Air target abnormal behavior monitoring method and device and storage medium

By acquiring and preprocessing the spatiotemporal information and environmental characteristics of the aerial target, and combining the abnormal behavior of the aerial targets to monitor abnormal behaviors, the problem of inaccurate judgment caused by environmental factors in the existing technology is solved, and the accuracy and safety of monitoring are improved. prevents sensitive targets from being circumvented.

CN118916808AInactive Publication Date: 2024-11-08SICHUAN CREATIVE TECH CO LTD

Patent Information

Application Number
CN202410934280.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing aerial target abnormal behavior monitoring methods do not consider environmental factors, resulting in inaccurate judgment of sensitive targets, and the judgment rules based on static parameters are easily evaded by criminals, and the monitoring is not accurate and reliable enough.

Method used

By obtaining spatiotemporal information and basic attribute information, preprocessing and extracting environmental features, combining anomaly analysis models for abnormal detection, including analysis of numerical abnormalities, action abnormalities, distance abnormalities, out-of-bounds abnormalities and combination abnormalities, improving the accuracy and safety of monitoring sex.

Benefits of technology

Improve the accuracy and reliability of aerial target abnormal behavior monitoring, reduce misjudgment and misreport, enhance the monitoring ability of sensitive targets, prevent the possibility of evading monitoring by using simple judgment rules, and improve the comprehensive performance of the monitoring system. and security.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916808A_ABST
    Figure CN118916808A_ABST
Patent Text Reader

Abstract

The invention discloses an air target abnormal behavior monitoring method and device and a storage medium, and the method comprises the following steps: obtaining to-be-processed data which comprises spatio-temporal information and basic attribute information; preprocessing the to-be-processed data to obtain preprocessed data; calling an environment information base, matching environment information according to the spatio-temporal information, and extracting environment features; inputting the pre-processed data and the environment characteristics by using an abnormal analysis model so as to carry out abnormal behavior detection; and finally, when an abnormal detection result is confirmed as an abnormal behavior, sending out an abnormal early warning prompt. Spatio-temporal information, environmental characteristics and basic attribute information are considered, the accuracy and reliability of the monitoring system are effectively improved, and an effective means is provided for timely identification and processing of abnormal behaviors.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of aviation, and in particular to a method, device and storage medium for monitoring abnormal behavior of aerial targets. Background Art

[0002] With the development of modern aviation technology, the monitoring of abnormal behaviors of aerial targets (such as yaw, circling, dive, etc.) has become increasingly important. Although many countries have invested a lot of manpower and material resources in analyzing the behaviors of aerial targets, there is still no reliable solution.

[0003] The existing monitoring of abnormal behavior of aerial targets usually adopts the method of trajectory analysis. Due to limited monitoring and management resources, only relatively strict abnormal behavior judgment of targets is performed through target trajectories. Therefore, the analysis of target trajectories is particularly important.

[0004] The existing target trajectory analysis method is to monitor the behavior trajectory of aerial targets, calibrate sensitive targets based on the comprehensive assessment of their behavior trajectory characteristics and abnormal attribute characteristics, and then assign and update the sensitivity of the targets based on their real-time behavior trajectories, and determine them as key targets when the sensitivity exceeds the threshold. However, when determining sensitive targets, the impact of environmental factors on target behavior is not considered, resulting in inaccurate judgment of sensitive targets. At the same time, the existing sensitive target judgment rules based on static parameters are relatively simple, and criminals may use the sensitive target judgment rules to circumvent these judgment indicators to hide the real sensitive targets within the normal rule range, thereby greatly reducing the probability of being monitored. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a method, device and storage medium for monitoring abnormal behavior of aerial targets, so as to solve the limitations of abnormal behavior monitoring of aerial targets and inaccurate judgment of sensitive targets, and prevent sensitive targets from evading monitoring.

[0006] A method for monitoring abnormal behavior of an aerial target, characterized in that the method comprises the following steps:

[0007] S1, obtaining data to be processed, wherein the data to be processed includes spatiotemporal information and basic attribute information;

[0008] S2, preprocessing the data to be processed to obtain preprocessed data;

[0009] S3, calling the environmental information library, matching the environmental information according to the preprocessed data, and extracting environmental features from the environmental information;

[0010] S4, calling the anomaly analysis model, inputting the preprocessed data and environmental features into the anomaly analysis model, thereby performing anomaly detection on the abnormal behavior;

[0011] S5, when the abnormal detection result is an abnormal behavior, an abnormal warning prompt is issued for the abnormal detection result.

[0012] Furthermore, the step S2 includes the following steps:

[0013] S21, removing outliers from the data to be processed;

[0014] S22, performing numerical anomaly detection and filtering correction on the data to be processed after outliers are removed to obtain preprocessed data;

[0015] Furthermore, the anomaly analysis model includes six types of processing methods: numerical anomaly, action anomaly, distance anomaly, cross-boundary anomaly, route anomaly analysis and combined anomaly.

[0016] Furthermore, the numerical anomaly includes clustering algorithm, tree model, classification algorithm and dimensionality reduction algorithm.

[0017] Furthermore, the action anomaly analysis calculates the minimum turning radius of the target and matches the target's route path using a similar time series matching algorithm to identify specific actions, where the minimum turning radius is calculated as follows: g is the gravitational acceleration constant, R min is the minimum turning radius, v is the speed, Φ max is the maximum slope angle, and the calculation formula of the similar time series matching algorithm is as follows: Among them, φ(T i ∩E j ) represents the track T i With track E j The total length of the intersecting subpaths, φ(T i ∪E j ) represents the track T i With track E j The total path length.

[0018] Furthermore, the distance anomaly is calculated and determined by the following formula: r is the radius of the earth, and the distance d between two points is determined by calculation AB This can determine whether there is a distance anomaly.

[0019] Furthermore, the out-of-bounds exception is determined by the following formula: Where (x0, y0) is the coordinate of the center of the circle, α is the rotation angle of the circle, and b is the radius of the circle.

[0020] Furthermore, the combined anomaly is to combine the numerical anomaly, the action anomaly, the distance anomaly and the cross-boundary anomaly to perform a comprehensive anomaly analysis.

[0021] In addition, the present invention discloses a device for monitoring abnormal behavior of aerial targets, the device comprising:

[0022] An acquisition unit, used for acquiring data to be processed, wherein the data to be processed includes spatiotemporal information and basic attribute information;

[0023] A preprocessing unit, used for preprocessing the data to be processed to obtain preprocessed data;

[0024] An environmental feature extraction unit is used to call an environmental information library, match environmental information according to the preprocessed data, and extract environmental features from the environmental information;

[0025] The anomaly analysis unit is used to call the anomaly analysis model and input the preprocessed data and environmental characteristics into the anomaly analysis model, so as to perform anomaly detection on abnormal behaviors;

[0026] The abnormal warning unit is used to issue an abnormal warning prompt for the abnormal detection result when the abnormal detection result is an abnormal behavior.

[0027] In addition, the present invention discloses a computer-readable storage medium, which stores one or more instructions, and the one or more instructions are suitable for being loaded by a processor and executing the method for monitoring abnormal behavior of aerial targets as described in any one of claims 1 to 8.

[0028] The present application provides a method, device and storage medium for monitoring abnormal behavior of aerial targets, which improves the accuracy and reliability of monitoring abnormal behavior of aerial targets, reduces the possibility of misjudgment and omission, and makes up for the limitations of existing monitoring methods. It can more comprehensively consider the impact of environmental factors on target behavior, improve the comprehensive performance of the monitoring system, and at the same time enhance the monitoring capability of sensitive targets, prevent the possibility of sensitive targets evading monitoring by using simple judgment rules, and improve the safety and practicality of the monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic diagram of the device structure of the hardware operating environment involved in the implementation scheme of the present invention;

[0030] Figure 2 A schematic diagram of a flow chart of an implementation of the method for monitoring abnormal behavior of aerial targets of the present invention;

[0031] Figure 3 It is a schematic diagram of the structure of a device for monitoring abnormal behavior of aerial targets in an implementation scheme of the present invention. DETAILED DESCRIPTION

[0032] The following will be combined with the accompanying drawings of the present invention to clearly and completely describe the simulation technology route in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] With the rapid development of modern aviation technology, the monitoring of abnormal behavior of aerial targets has become increasingly important. Although countries have invested a lot of resources in behavior analysis, existing monitoring methods still face many challenges. At present, trajectory analysis is usually used for monitoring, but due to the limitation of monitoring and management resources, behavior judgment based solely on target trajectory is not accurate enough. Therefore, a method for monitoring abnormal behavior of aerial targets is proposed to address the limitations of existing monitoring methods. Please refer to the attached Figure 2 As shown, the method comprises the following steps:

[0035] Step S1, obtaining data to be processed, wherein the data to be processed includes spatiotemporal information and basic attribute information;

[0036] Specifically, first the system needs to collect data on aerial targets. In one or more embodiments, these data come from sensors such as radars and satellites, and are used to monitor and track the position and status of aerial targets. Extract spatiotemporal information from the acquired data, which includes the location time, longitude, latitude, altitude, heading, speed, altitude difference, and speed difference elements of the target track point. This information reflects the changes in the target in time and space and is dynamic time series data. At the same time, extract basic attribute information from the acquired data, which includes elements such as target track number, target type, target model, target country, target military and civilian nature, affiliated military branch, departure point, destination, etc. This information describes the basic attributes of the target, which generally will not change significantly and is relatively fixed target information. Integrate the sorted spatiotemporal information and basic information into data to be processed.

[0037] Step S2, preprocessing the data to be processed to obtain preprocessed data;

[0038] In step S2, the data to be processed is first preprocessed to obtain preprocessed data. This step includes two steps: outlier removal and numerical anomaly detection and filtering correction, so as to improve data quality and reduce the interference of data anomalies on the judgment of abnormal target behavior. Through outlier removal, numerical anomaly detection and filtering correction, the system can clean up and correct the noise and abnormal points in the data, making subsequent data analysis and anomaly detection more accurate and reliable. This step lays a good data foundation for the subsequent reasoning detection of abnormal target behavior and improves the reliability and practicality of the system. The specific steps are as follows:

[0039] Step S21, removing outliers from the data to be processed;

[0040] Specifically, in the step of eliminating wild values, it is necessary to delete the track information that exceeds the value range of the data element. This involves range detection of data such as position, speed, and altitude, and eliminating data that exceeds the range. For position information (longitude, latitude), it is limited to the range of the earth's surface. The value range of longitude is [-180,180] degrees, and the value range of latitude is [-90,90] degrees. Therefore, conditional statements are used to filter and delete data that exceeds the range. For heading information, the value range of heading is [0,360] degrees. For speed and altitude information, the specific value range needs to be set according to actual conditions. In one or more embodiments, the speed is limited to between [0,3000] km / h, and the altitude is limited to between [0,20000] m.

[0041] Step S22, performing numerical anomaly detection and filtering correction on the data to be processed after outliers are eliminated to obtain preprocessed data;

[0042] Specifically, in this step, a filtering and smoothing method is used to detect and correct numerical anomalies to reduce noise and abnormal points in the data. In one or more embodiments, this step is processed by using a Hampel filtering method. Specifically, the proportional factor W is calculated. i , if W i =(x i -m i ) / σ i >n σ , then the data point is considered to be an outlier and needs to be corrected. Among them, mi is the local median of the i-th data point, σ i is the corresponding standard deviation, κ is the correction factor, k = 1.4826, n σ is the threshold of the outlier point. The specific calculation formula is as follows:

[0043] m i =median(x i-k ,....,x i+k ),

[0044] σ i =k*median(x i-k -m i |,...,|x i+k -m i |),

[0045]

[0046] If W i >n σ Then x i It is considered an outlier and needs to be corrected. The correction method is to use the local median m i Replace the value of the outlier. Hampel filtering uses the median and standard deviation of the data to identify outliers. By calculating the median and standard deviation of local data points and defining the scaling factor W i To determine whether a data point is abnormal. If the deviation of a data point exceeds a certain threshold (for example, n σ ), it is considered as an outlier and needs to be corrected. The correction method is to replace the value of the outlier with the local median to eliminate the impact of the outlier on data analysis.

[0047] In summary, this step is to improve the quality and accuracy of data by cleaning and correcting the noise and anomalies in the data through preprocessing operations. Such a preprocessing process provides a reliable data basis for subsequent target abnormal behavior analysis and detection.

[0048] Step S3, calling the environmental information library, matching the environmental information according to the preprocessed data, and extracting the environmental features in the environmental information;

[0049] Specifically, the environmental information library is called, and the environmental information library is a data storage and management system specifically used to store various environmental data, including meteorological information, topographic information, human activity area information, bird migration path information, electromagnetic environment information and location information. The system matches the location of the target with the location information in the environmental information library according to the location information in the preprocessed data. Through this matching process, the system can determine the specific environmental information corresponding to the location of the target. When processing different types of environmental information, the system extracts corresponding environmental features according to the nature of the information. When the environmental information is meteorological information, the system obtains meteorological features such as atmospheric pressure, wind speed, wind direction, temperature, humidity, precipitation, etc. from the location of the target. These features have an important impact on the flight behavior of the aircraft. For example, wind speed and wind direction will affect the heading and speed of the aircraft, while temperature and humidity may affect the performance and stability of the aircraft. Secondly, if the environmental information includes topographic information, the system will extract the topographic features of the target location, such as mountains, rivers, lakes, and forests. These topographic features have an important impact on the flight path and attitude planning of the aircraft, because the ups and downs of the terrain and the changes in the topography will affect the route selection and flight altitude of the aircraft. In addition, when the environmental information includes human activity area information, the system will extract the characteristics of the human activity area around the target location, such as cities, farmlands, industrial areas, etc. These human activity area characteristics can help the system understand the environmental background and constraints of the target, such as whether the target is flying in a busy urban area, or whether it is flying over an industrial area. In addition to the aforementioned meteorological information, topographic information and human activity area information, the system will also extract corresponding environmental features based on other types of environmental information. For example, if the environmental information includes bird migration path information, the system will extract relevant features such as bird migration paths that need to be considered at the target location. These features help the system understand the possible bird migration routes around the target to avoid flight behaviors that conflict with bird migration. In addition, if the environmental information includes electromagnetic environment information, the system will extract characteristics such as electromagnetic distribution at the target location. These features have an important impact on the communication and navigation systems of the aircraft, because electromagnetic interference may affect the signal reception and transmission of the aircraft, and even cause communication interruption or navigation deviation. Finally, if the environmental information includes location information, the system will extract specific location features corresponding to the target location, such as airport airspace, air defense identification zone, cruise area, etc. These location features can help the system more comprehensively understand the environmental context and constraints of the target, such as whether the target has entered a restricted flight area or a restricted flight area, and whether there are possible flight obstacles or conflict areas around the target.

[0050] Step S4, calling the anomaly analysis model, inputting the preprocessed data and environmental features into the anomaly analysis model, thereby performing anomaly detection on the abnormal behavior;

[0051] In step S4, the anomaly analysis model is called and the preprocessed data and the extracted environmental features are input into the model to detect abnormal behavior. This means that the system passes the state data of the target and the characteristics of its environment to the anomaly analysis model, allowing the model to conduct comprehensive analysis and processing.

[0052] Specifically, the anomaly analysis model includes five types of processing methods: numerical anomaly, action anomaly, distance anomaly, out-of-bounds anomaly, and combined anomaly. It includes the feature data required for these types of abnormal behavior recognition and detection; the specific target abnormal behavior analysis methods are as follows;

[0053]

[0054] Numerical anomaly detection is anomaly detection for numerical variables, and its purpose is to identify abnormal values ​​in the data that are obviously inconsistent with the expected pattern. This process involves developing, testing, and packaging general statistical anomaly detection algorithms, and publishing these algorithms as atomic components to the target behavioral anomaly analysis model.

[0055] In numerical anomaly detection, an adaptive anomaly detection algorithm can be selected according to the needs of the actual scenario and the statistical characteristics of the data, and the final anomaly judgment criteria can be adjusted. Specifically, in one or more embodiments, the anomaly detection algorithm includes a clustering algorithm (such as KNN and DBSCAN), a tree model (such as isolation forest), a classification algorithm (such as SVM) and a dimensionality reduction algorithm (such as PCA). These algorithms can effectively identify outliers in numerical variables through different principles and methods. Therefore, a suitable algorithm can be selected according to the specific situation, and the parameters can be adjusted according to the characteristics of the data set to achieve accurate detection and identification of target anomalies. The numerical anomaly detection algorithm is as follows:

[0056]

[0057] It is worth mentioning that in terms of application scenarios, clustering algorithms are often used to detect outlier values ​​that cannot be clustered. For example, when there are sample points with obvious irregular distribution or far away from the cluster center in the data set, the clustering algorithm can better identify these outliers. By clustering the data, the clustering algorithm can help users find potential anomalies in the data and provide effective support for anomaly detection; and the tree model described in the table is usually used to identify anomalies in the case where the anomaly index is greater than the threshold in the hyperplane. For example, when faced with a data set in which some sample points show obvious separation or abnormal distribution in the feature space, the tree model can quickly separate these anomalies. Specifically, when constructing multiple independent random trees, the isolation forest can quickly identify those anomalies that are different from the normal pattern. By recursively dividing the data set into left and right subtrees and calculating the path length of each sample point in the tree, the isolation forest can effectively identify anomalies. In the branching process of the tree, anomalies are usually separated faster than normal points because they are often sparser and farther away from other points. Therefore, the tree model can better identify these anomalies when there are obvious abnormal distributions or separations in the data set. By identifying these anomalies, the tree model provides effective support for anomaly detection and helps users discover potential anomalies in the data; and the SVM (support vector machine) described in the above table is an algorithm suitable for detecting external anomalies in hyperplane classification. In this scenario, SVM is used to find the best hyperplane in the data set, divide the data into two categories, and identify anomalies from them. The core idea of ​​SVM is to classify by finding a hyperplane that maximizes the interval between categories. Specifically, SVM tries to find a separating hyperplane that maximizes the distance from the sample point closest to the hyperplane (called support vector) to the hyperplane. Such a hyperplane can form an optimal dividing line between the two categories and has good generalization ability for new sample points. In terms of anomaly detection, SVM can identify anomalies that do not conform to the normal category distribution by dividing the hyperplane. Specifically, anomalies are usually located in a distant area of ​​the data set and are far away from the hyperplane, so they may become outliers. By finding the best hyperplane, SVM can effectively separate these abnormal points from normal points and identify them as abnormal points; finally, in the application scenario, PCA (Principal Component Analysis) is suitable for eigenvalue anomaly detection after reconstructing the main features of the data. Specifically, PCA maps the original data into a low-dimensional space through linear transformation, and then identifies the outliers based on the variance of the principal components. First, PCA calculates the covariance matrix of the original data and finds a set of new orthogonal bases, namely the principal components, so that the data has the largest variance in the new coordinate system. In this way, the original high-dimensional data is mapped to a space with a lower dimension while retaining the data information as much as possible.Next, PCA performs eigenvalue anomaly detection by selecting the parts with larger variance in the principal components. Outliers usually cause larger variance in the data, so they may have larger eigenvalues ​​in the principal components. By identifying the parts with larger eigenvalues ​​in the principal components, PCA can effectively identify outliers in the data. Finally, based on the size of the eigenvalues ​​in the principal components, PCA can classify data points into normal values ​​and outliers. Typically, parts with larger eigenvalues ​​in the principal components may be considered outliers because they contribute more to the variability of the data and may indicate abnormal patterns or abnormal behavior in the data.

[0058] In general, numerical anomaly processing in anomaly analysis models includes clustering algorithms, tree models, classification algorithms, and dimensionality reduction algorithms. These algorithms and methods can detect anomalies for numerical variables and identify anomalies in data sets through different data processing and model training. For example, clustering algorithms identify outliers that are far from the cluster center by clustering data; tree models such as isolation forests quickly and effectively identify anomalies by constructing random forests and calculating path lengths; classification algorithms such as SVM can find the best hyperplane in the data set to divide the data into two categories and identify anomalies from them; dimensionality reduction algorithms such as PCA map the original data to a low-dimensional space through linear transformation, and then identify outliers based on the variance of the principal components. These algorithms and methods work together in anomaly analysis models to provide a variety of anomaly detection options to deal with numerical anomalies in different scenarios.

[0059] Motion anomaly analysis aims to analyze specific actions performed by a target within a specific time period. For example, when a target performs actions such as circling or turning back, these actions may indicate abnormal behavior or intentions of the target. In this process, the turning radius is considered an important parameter because it reflects the motion characteristics of the target when performing specific actions. Specifically, the minimum turning radius (R min ) is defined as the target at a given speed (v) and maximum slope angle (Φ max ) is the minimum turning radius that can be achieved under the condition of . Its calculation formula is: Here, g is the gravitational acceleration constant.

[0060] In the motion anomaly analysis, in one or more embodiments, a similar time series matching algorithm is used to perform rule matching on the target's flight path. For example, assuming that a user has defined a set of templates E representing the circling trajectory in advance. i ={s1,s2,....,s n}, we can determine whether the target is performing a hovering action by calculating the similarity between the current track and the template. The similarity calculation formula is as follows: Among them, φ(Ti ∩E j ) represents the track T i With track E j The total length of the intersecting subpaths, φ(T i ∪E j ) represents the track T i With track E j By calculating the similarity between the current track and the template track, we can know whether the current track is similar to the circling track. If the similarity is higher than the preset threshold, it can be determined that the target is performing a circling action. This method based on similarity calculation can effectively detect abnormal behavior of the target's action and provide a reliable analysis method.

[0061] In general, the motion anomaly processing in the anomaly analysis model includes two key steps: first, calculating the minimum turning radius of the target to reflect its motion characteristics under specific speed and slope conditions; second, using the similar time series matching algorithm to match the target's flight path to identify specific actions, such as circling, turning back, etc. Through these steps, motion anomaly analysis can effectively detect and identify specific actions performed by the target within a specific period, providing a reliable analysis method for the discovery of abnormal behavior.

[0062] Distance anomaly analysis is an important part of the anomaly analysis model. It identifies possible anomalies by calculating the distance between two points in geographic space. In this dimension of processing, in one or more embodiments, the distance calculation formula is the Haversine formula. This formula is used to calculate the spherical arc distance between two points on the earth's surface. Its basic principle is to calculate the shortest distance between two points on the sphere based on the longitude and latitude coordinates. Specifically, two points A and B are known, and their longitude and latitude are respectively and The distance d between the two points can be calculated using the Haversine formula AB , the formula is as follows: Where r is the radius of the earth. By considering the earth as a sphere, the arc distance between two points can be calculated using the difference in longitude and latitude based on the trigonometric relationship on the sphere. This can accurately reflect the actual distance between two points on the earth's surface, thus providing a reliable mathematical basis for the detection of distance anomalies.

[0063] In general, distance anomaly analysis uses the Haversine formula to calculate the spherical arc distance between two points on the earth's surface in the anomaly analysis model to identify possible anomalies. This formula calculates the shortest distance between two points on a sphere based on longitude and latitude coordinates, using the principles of spherical trigonometry. By calculating the distance between two points in geographic space, possible anomalies can be detected, such as the target moving an unreasonable distance in a short period of time or an abnormal trajectory path. Distance anomaly analysis provides an important mathematical basis and analysis method for the identification of abnormal behavior.

[0064] Out-of-bounds anomaly analysis is an important component of the anomaly analysis model, which is used to determine whether the target location point is located in a specified area. In one or more embodiments, it is implemented by ray method, vector cross multiplication judgment method and area sum judgment method. When judging whether the target location point is within a circular or elliptical area, the following calculation formula is used: Among them, (x0, y0) is the coordinate of the center of the circle, α is the rotation angle of the circle, and b is the radius of the circle. This calculation formula is based on the equation of the ellipse and is used to determine whether a given point (x, y) is inside the ellipse. Using the distance formula from a point to an ellipse, if the distance is less than the radius of the ellipse, the point is inside the ellipse; otherwise, it is outside the ellipse. This method uses the properties of the ellipse in geometry to determine whether the target location point is within the specified area, providing an effective tool for anomaly analysis.

[0065] In general, out-of-bounds anomaly analysis is an important component of the anomaly analysis model, which is used to determine whether the target location point is within the specified area. In one or more embodiments, the processing method includes a ray method, a vector cross product discrimination method, or an area sum discrimination method. In specific cases, for example, it is determined whether the target location point is within a circular or elliptical area. Through these methods, the anomaly analysis model can effectively identify whether the target location point is out of bounds, providing an important basis and analysis means for the detection of abnormal behavior.

[0066] Route anomaly analysis is a method used to determine whether the target flight route is in compliance with the prescribed route. In aviation monitoring, the route specifies a specific path that the aircraft can safely fly, and a route anomaly may indicate that the aircraft has deviated from these safe paths. In route anomaly analysis, the location and shape of the prescribed route are first determined based on environmental features. These environmental features are extracted from various types of environmental information and are extracted accordingly based on the nature of the information. For example, when the system obtains meteorological information, meteorological features such as atmospheric pressure, wind speed, wind direction, temperature, humidity, and precipitation can be obtained. These meteorological features have an important impact on the flight of the aircraft, such as wind direction and wind speed can affect the heading of the aircraft. In addition, relevant features are extracted based on topographic information, such as mountains, rivers, lakes, forests, etc. The ups and downs of the terrain and changes in the topography can affect the route selection and flight altitude of the aircraft. In addition, human activity area information such as cities, farmlands, industrial areas, etc. is extracted to obtain the characteristics of the human activity areas around the target location. These human activity area features help the system understand the environmental context and constraints of the target, such as whether the target is flying in a busy urban area or flying over an industrial area. In order to detect route anomalies, in one or more embodiments, the Euclidean distance algorithm is used for analysis. Route anomaly analysis uses the Euclidean distance algorithm to calculate the straight-line distance between two points. In this case, the Euclidean distance is used to measure the distance between a point on the target flight route and a point on the specified route. The calculation formula of the Euclidean distance is as follows: (x1, y1, z1) are the coordinates of a point on the target flight route, and (x2, y2, z2) are the coordinates of a point on the prescribed route. By calculating the Euclidean distance between each pair of points, a series of distance values ​​are obtained. Next, a threshold is determined, which is used to determine what kind of distance is considered abnormal. If the distance between each pair of points is less than this threshold, then we can assume that the target flight route is consistent with the prescribed route and there is no abnormality. On the contrary, if there is at least one pair of points whose distance exceeds the threshold, then we will determine this situation as a route abnormality.

[0067] Combined anomalies refer to situations where multiple abnormal situations occur at the same time or in the same event. According to the anomaly analysis method described above, different types of anomalies can be combined for comprehensive analysis to identify more complex and hidden abnormal behaviors. For example, assume that the system uses four methods: numerical anomaly analysis, action anomaly analysis, distance anomaly analysis, and cross-border anomaly analysis. At a certain moment, the system detects that the speed of the target suddenly increases, and the target begins to perform abnormal actions, and its position point exceeds the specified area. In this case, these abnormal situations can be considered comprehensively to determine whether there is a combined anomaly. In the specific analysis process, the numerical anomaly is first detected, indicating that the speed of the target has changed abnormally; then the action anomaly analysis is triggered, and it is found that the target begins to perform unusual actions; at the same time, the distance anomaly analysis shows that the distance between the target and the reference point has increased abnormally; and finally, the cross-border anomaly analysis confirms that the target's position point has crossed the boundary. Comprehensively considering these abnormal situations, it can be judged that the target has a combined anomaly, that is, its speed increases abnormally, performs abnormal actions, and the position point crosses the boundary, involving more serious abnormal behaviors.

[0068] At the same time, in some cases, a single anomaly analysis cannot fully reveal the potential abnormal behavior of the target, especially when the abnormal behavior of the target is secretive or complex. At this time, it is necessary to consider multiple abnormal situations to more accurately identify abnormal behavior, which is the importance of combined anomaly analysis. For example, in detecting abnormal behavior of an aircraft. Single anomaly analysis only focuses on the abnormal speed or altitude of the aircraft, but these anomalies may not fully reflect the potential abnormal behavior of the aircraft, such as operation under specific conditions. However, if multiple abnormal situations are combined for analysis, more accurate conclusions will be obtained. For example, speed anomaly analysis, altitude anomaly analysis, position anomaly analysis and action anomaly analysis are performed simultaneously. At a certain moment, it is detected that the speed of the aircraft suddenly increases, and at the same time, the altitude of the aircraft changes abnormally, its position point also deviates from the prescribed route, and the aircraft also performs abnormal actions. Considering each abnormal situation alone cannot provide enough information to determine whether the aircraft has abnormal behavior. For example, the increase in speed and abnormal changes in altitude may be external manifestations caused by special tasks, but considering these abnormal situations in combination can more accurately determine whether the aircraft is in an abnormal state. In this example, a single anomaly analysis may lead to misjudgment or missed judgment due to lack of comprehensive information, but combined anomaly analysis can more comprehensively consider abnormal situations in multiple aspects, thereby improving the ability to identify potential abnormal behaviors. Therefore, for some hidden or complex abnormal behaviors, the effect of a single anomaly analysis may not be obvious, while combined anomaly analysis can more accurately determine the abnormal state of the target.

[0069] Through the comprehensive analysis of different types of anomalies, we can have a more comprehensive understanding of the target's behavior and improve the ability to detect and identify abnormal behaviors. Therefore, combined anomaly analysis is of great significance in anomaly detection systems, which can help users discover and respond to complex anomalies in a timely manner.

[0070] In summary, the anomaly analysis model is a model used to detect abnormal behavior of the target. It combines multiple anomaly detection methods, including numerical anomaly analysis, action anomaly analysis, distance anomaly analysis, and cross-boundary anomaly analysis. The model receives input from preprocessed data and environmental features, and identifies the abnormal behavior of the target by analyzing these data. First, the anomaly analysis model receives preprocessed data, which has undergone preprocessing steps such as outlier removal, numerical anomaly detection, and filtering correction to reduce noise interference in anomaly detection and improve data quality. The preprocessed data contains the temporal and spatial information and basic attribute information of the target, such as position, speed, height, etc., as well as environmental information, such as meteorological information and position information. Secondly, the anomaly analysis model calls various anomaly detection algorithms to perform anomaly detection based on preprocessed data and environmental features. These algorithms include numerical anomaly detection algorithms, action anomaly analysis algorithms, distance anomaly analysis algorithms, and cross-boundary anomaly analysis algorithms. By comprehensively analyzing these anomaly detection results, the model can identify the abnormal behavior of the target. In summary, the anomaly analysis model can effectively identify the abnormal behavior of the target by integrating multiple anomaly detection methods and combining preprocessed data and environmental features, thereby improving the accuracy and reliability of anomaly detection.

[0071] Step S5: When the abnormal detection result is an abnormal behavior, an abnormal warning prompt is issued for the abnormal detection result.

[0072] In step S5, when the anomaly detection model detects abnormal behavior of the target, the abnormal behavior warning process is triggered. This warning process is carried out through a pre-designed behavior anomaly warning processing module. When abnormal behavior is detected, the system immediately generates abnormal behavior warning messages, which are generated in a pre-defined standard message format. The abnormal behavior warning message contains detailed information about the abnormal behavior, such as the type of abnormality, target information, timestamp, etc. These messages will be distributed to relevant business personnel so that they can take appropriate actions to deal with the abnormal situation. In addition, the warning message will also be passed to the back-end monitoring equipment to provide an alarm prompt. The alarm prompt can take a variety of forms, including text display, sound prompt, graphical display, etc. Through such an early warning system, timely monitoring and warning of potential threat targets or sensitive area targets can be achieved to ensure that necessary measures are taken in time to deal with abnormal situations and ensure the safety of relevant areas. At the same time, the rapid response capability of the early warning system also helps to reduce the losses and risks that may be caused by abnormal behavior.

[0073] Based on the description of the above-mentioned aerial target abnormal behavior monitoring method embodiment, the present application embodiment also discloses an aerial target abnormal behavior monitoring device, and the aerial target abnormal behavior monitoring method device can be a computer program (including program code) for running the above-mentioned aerial target abnormal behavior monitoring method. Figure 3 As shown, the aerial target abnormal behavior monitoring method device can run the following units:

[0074] An acquisition unit 110 is used to acquire data to be processed, where the data to be processed includes spatiotemporal information and basic attribute information;

[0075] A preprocessing unit 120, used for preprocessing the data to be processed to obtain preprocessed data;

[0076] The environmental feature extraction unit 130 is used to call the environmental information library, match the environmental information according to the data to be processed, and extract the environmental features in the environmental information;

[0077] The abnormality analysis unit 140 is used to call the abnormality analysis model, input the preprocessed data and environmental characteristics into the abnormality analysis model, and thus perform abnormality detection on abnormal behavior;

[0078] The abnormal warning unit 150 is used to issue an abnormal warning prompt for the abnormal detection result when the abnormal detection result is an abnormal behavior.

[0079] In addition, a computer-readable storage medium is also provided in an embodiment of the present application, and the computer-readable storage medium includes a computer-readable storage medium on which an abnormal behavior monitoring program for an aerial target and target information data are stored. The computer-readable storage medium may be an attached Figure 1 The memory in the terminal shown may also be at least one of a ROM (Read-Only Memory) / RAM (Random Access Memory), a magnetic disk, and an optical disk, and the readable storage medium includes a number of instructions for enabling an aerial target abnormal behavior monitoring device having a processor to execute the aerial target abnormal behavior monitoring method described in the present application.

[0080] In addition, in order to achieve the above purpose, the present application also provides an aerial target abnormal behavior monitoring device, please refer to the attached Figure 1As shown, the abnormal behavior monitoring device for aerial targets is a device for monitoring and identifying abnormal behavior of aerial targets. The device includes a memory, a processor, and an abnormal behavior monitoring program for aerial targets stored in the memory. First, the memory stores various data and programs. In this device, the memory stores the track information, environmental information, and abnormal behavior monitoring program for aerial targets. The track information includes information such as the position, speed, and altitude of the target, and the environmental information includes the meteorological conditions and geographical location of the target's environment. Secondly, the processor is responsible for performing various calculation and processing tasks. The processor contains programs in the memory and instructions from the outside. The processor analyzes and processes the data stored in the memory according to the pre-designed abnormal behavior monitoring program for aerial targets. The processor can also execute instructions from the outside to cope with different monitoring needs and changes in situations. In addition to the memory and the processor, the abnormal behavior monitoring device for aerial targets also includes a network communication module, a user interface module, and a system interface module. The network communication module is responsible for the communication between the device and the external network in order to obtain real-time data and transmit warning information. The user interface module provides an interface for the user to interact with the device, allowing the user to easily monitor and manage the operating status of the device. The system interface module is used to exchange and communicate data with other systems or devices. In summary, the aerial target abnormal behavior monitoring device can monitor and identify the abnormal behavior of aerial targets in real time through the data and programs stored in the memory and the computing and processing capabilities of the processor. At the same time, the addition of the network communication module, user interface module and system interface module enables the device to have better communication and interaction capabilities, and can respond to various monitoring needs and abnormal situations in a timely manner.

[0081] The above is only a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein, and should not be regarded as excluding other embodiments, but can be used in various other combinations, modifications and environments, and can be modified within the scope of the concept described herein through the above teachings or the technology or knowledge in the relevant field. The changes and modifications made by those skilled in the art shall be within the scope of protection of the claims attached to the present invention if they do not deviate from the spirit and scope of the present invention.

Claims

1. A method for monitoring abnormal behavior of aerial targets, characterized in that: The method comprises the following steps: S1, obtaining data to be processed, wherein the data to be processed includes spatiotemporal information and basic attribute information; S2, preprocessing the data to be processed to obtain preprocessed data; S3, calling the environmental information library, matching the environmental information according to the preprocessed data, and extracting environmental features from the environmental information; S4, calling the anomaly analysis model, inputting the preprocessed data and environmental features into the anomaly analysis model, thereby performing anomaly detection on the abnormal behavior; S5, when the abnormal detection result is an abnormal behavior, an abnormal warning prompt is issued for the abnormal detection result.

2. The method for monitoring abnormal behavior of an aerial target according to claim 1, characterized in that: The S2 step includes the following steps: S21, removing outliers from the data to be processed; S22, performing numerical anomaly detection and filtering correction on the data to be processed after outliers are eliminated to obtain preprocessed data.

3. The method for monitoring abnormal behavior of an aerial target according to claim 2, characterized in that: The anomaly analysis model includes six processing methods: numerical anomaly analysis, action anomaly analysis, distance anomaly analysis, cross-boundary anomaly analysis, route anomaly analysis and combined anomaly analysis.

4. The method for monitoring abnormal behavior of an aerial target according to claim 3, characterized in that: The numerical anomaly analysis calculation includes clustering algorithm, tree model, classification algorithm and dimension reduction algorithm.

5. The method for monitoring abnormal behavior of an aerial target according to claim 3, characterized in that: The action anomaly analysis calculates the minimum turning radius of the target and matches the target's route path using a similar time series matching algorithm to identify specific actions, where the minimum turning radius is calculated as follows: g is the gravitational acceleration constant, R min是 Minimum turning radius, v is the speed, Φ max is the maximum slope angle, and the calculation formula of the similar time series matching algorithm is as follows: Among them, φ(T i ∩E j ) represents the track T i With track E j The total length of the intersecting subpaths, φ(T i ∪E j ) represents the track T i With track E j The total path length.

6. The method for monitoring abnormal behavior of an aerial target according to claim 3, characterized in that: The distance anomaly analysis is calculated and judged by the following formula: r is the radius of the earth, and the distance d between two points is determined by calculation AB This can determine whether there is a distance anomaly.

7. The method for monitoring abnormal behavior of an aerial target according to claim 3, characterized in that: The out-of-bounds exception analysis determines whether it is out-of-bounds by the following formula: Where (x0, y0) is the coordinate of the center of the circle, α is the rotation angle of the circle, and b is the radius of the circle.

8. The method for monitoring abnormal behavior of an aerial target according to claim 3, characterized in that: The combined anomaly analysis is to combine the numerical anomaly analysis, the action anomaly analysis, the distance anomaly analysis, and the cross-boundary anomaly analysis to perform a comprehensive anomaly analysis.

9. An aerial target abnormal behavior monitoring device, characterized in that: The device comprises: An acquisition unit, used for acquiring data to be processed, wherein the data to be processed includes spatiotemporal information and basic attribute information; A preprocessing unit, used for preprocessing the data to be processed to obtain preprocessed data; An environmental feature extraction unit is used to call an environmental information library, match environmental information according to the data to be processed, and extract environmental features from the environmental information; The anomaly analysis unit is used to call the anomaly analysis model and input the preprocessed data and environmental characteristics into the anomaly analysis model, so as to perform anomaly detection on abnormal behaviors; The abnormal warning unit is used to issue an abnormal warning prompt for the abnormal detection result when the abnormal detection result is an abnormal behavior.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores one or more instructions, and the one or more instructions are suitable for being loaded by a processor and executing the method for monitoring abnormal behavior of aerial targets as described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • Unmanned aerial vehicle system state monitoring and early warning system based on artificial intelligence

    CN113657665A

  • Ship abnormal behavior identification method and device, terminal and storage medium

    CN114742128A

  • Target behavior abnormity early warning processing system and method

    CN115456075A

Cited By

  • Unmanned aerial vehicle locking monitoring method and system based on cloud control platform

    CN119782846A