target in the three-dimensional digital earth model is used to calculate the distance between it and all towering surface elements. The calculation is based on the straight-line distance or the surface elevation difference, and the spatial geometry algorithm is used for precise measurement. Subsequently, based on these distance data, near-surface airflow disturbance analysis is performed. Using meteorological data and aerodynamic theory, near-surface airflow disturbance data is generated by simulating the air flow around the flying target. Finally, combining the airflow disturbance data and the dynamic response characteristics of the flying target, the probability of flight loss of control is estimated by applying statistical methods to form flight loss of control probability data. All data processing and calculations are completed through dedicated analysis tools to ensure the accuracy and reliability of the results.

[0079] Step S3: constructing a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model;

[0080] In an embodiment of the present invention, a flight target state recognition model is constructed based on the Q-Learning algorithm. First, the previously obtained flight out-of-control probability data is used as the input of the Q-Learning algorithm, where the state of the flight target is represented as the state of the environment. Each state is paired with its corresponding flight out-of-control probability as the initial data for training the Q-Learning algorithm. By defining the state space, action space and reward function, the Q-Learning model is trained to optimize the strategy, thereby identifying the risks of the flight target in different states. During the training process, a method of iteratively updating the Q value table is used to gradually improve the model's ability to accurately identify the flight target state. After completing the model training, a Q-Learning model that can predict the state of the flight target is obtained, which can evaluate the flight risk according to different flight states and environmental conditions.

[0081] Step S4: predict the risky flight intention of the low-altitude flying target according to the flight target state recognition model, obtain the target risky flight intention data, and feed back the target risky flight intention data to the terminal.

[0082] In an embodiment of the present invention, a constructed flight target state recognition model is used to predict risky flight intentions. First, real-time flight target data is input into the recognition model. The model analyzes the current state of the flight target based on the input data, and predicts future flight intentions in combination with the flight environment information. The model predicts its risky flight intentions based on the state and behavior of the flight target, such as approaching obstacles or deviating from the scheduled route. The prediction results are used to generate target risk flight intention data, specifically including the predicted risk type and probability of occurrence. Finally, the target risk flight intention data is fed back to the terminal in real time through the wireless communication system for subsequent processing or alarm triggering. All prediction and feedback operations must ensure the timeliness and accuracy of data transmission.

[0083] Preferably, step S1 comprises the following steps:

[0084] Step S11: Acquire low-altitude monitoring range data;

[0085] Step S12: collecting monitoring range environment data for the low-altitude monitoring range data to obtain monitoring range environment data;

[0086] Step S13: marking the high-rise surface elements on the low-altitude monitoring range data according to the monitoring range environment data to obtain high-rise surface element marking data.

[0087] In the embodiment of the present invention, a variety of remote sensing technologies are used to collect surface elevation data of a designated area. The target area is scanned to collect high-precision three-dimensional terrain data. At the same time, synthetic aperture radar (SAR) is used to detect the surface to supplement the shortcomings of laser radar under certain conditions. All collected data are transmitted to the processing system through a data storage device to form a comprehensive low-altitude monitoring range data set. This data set contains detailed information on the height change of the surface, the distribution of objects and their detailed information in three-dimensional space, providing basic data for subsequent processing. The low-altitude monitoring range data is collected for monitoring range environmental data. First, the low-altitude monitoring range data is input into an environmental data collection system, which includes meteorological sensors and environmental monitoring equipment. The real-time meteorological information provided by these devices, such as wind speed, temperature and humidity, and other environmental factors, such as vegetation coverage and surface material type, will affect the environmental conditions of low-altitude flight. Using this information, the environmental data is combined with the terrain data through data fusion technology to form monitoring range environmental data. These environmental data cover various factors that affect low-altitude flight and provide comprehensive background information for further analysis. The low-altitude monitoring range data is marked with high-rise surface elements according to the monitoring range environmental data. First, the surface materials and environmental characteristics in the environmental data are used to identify and mark the tall surface elements in the low-altitude monitoring range data. The terrain analysis algorithm is used to extract areas with abnormal surface heights, which are usually marked as tall surface elements. The feature recognition algorithm based on environmental data is further applied to classify these tall areas, such as buildings, towers or natural terrain. Finally, the tall surface element labeling data is generated, which contains the specific location and attribute information of all identified tall surface elements, providing basic data for flight target identification and risk assessment.

[0088] Preferably, step S2 comprises the following steps:

[0089] Step S21: Real-time monitoring of low-altitude flying targets in the low-altitude monitoring range data is performed by radar to obtain real-time monitoring data of low-altitude flying targets;

[0090] Step S22: Calculate the spatial relative distance of the high-rise ground element marking data according to the real-time monitoring data of the low-altitude flying target to obtain the spatial relative distance data between the flying target and the ground element;

[0091] Step S23: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data;

[0092] Step S24: Estimate the flight out-of-control probability based on the near-surface airflow disturbance data and the real-time monitoring data of the low-altitude flying target to obtain flight out-of-control probability data.

[0093] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0094] Step S21: Real-time monitoring of low-altitude flying targets in the low-altitude monitoring range data is performed by radar to obtain real-time monitoring data of low-altitude flying targets;

[0095] In an embodiment of the present invention, a radar is used to monitor low-altitude flying targets in the low-altitude monitoring range data in real time. A radar system installed on a ground monitoring platform is used to continuously scan the low-altitude area. These radar systems can detect and track flying targets within the monitoring range in real time and obtain information such as their position, speed, and altitude. The radar system sends an electromagnetic wave signal and receives an echo signal reflected from the flying target. The spatial coordinates and dynamic parameters of the target are calculated based on the time delay and frequency change of the signal. All acquired real-time data will be recorded and stored to form real-time monitoring data of low-altitude flying targets.

[0096] Step S22: Calculate the spatial relative distance of the high-rise ground element marking data according to the real-time monitoring data of the low-altitude flying target to obtain the spatial relative distance data between the flying target and the ground element;

[0097] In an embodiment of the present invention, the spatial relative distance calculation of the towering surface element marking data is performed based on the real-time monitoring data of low-altitude flight targets. The flight target position coordinates in the real-time monitoring data of low-altitude flight targets are compared with the element position coordinates in the towering surface element marking data. In a specific implementation, the straight-line distance between each flight target and the ground object marked as a towering surface element is calculated. The distance calculation is performed through a three-dimensional coordinate system, and the spatial relative distance data between the flight target and each towering surface element is obtained by taking into account the height change of the terrain. This data will be used for subsequent airflow disturbance analysis and flight risk assessment.

[0098] Step S23: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data;

[0099] In an embodiment of the present invention, near-surface airflow disturbance analysis is performed on real-time monitoring data of low-altitude flying targets based on the spatial relative distance data of the flying target and the surface elements. First, the spatial relative distance data of the flying target and the surface elements is input into the airflow disturbance analysis model. The model uses computational fluid dynamics (CFD) technology or wind tunnel test data to simulate the disturbance of near-surface airflow, considering the influence of the relative position and relative movement between the flying target and the surrounding towering surface elements on the airflow. By analyzing the changes in the airflow field around the flying target, near-surface airflow disturbance data is generated, which reflects the influence of airflow instability caused by terrain features on the flying target.

[0100] Step S24: Estimate the flight out-of-control probability based on the near-surface airflow disturbance data and the real-time monitoring data of the low-altitude flying target to obtain flight out-of-control probability data.

[0101] In an embodiment of the present invention, the probability of flight loss of control is estimated based on the real-time monitoring data of low-altitude flying targets according to the near-surface airflow disturbance data. Combined with the near-surface airflow disturbance data, the stability of the flying target in the current airflow environment is evaluated. The specific method includes applying a flight mechanics model to quantify the impact of airflow disturbance on the flying target and calculate the risk of loss of control of the flying target under different airflow conditions. The generated loss of control probability data reflects the possibility of loss of control of the flying target in the current environment. These data will be used for risk warning and flight safety management to help take necessary flight adjustments or safety measures.

[0102] Preferably, step S23 includes the following steps:

[0103] Step S231: extracting the flight speed of the low-altitude flying target real-time monitoring data to obtain the target flight speed data;

[0104] Step S232: Decomposing the multi-directional airflow interference vector in the flight path based on the target flight speed data and the flight target-surface element space relative distance data to obtain the multi-directional airflow interference vector data;

[0105] Step S233: performing interference airflow manifold feature recognition on the multi-directional airflow interference vector data to obtain interference airflow manifold feature data;

[0106] Step S234: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target according to the interfering airflow manifold characteristic data to obtain near-surface airflow disturbance data.

[0107] In an embodiment of the present invention, the flight speed is extracted from the real-time monitoring data of low-altitude flying targets. First, the linear velocity component of the flying target in three-dimensional space is extracted through the velocity information in the real-time monitoring data of the flying target obtained by the radar system. The frequency shift effect of the radar signal is used to measure the speed of the flying target through the Doppler effect to obtain the velocity components of the target in the horizontal and vertical directions. These velocity components are integrated to calculate the overall flight speed data of the target. This data will provide necessary dynamic information for subsequent airflow interference analysis. Based on the target flight speed data and the spatial relative distance data of the flying target-surface element, the multi-directional airflow interference vector decomposition in the flight path process is performed. First, according to the speed data of the flying target and its spatial relative distance with the surface element, a flight path model is constructed. The multi-directional airflow interference encountered by the flying target in the flight path process is decomposed by the airflow interference vector decomposition method to obtain the airflow interference components in different directions. Through this method, the influence of airflow interference on the flying target can be understood in detail, and multi-directional airflow interference vector data can be generated. The multi-directional airflow interference vector data is subjected to the interference airflow manifold feature recognition. The multi-directional airflow interference vector data is processed by a fluid dynamics analysis method to identify the manifold features of the airflow disturbance. Specifically, it includes analyzing the pattern and shape of airflow disturbances, such as vortices, laminar flow or turbulence. By identifying the manifold characteristics of the interfering airflow, the manifold characteristic data of the interfering airflow is obtained, which describes the specific flow pattern of the airflow disturbance and its impact on the flight target. According to the manifold characteristic data of the interfering airflow, the near-surface airflow disturbance analysis is performed on the real-time monitoring data of the low-altitude flight target. Combined with the manifold characteristic data of the interfering airflow, the stability of the flight target in the current airflow environment is evaluated. The actual impact of the airflow manifold characteristics on the flight target is analyzed, and the degree of interference of the airflow disturbance on the flight path is calculated using the airflow disturbance model. The generated near-surface airflow disturbance data reflects the airflow instability encountered by the flight target under the current environmental conditions, providing a basis for further flight safety assessment.

[0108] Preferably, step S24 comprises the following steps:

[0109] Step S241: identifying the wind cut behavior during the flight path of the low-altitude flight target real-time monitoring data according to the near-surface airflow disturbance data to obtain wind cut behavior identification data;

[0110] Step S242: performing flight airflow load simulation calculation on the low-altitude flight target real-time monitoring data according to the wind cut behavior recognition data to obtain the target flight airflow load data;

[0111] Step S243: performing load intensity distribution analysis on the target flight airflow load data to obtain load intensity distribution data;

[0112] Step S244: performing out-of-control behavior pre-identification of the low-altitude flying target according to the load intensity distribution data and the wind cut behavior identification data to obtain out-of-control behavior pre-identification data;

[0113] Step S245: Estimate the flight out-of-control probability of the low-altitude flight target real-time monitoring data according to the out-of-control behavior pre-identification data, the load intensity distribution data, and the wind cut behavior identification data to obtain the flight out-of-control probability data.

[0114] In an embodiment of the present invention, the wind cut behavior is identified for the real-time monitoring data of low-altitude flight targets according to the near-surface airflow disturbance data, thereby obtaining wind cut behavior identification data. The specific operation includes collecting the wind speed, wind direction and airflow disturbance characteristics around the low-altitude flight path by deploying a laser radar array and a high-frequency ultrasonic airflow detection device. These data are processed by optical interferometry to identify the velocity gradient and direction difference of the sudden change in the airflow to lock the existence area of ​​the wind cut. Using data analysis and vortex feature extraction technology, the abnormal airflow pattern in the monitoring data is refined and analyzed to determine the intensity, position and diffusion range of the wind cut behavior on the target flight path. In order to improve the recognition accuracy, the real-time data cross-validation technology is used to spatially superimpose the data of multiple sensors to eliminate the interference of false alarms and missed alarms, and the output result is the wind cut behavior identification data, providing detailed wind cut features, including intensity, position and degree of influence. According to the wind cut behavior identification data, the flight airflow load simulation calculation is performed on the real-time monitoring data of the low-altitude flight target to generate the target flight airflow load data. First, combined with the characteristic information of wind cut identification, the identified wind shear area is input into the high-precision airflow load simulation system. This system uses a refined aerodynamic calculation method to calculate the airflow pressure, shear force and lateral disturbance of the flying target point by point. During the simulation process, the dynamic load imposed on the surface of the aircraft by the airflow in the wind shear area is distributed and calculated through the fluid mechanics equation and the large eddy simulation method, with special attention to the instantaneous impact of the airflow on the wings, tail and fuselage. After the analysis and calculation are completed, the load data is presented in the form of a three-dimensional spatial distribution diagram, which carefully depicts the load intensity and direction borne by each key part, providing input data for the subsequent intensity distribution analysis. The load intensity distribution analysis is performed on the target flight airflow load data to obtain the load intensity distribution data. In the specific implementation process, the dynamic load of the aircraft in the wind shear area is visualized by using high-resolution load distribution imaging technology. First, the airflow load data is sampled at high frequency, and the data is converted into a load action time series. The intensity characteristics of the load at different frequencies are analyzed by Fourier transform, and the peak area and duration of the load intensity are identified. Then, the load distribution automatic segmentation algorithm is used to mark the high-intensity load area, and the stress concentration and diffusion law of each load point are statistically counted. During the analysis process, the load distribution error is corrected by comparing the load historical data to ensure the accuracy and practicality of the data. The load intensity distribution data finally formed describes in detail the load concentration of each part and the dynamic evolution trend of the load, laying a data foundation for the pre-identification of out-of-control behavior. According to the load intensity distribution data and wind cut behavior identification data, the out-of-control behavior of low-altitude flying targets is pre-identified to obtain the pre-identification data of out-of-control behavior. The specific operation includes first jointly analyzing the load intensity distribution data and the wind cut behavior data, and using the stress-strain correspondence to evaluate the force response of the aircraft in the key disturbance area.Through stress concentration and strain deviation, the dangerous areas that cause overload or reduced controllability of the aircraft structure are identified. Using the critical stress threshold identification algorithm, the control performance of the aircraft under different load scenarios is judged point by point, and the critical points that lead to loss of control in the flight path are identified. In the identification process, special attention is paid to the slight deviation changes of the aircraft's attitude and trajectory to judge the potential risk of loss of control in advance, and the pre-identification data of the loss of control behavior is output, including the specific loss of control area, time period and triggering conditions. Through the pre-identification data of the loss of control behavior, the load intensity distribution data and the wind cut behavior identification data, the real-time monitoring data of low-altitude flight targets are used to estimate the flight loss of control probability, so as to obtain the flight loss of control probability data. First, the pre-identification data is superimposed with the real-time monitoring data, and the spatial risk mapping technology is used to align each loss of control risk point with the flight path. Then, the loss of control probability calculation formula is used to quantify the triggering conditions of each risk point, and the loss of control probability value of each position is calculated by combining key parameters such as load intensity, wind shear intensity and flight attitude deviation. Through multi-dimensional data fusion, a flight loss of control probability map is generated to evaluate the dynamic stability of the aircraft in a wind shear environment as a whole. The final flight out-of-control probability data includes a detailed probability distribution diagram, risk level classification and warning time of out-of-control occurrence, providing a decision-making basis for flight safety control.

[0115] Preferably, pre-identifying the out-of-control behavior of a low-altitude flying target based on the load intensity distribution data and the wind cut behavior identification data comprises the following steps:

[0116] Perform airflow directionality analysis on the wind cutout behavior recognition data to obtain wind cutout airflow directionality data;

[0117] According to the wind cutout airflow directionality data, the air pressure intensity of the three-dimensional cross section of the wind cutout behavior identification data is calculated to obtain the air pressure intensity data of the three-dimensional cross section of the space;

[0118] Based on the air pressure intensity data of the three-dimensional space section and the airflow directionality data of the wind cut, the maximum flight elevation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum flight elevation angle data of the airflow impact;

[0119] Based on the air pressure intensity data of the three-dimensional cross section and the airflow directionality data of the wind cut, the maximum lateral / longitudinal deviation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum lateral / longitudinal deviation angle data of the airflow impact;

[0120] According to the maximum flight elevation angle data of airflow impact and the maximum lateral / longitudinal deviation angle data of airflow impact, the out-of-control behavior pre-identification of the low-altitude flying target is performed to obtain the out-of-control behavior pre-identification data.

[0121] In the embodiment of the present invention, the directionality of the airflow is finely measured by a high-frequency vector wind speed sensor and a laser Doppler velocimeter. First, the monitored wind speed and wind direction data are time-synchronized to ensure the consistency of each data point. Then, the vector decomposition method is used to divide the airflow directionality into multiple components, and the dynamic evolution characteristics of the airflow are obtained by analyzing the dominant direction of the airflow, the speed change, and the distribution of the vortex area section by section. In this process, a detailed analysis is particularly performed on the sharp change area of ​​wind shear to identify the main directional characteristics of the wind cut behavior. The result is output as wind cut airflow directionality data, including the main flow direction, rotation direction, and flow velocity change law of the airflow. Based on the wind cut airflow directionality data, the air pressure intensity of the three-dimensional section of the wind cut behavior identification data is calculated. First, the three-dimensional space grid division technology is used to decompose the flight area into multiple sections to accurately locate the wind shear area. Subsequently, the air pressure data of each section is collected in real time through a high-precision pressure sensor network, and the air pressure intensity of each section is calculated using a numerical integration method in combination with the airflow directionality data. Especially in areas with severe wind shear, encrypted grid division and multiple measurements are used to ensure the accuracy of the data. Through section-by-section analysis, the pressure intensity distribution map of each section is generated, and all section data are integrated to obtain complete spatial three-dimensional section pressure intensity data, which provides a basis for subsequent elevation and offset angle calculations. By combining the spatial three-dimensional section pressure intensity data and the wind cut airflow directionality data, the maximum flight elevation angle of the airflow impact of the low-altitude flying target is calculated. The dynamic fluid mechanics simulation device is used to input the air pressure and directionality data to simulate the airflow impact response of the flying target in different wind shear areas. By analyzing the air pressure and shear force on each part of the windward surface of the aircraft point by point, the maximum elevation angle of the flying target under a specific airflow environment is calculated. Using a nonlinear simulation calculation method, the critical value of the elevation angle is gradually approached to ensure the accuracy of the measurement results. The output maximum flight elevation angle data of the airflow impact describes in detail the maximum safe elevation angle value that the flying target can reach under the influence of the wind cut, and identifies the boundary conditions of flight control. The maximum lateral and longitudinal offset angles of airflow impact on low-altitude flying targets are calculated using the air pressure intensity data of three-dimensional cross-sections and the airflow directionality data of wind cuts. The lateral and longitudinal force states of the aircraft are modeled and calculated through airflow intensity and directionality using the equilibrium moment analysis method. For areas with large changes in air pressure intensity, the moment calculation process is further refined to capture the offset angle of the aircraft under extreme conditions. Through iterative numerical solution, the maximum lateral and longitudinal offset angles of the aircraft in a specific airflow environment are calculated, and the output result is the maximum lateral / longitudinal offset angle data of airflow impact. The data details the critical angle value of the offset and the relevant airflow impact parameters, laying a data foundation for the pre-assessment of the risk of loss of control. Based on the maximum flight elevation angle data of airflow impact and the maximum lateral / longitudinal offset angle data of airflow impact, the out-of-control behavior of low-altitude flying targets is pre-identified.By establishing a dynamic correlation matrix between the elevation angle and the offset angle, the stability of the aircraft under different elevation angles and offset conditions is analyzed. These data are compared with the design limit parameters of the aircraft to identify the critical conditions that trigger loss of control. Through continuous risk assessment and critical state identification, the safe flight envelope of the aircraft under different wind shear conditions is determined. The output out-of-control behavior pre-identification data includes the potential trigger points, degree of danger and response strategies of the out-of-control behavior, providing key data support for the optimization of the flight control system and the early warning system.

[0122] Preferably, step S3 comprises the following steps:

[0123] Step S31: Predicting the flight target deviation trajectory based on the near-surface airflow disturbance data and the flight loss of control probability data to obtain flight target deviation trajectory prediction data;

[0124] Step S32: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data to obtain deviation trajectory smoothing data;

[0125] Step S33: constructing a flight target state recognition model based on the flight out of control probability data and the deviation trajectory smoothing data based on the Q-Learning algorithm to obtain a flight target state recognition model.

[0126] In the embodiment of the present invention, the flight target offset trajectory is estimated by a multi-dimensional space vector calculation method. First, the near-surface airflow disturbance data is matched with the out-of-control probability data, and the eddy current characteristic extraction algorithm is used to analyze the real-time impact of the airflow on the aircraft. Subsequently, the trajectory deduction formula is used to dynamically simulate the offset path of the aircraft under different airflow disturbances in combination with the speed, angle and inertial parameters of the flight target, and the coordinate sequence of continuous trajectory points is output. The offset trend and rate change of the flight target in the airflow mutation area are analyzed by high-frequency data interpolation and numerical differentiation. The resulting flight target offset trajectory estimation data includes the coordinate set, rate change characteristics and offset directionality of the target offset trajectory, and the estimated offset trajectory of the flight target in a complex airflow environment is fully displayed. The trajectory smoothing analysis is performed on the flight target offset trajectory estimation data. The weighted moving average algorithm and Kalman filtering technology are used to process the estimated trajectory to eliminate abnormal beating data and noise interference caused by airflow mutation. First, the trajectory point set is analyzed for continuity, the coherence and rationality of the trajectory are determined, and the speed and direction of the offset trajectory are adjusted using a smoothing window. Through multi-scale piecewise smoothing, the stability and accuracy of the trajectory are ensured, and the large deviation points are corrected. Subsequently, the quadratic spline interpolation method is used to fit the smooth trajectory to generate a smoothed trajectory line. The output offset trajectory smoothing data contains an optimized trajectory point sequence, which eliminates the deviation caused by random disturbances, shows the actual deviation trend of the flight target under airflow disturbances, and provides a more accurate trajectory basis for subsequent flight state identification. The flight target state recognition model is constructed by the Q-Learning algorithm. First, the flight out-of-control probability data and the offset trajectory smoothing data are input into the state space of the Q-Learning algorithm to define the state of the aircraft under different combinations of trajectories and out-of-control probabilities. According to the algorithm's reinforcement learning mechanism, the state-action pair is set, and the training is carried out with the goal of maximizing flight stability, and the response strategy under each state is gradually optimized. Through multiple rounds of simulation, the reward function is repeatedly adjusted, with particular attention paid to the identification of the out-of-control precursor state and the improvement of the response behavior. During the training process, adaptive step size control and dynamic exploration rate adjustment are used to ensure the generalization ability of the model under various airflow disturbance conditions. The final output flight target state recognition model can identify the stability state of the aircraft in different disturbance environments in real time and give corresponding adjustment suggestions. The model includes a state mapping matrix, an optimal response strategy set, and a real-time decision feedback mechanism, providing reliable data support for the identification and risk control of low-altitude flight targets.

[0127] Preferably, step S32 includes the following steps:

[0128] Step S321: performing trajectory offset time series difference calculation on the flight target offset trajectory estimation data to obtain trajectory offset time series difference data;

[0129] Step S322: performing a displacement discontinuity analysis on the flight target displacement trajectory estimation data according to the trajectory displacement time series difference data to obtain trajectory displacement discontinuity data;

[0130] Step S323: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data according to the trajectory deviation discontinuity data to obtain deviation trajectory smoothing data.

[0131] In an embodiment of the present invention, the trajectory offset time series difference calculation is performed on the flight target offset trajectory estimation data to identify the change rate and trend of the trajectory at different time nodes. First, the coordinate information of each time point is extracted from the offset trajectory estimation data, and the offset between adjacent time points is calculated using the time series difference method. The specific operation includes performing coordinate axis difference calculation on the continuous coordinate points of the trajectory, and analyzing the change characteristics of the target offset rate by calculating the displacement difference in each time interval. Subsequently, the time series difference data is standardized to ensure that the offset characteristics of different time periods are comparable, and a set of trajectory offset time series difference data including time points, offsets and rate changes is formed. This data provides a basis for subsequent trajectory discontinuity analysis, with special attention to the identification of abnormal offset behavior and mutation points. The trajectory offset time series difference data is used to perform discontinuity analysis on the flight target offset trajectory estimation data to evaluate the continuity and stability of the trajectory. First, the time series difference data is checked for smoothness to identify the mutation offset points and their time series positions. By defining a critical offset threshold, the offset points that exceed the threshold are marked as discontinuity abnormal points. Subsequently, the time window segmentation method is used to segment the trajectory data by time, and the offset consistency and fluctuation characteristics within each segment are calculated. For areas with large fluctuations or prominent intermittent offsets, in-depth analysis is performed to evaluate the impact of discontinuity on the stability of the aircraft. The output trajectory offset discontinuity data includes the location of the discontinuity point, the discontinuity amplitude and the frequency, which shows the potential discontinuity characteristics in the trajectory offset process and provides a positioning basis for subsequent smoothing. The estimated data of the flight target offset trajectory is smoothed according to the trajectory offset discontinuity data, aiming to reduce the impact of the discontinuity offset on the overall trajectory. First, for the marked discontinuous offset points, the local regression smoothing method (LOWESS) is used to correct the data and refine the discontinuous area. For the continuous area of ​​the trajectory, the weighted average method is applied to balance the trajectory trends between different points. In particular, in areas with strong discontinuity, interpolation points are added to supplement the detailed information lost during the smoothing process. The overall trajectory is quadratically fitted to eliminate irregular fluctuations and finally generate an optimized trajectory path. The output offset trajectory smoothing data includes a smoothed trajectory point set, optimized trajectory lines and smoothness evaluation indicators, providing a more stable trajectory basis for the state recognition of low-altitude flying targets.

[0132] Preferably, step S33 includes the following steps:

[0133] Step S331: Calculate the local risk flight curvature on the flight out of control probability data and the deviation trajectory smoothing data to obtain local risk flight curvature data;

[0134] Step S332: dividing the local risk flight curvature data into a training set and a test set to obtain a local risk flight curvature training set and a local risk flight curvature test set;

[0135] Step S333: constructing an initial recognition model of the flight target state for the local risk flight curvature training set based on the Q-Learning algorithm to obtain an initial recognition model of the flight target state;

[0136] Step S334: Perform model optimization test on the initial recognition model of the flight target state through the local risk flight curvature test set to obtain the flight target state recognition model.

[0137] In an embodiment of the present invention, the local risk flight curvature is calculated for the flight out of control probability data and the offset trajectory smoothing data to identify the risk level of low-altitude flight targets in different path segments. First, the second-order derivative of the offset trajectory smoothing data is calculated to extract the curvature information of each segment of the trajectory, and the local risk curvature value of the trajectory point is quantified in combination with the flight out of control probability data. During the operation, the curvature calculation involves a quantitative evaluation of the curvature of the trajectory line, and each trajectory segment is analyzed using the arc fitting method to obtain the curvature radius of each segment, and the correlation between the risk factor and the curvature is calculated. Comparative analysis shows that the larger the curvature, the higher the risk of out of control. Finally, the local risk flight curvature data is generated by combining the local curvatures and the corresponding out of control probability values. This data is used to identify the key points of the high-risk flight path and provide a reference for the subsequent model construction. The local risk flight curvature data is divided to establish a training set and a test set for model construction and verification. First, all the curvature data are arranged in chronological order, and the data are divided into a training set and a test set by random segmentation or time series segmentation. The training set is used for model construction, and the test set is used for subsequent model verification. The specific operation includes randomly shuffling the local risk flight curvature data into a training set and a test set in a ratio of 80:20, and ensuring that the data distribution in the training set and the test set is balanced. After the division, descriptive statistics are performed on the data characteristics of the training set and the test set to ensure that the two parts of the data contain enough representative risk curvature characteristics. The divided data sets provide basic support for the model construction and optimization test of the Q-Learning algorithm. The Q-Learning algorithm is used to build a model for the local risk flight curvature training set to preliminarily identify the state of low-altitude flying targets. The operation steps include: first, initialize the state space, action space and reward mechanism of Q-Learning, and set the state transition condition and reward function according to the data characteristics of the local risk flight curvature training set. Then, through algorithm iteration, learn the correlation between local curvature and flight state, and update the Q value matrix. In each training round, the algorithm simulates the decision-making of the input curvature data, adjusts the recognition strategy, and outputs the best state recognition strategy until the Q value converges. During the training process, the key features of the trajectory data, such as the curvature mutation point and the risk of loss of control, are used as input to continuously optimize the recognition accuracy of the Q-Learning algorithm. The final initial recognition model of the flight target state provides a preliminary analysis framework for the accurate determination of the flight state. The initial recognition model of the flight target state is optimized and tested using the local risk flight curvature test set to verify and adjust the recognition ability of the model. First, the data in the test set is input into the initial recognition model, and the recognition results of the model are evaluated for accuracy, recall rate, and false positive rate to quantify the recognition effect of the model under different risk curvatures.In response to the errors found during the test, the model was adjusted by adjusting the model parameters, optimizing the reward mechanism, and incrementally updating the Q value table. During the test, special attention was paid to the recognition accuracy of the high curvature area, and the model's responsiveness to extreme deviations and out-of-control conditions was improved by refining the Q value update strategy. After completing the test and optimization, the output flight target state recognition model has higher recognition accuracy and robustness, ensuring that the dynamic state recognition of low-altitude flying targets achieves the expected effect.

[0138] Preferably, the present invention further provides a low-altitude flying target recognition system based on a three-dimensional digital earth, which is used to execute the low-altitude flying target recognition method based on a three-dimensional digital earth as described above, and the low-altitude flying target recognition system based on a three-dimensional digital earth comprises:

[0139] The surface element marking module is used to obtain low-altitude monitoring range data; the low-altitude monitoring range data is marked with high-altitude surface elements to obtain high-altitude surface element marking data;

[0140] The flight out-of-control probability estimation module is used to calculate the spatial relative distance of the towering surface element marking data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; and perform flight out-of-control probability estimation based on the near-surface airflow disturbance data to obtain flight out-of-control probability data;

[0141] A flight target state recognition model building module is used to build a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model;

[0142] The risk flight intention prediction module is used to predict the risk flight intention of low-altitude flying targets according to the flight target state recognition model, obtain the target risk flight intention data, and feed back the target risk flight intention data to the terminal.

[0143] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0144] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for identifying low-altitude flying targets based on three-dimensional digital earth, characterized in that: The following steps are involved: Step S1: acquiring low-altitude monitoring range data; marking the high-rise surface elements on the low-altitude monitoring range data to obtain high-rise surface element marking data; Step S2: Calculate the spatial relative distance of the towering surface element marker data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; estimate the flight out-of-control probability based on the near-surface airflow disturbance data to obtain the flight out-of-control probability data; Step S3: constructing a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model; Step S4: predict the risky flight intention of the low-altitude flying target according to the flight target state recognition model, obtain the target risky flight intention data, and feed back the target risky flight intention data to the terminal.

2. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 1 is characterized in that: Step S1 includes the following steps: Step S11: Acquire low-altitude monitoring range data; Step S12: collecting monitoring range environment data for the low-altitude monitoring range data to obtain monitoring range environment data; Step S13: marking the high-rise surface elements on the low-altitude monitoring range data according to the monitoring range environment data to obtain high-rise surface element marking data.

3. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 1 is characterized in that: Step S2 includes the following steps: Step S21: Real-time monitoring of low-altitude flying targets in the low-altitude monitoring range data is performed by radar to obtain real-time monitoring data of low-altitude flying targets; Step S22: Calculate the spatial relative distance of the high-rise ground element marking data according to the real-time monitoring data of the low-altitude flying target to obtain the spatial relative distance data between the flying target and the ground element; Step S23: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; Step S24: Estimate the flight out-of-control probability based on the near-surface airflow disturbance data and the real-time monitoring data of the low-altitude flight target to obtain flight out-of-control probability data.

4. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 3 is characterized in that: Step S23 includes the following steps: Step S231: extracting the flight speed of the low-altitude flying target real-time monitoring data to obtain the target flight speed data; Step S232: Decomposing the multi-directional airflow interference vector in the flight path based on the target flight speed data and the flight target-surface element space relative distance data to obtain the multi-directional airflow interference vector data; Step S233: performing interference airflow manifold feature recognition on the multi-directional airflow interference vector data to obtain interference airflow manifold feature data; Step S234: performing near-surface airflow disturbance analysis on the real-time monitoring data of the low-altitude flying target according to the interfering airflow manifold characteristic data to obtain near-surface airflow disturbance data.

5. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 3 is characterized in that: Step S24 includes the following steps: Step S241: identifying the wind cut behavior during the flight path of the low-altitude flight target real-time monitoring data according to the near-surface airflow disturbance data to obtain wind cut behavior identification data; Step S242: performing flight airflow load simulation calculation on the low-altitude flight target real-time monitoring data according to the wind cut behavior recognition data to obtain the target flight airflow load data; Step S243: performing load intensity distribution analysis on the target flight airflow load data to obtain load intensity distribution data; Step S244: performing out-of-control behavior pre-identification of the low-altitude flying target according to the load intensity distribution data and the wind cut behavior identification data to obtain out-of-control behavior pre-identification data; Step S245: Estimate the flight out-of-control probability of the low-altitude flight target real-time monitoring data according to the out-of-control behavior pre-identification data, the load intensity distribution data, and the wind cut behavior identification data to obtain the flight out-of-control probability data.

6. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 4 is characterized in that: The pre-identification of the out-of-control behavior of low-altitude flying targets based on the load intensity distribution data and wind cut behavior identification data includes the following steps: Perform airflow directionality analysis on the wind cutout behavior recognition data to obtain wind cutout airflow directionality data; According to the wind cutout airflow directionality data, the air pressure intensity of the three-dimensional cross section of the wind cutout behavior identification data is calculated to obtain the air pressure intensity data of the three-dimensional cross section of the space; Based on the air pressure intensity data of the three-dimensional space section and the airflow directionality data of the wind cut, the maximum flight elevation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum flight elevation angle data of the airflow impact; Based on the air pressure intensity data of the three-dimensional cross section and the airflow directionality data of the wind cut, the maximum lateral / longitudinal deviation angle of the airflow impact of the low-altitude flying target is calculated to obtain the maximum lateral / longitudinal deviation angle data of the airflow impact; According to the maximum flight elevation angle data of airflow impact and the maximum lateral / longitudinal deviation angle data of airflow impact, the out-of-control behavior pre-identification of the low-altitude flying target is performed to obtain the out-of-control behavior pre-identification data.

7. The low-altitude flying target recognition method based on three-dimensional digital earth according to claim 1 is characterized in that: Step S3 includes the following steps: Step S31: Predicting the flight target deviation trajectory based on the near-surface airflow disturbance data and the flight loss of control probability data to obtain flight target deviation trajectory prediction data; Step S32: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data to obtain deviation trajectory smoothing data; Step S33: constructing a flight target state recognition model based on the flight out of control probability data and the deviation trajectory smoothing data based on the Q-Learning algorithm to obtain a flight target state recognition model.

8. The method for identifying low-altitude flying targets based on three-dimensional digital earth according to claim 7 is characterized in that: Step S32 includes the following steps: Step S321: performing trajectory offset time series difference calculation on the flight target offset trajectory estimation data to obtain trajectory offset time series difference data; Step S322: performing a displacement discontinuity analysis on the flight target displacement trajectory estimation data according to the trajectory displacement time series difference data to obtain trajectory displacement discontinuity data; Step S323: performing trajectory smoothing analysis on the flight target deviation trajectory estimation data according to the trajectory deviation discontinuity data to obtain deviation trajectory smoothing data.

9. The method for identifying low-altitude flying targets based on three-dimensional digital earth according to claim 7, characterized in that: Step S33 includes the following steps: Step S331: Calculate the local risk flight curvature on the flight out of control probability data and the deviation trajectory smoothing data to obtain local risk flight curvature data; Step S332: dividing the local risk flight curvature data into a training set and a test set to obtain a local risk flight curvature training set and a local risk flight curvature test set; Step S333: constructing an initial recognition model of the flight target state for the local risk flight curvature training set based on the Q-Learning algorithm to obtain an initial recognition model of the flight target state; Step S334: Perform model optimization test on the initial recognition model of the flight target state through the local risk flight curvature test set to obtain the flight target state recognition model.

10. A low-altitude flying target recognition system based on three-dimensional digital earth, characterized in that: For executing the low-altitude flying target recognition method based on three-dimensional digital earth as claimed in claim 1, the low-altitude flying target recognition system based on three-dimensional digital earth comprises: The surface element marking module is used to obtain low-altitude monitoring range data; the low-altitude monitoring range data is marked with high-altitude surface elements to obtain high-altitude surface element marking data; The flight out-of-control probability estimation module is used to calculate the spatial relative distance of the towering surface element marking data to obtain the flight target-surface element spatial relative distance data; perform near-surface airflow disturbance analysis based on the flight target-surface element spatial relative distance data to obtain near-surface airflow disturbance data; and perform flight out-of-control probability estimation based on the near-surface airflow disturbance data to obtain flight out-of-control probability data; A flight target state recognition model building module is used to build a flight target state recognition model based on the flight out of control probability data based on the Q-Learning algorithm to obtain a flight target state recognition model; The risk flight intention prediction module is used to predict the risk flight intention of low-altitude flying targets according to the flight target state recognition model, obtain the target risk flight intention data, and feed back the target risk flight intention data to the terminal.

Images