A low-altitude aircraft situation awareness and conflict warning method based on multi-source data fusion

Abstract

The application discloses a low-altitude aircraft situation awareness and conflict early warning method based on multi-source data fusion, relates to the technical field of low-altitude airspace monitoring and flight safety guarantee, and through a dynamic partition mechanism driven by a wind field and a terrain, airspace division can be in real time response to environmental changes, adaptive space units are provided for subsequent data processing, and the pertinence of multi-source data fusion and resource allocation efficiency are remarkably improved; for a data sparse area, a low-rank matrix filling technology is adopted to reconstruct an aircraft track, internal relations and motion continuity among multi-dimensional states are fully utilized, and the defects of insufficient accuracy of a traditional interpolation method are effectively overcome, and continuous and reliable track data can still be recovered under the condition that observation is seriously missing.

Description

Technical Field

[0001] This invention relates to the field of low-altitude airspace surveillance and flight safety assurance technology, and in particular to a method for low-altitude aircraft situational awareness and conflict early warning based on multi-source data fusion. Background Technology

[0002] As a crucial extension of aviation activities, low-altitude airspace has seen a significant increase in operational density and complexity in recent years due to the rapid development of emerging industries such as general aviation, drone logistics, and urban air traffic. This places higher demands on the real-time situational awareness and conflict early warning capabilities of low-altitude aircraft. Currently, surveillance and conflict detection technologies for low-altitude airspace mainly rely on data fusion from multiple sources, including Automatic Dependent Surveillance-Broadcast (ADS-B) and primary / secondary radar. Existing technologies employ two main approaches: one involves structuring the airspace by establishing fixed geographic grids or pre-divided static sectors, and calculating the proximity between pairs of aircraft based on their reported position and speed information to assess conflict risk; the other approach introduces trajectory prediction models, such as Kalman filtering and interactive multi-model systems, to extrapolate the future trajectory of individual aircraft and assess collision probabilities. Furthermore, some studies have attempted to incorporate environmental factors (such as wind fields and terrain) into airspace management, constructing environmental field models to assist in flight path planning or conflict avoidance. While these methods have achieved some success in specific scenarios, they still have systemic limitations when facing complex low-altitude environments.

[0003] However, existing technologies have significant shortcomings in the collaborative mechanisms for multi-source data fusion and conflict detection. First, static or quasi-static airspace partitioning methods are difficult to adapt to the differences in perception characteristics caused by rapid changes in wind fields and terrain undulations in low-altitude environments. This results in a non-uniform distribution of data reliability and completeness in different regions, which in turn affects the allocation of computational resources and the adaptation of algorithm parameters in subsequent processing units. Second, in data-sparse areas, due to communication obstruction, sensor coverage blind spots, or intermittent target disconnection, aircraft status data often suffers from temporal gaps. Existing track processing methods typically use interpolation or single-target motion models to fill in the gaps, failing to utilize the potential motion correlations between multiple targets within the region, thus limiting reconstruction accuracy. Third, traditional full-pair conflict detection methods face computational explosion problems when the number of aircraft increases, while hierarchical fusion strategies often use fixed grids as intermediate processing units, failing to link with dynamic airspace partitioning. The parameter transfer between coarse screening and fine calculation lacks an adaptive mechanism, making it difficult to achieve a balance between real-time performance and accuracy. Finally, most existing conflict warning models are based on linear extrapolation based on the assumption of uniform linear motion, ignoring the acceleration changes of aircraft during low-altitude maneuvers, which leads to the accumulation of prediction errors. Moreover, the warning output often only includes a conflict indicator and lacks a precise description of the conflict location and the time of occurrence, making it difficult to support the accurate decision-making of automatic collision avoidance systems or controllers. Summary of the Invention

[0004] The purpose of this section is to outline some aspects of the embodiments of the present invention and to briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this section, the abstract and title of the invention. Such simplifications or omissions shall not be used to limit the scope of the present invention.

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a method for low-altitude aircraft situational awareness and conflict early warning based on multi-source data fusion, characterized by comprising: receiving multi-source data streams; dynamically partitioning the low-altitude airspace according to wind field information and terrain data to generate multiple airspace partitions with different sensing characteristics; for data-sparse partitions within the airspace partitions, reconstructing missing aircraft state data using low-rank matrix filling technology to form a complete airspace situational matrix; based on the airspace situational matrix, employing a hierarchical partitioning fusion strategy to perform coarse-grained fusion globally to screen candidate conflict aircraft pairs, and performing high-precision fusion within the local partitions where the candidate conflict aircraft pairs are located to obtain accurate relative motion parameters; calculating the minimum predicted distance between aircraft based on the accurate relative motion parameters; and generating a conflict early warning signal and associating it with the corresponding aircraft identifier when the minimum predicted distance is less than a safety threshold.

[0007] As a preferred embodiment of the present invention, the multi-source data stream includes broadcast automatic correlation surveillance data, radar data, and meteorological data; the meteorological data includes wind field information and terrain data.

[0008] In a preferred embodiment of the present invention, the dynamic partitioning includes: receiving the multi-source data stream, extracting wind field vector data from the meteorological data, extracting terrain elevation data from the multi-source data stream, mapping the wind field vector data and the terrain elevation data to a unified geographic grid coordinate system, generating gridded environmental data containing wind field vectors and terrain elevations; calculating the wind field variance and terrain undulation within each geographic grid, inputting the wind field variance and the terrain undulation as partitioning feature values ​​into a pre-established clustering model, and outputting an initial partitioning label for each geographic grid; merging adjacent geographic grids with the same initial partitioning label according to the initial partitioning label to form multiple consecutive spatial partitions, and generating a corresponding partitioning identifier for each spatial partition.

[0009] As a preferred embodiment of the present invention, the pre-established clustering model is K-means or DBSCAN.

[0010] As a preferred embodiment of the present invention, the formation of a complete airspace situation matrix includes: filtering all airspace partitions according to a preset sparsity criterion, determining the sparse partitions that require data reconstruction, and recording a list of identifiers for the sparse partitions; for each sparse partition, obtaining the identifiers of all aircraft currently located within the sparse partition from the broadcast automatic correlation surveillance data and radar data, and extracting a sequence of discrete state points within a historical time window for each aircraft; the discrete state points include timestamps, longitude, latitude, and altitude; for each aircraft, arranging the discrete state points in chronological order. A sparse data matrix is ​​constructed, where the rows of the sparse data matrix correspond to uniform sampling times within a historical time window, and the columns correspond to the three state parameters of longitude, latitude, and altitude. Utilizing the continuity and dynamic characteristics of the aircraft's trajectory, a low-rank matrix filling technique is used to fill in the missing elements in the sparse data matrix, generating a complete trajectory data matrix for the aircraft. The complete trajectory data of all aircraft within the sparse partitions are merged with the original observation data of aircraft within the non-sparse partitions to form a globally unified airspace situation dataset. This dataset contains continuous position information of each aircraft at every moment within the historical window.

[0011] As a preferred embodiment of the present invention, the preset sparsity determination criteria include: the number of aircraft in the sparse partition at the current moment is lower than a first threshold, or the data missing rate of aircraft in the partition within the historical time window is higher than a second threshold.

[0012] As a preferred embodiment of the present invention, the hierarchical partitioning fusion strategy includes: extracting the position coordinates and velocity vectors of all aircraft at the current moment from the airspace situation dataset; pairing all aircraft in pairs; calculating the relative distance and relative velocity between each pair of aircraft; marking aircraft pairs whose relative distance is less than a first distance threshold and whose relative velocity is greater than a first velocity threshold as candidate conflict aircraft pairs; generating a candidate conflict aircraft pair list; for each pair of aircraft in the candidate conflict aircraft pair list, extracting the historical state sequence of the two aircraft within a time window at the current moment from the airspace situation dataset based on the aircraft identifier; performing a relative motion coordinate system transformation on the extracted historical state sequences of the two aircraft, establishing a relative motion coordinate system with one of the aircraft as the origin, and calculating the relative motion parameters, including the relative velocity vector, relative acceleration vector, and relative heading angle.

[0013] In a preferred embodiment of the present invention, generating a conflict warning signal and associating it with the corresponding aircraft identifier includes: acquiring the relative motion parameters; assuming that the two aircraft maintain uniform acceleration motion during a future predicted time period; establishing a relative motion equation and solving it to obtain the relative position sequence at each future moment; traversing the relative position sequence, calculating the relative distance at each moment, extracting the minimum value as the minimum predicted distance, and comparing the minimum predicted distance with a preset safe distance threshold; when the minimum predicted distance is less than the preset safe distance threshold, extracting conflict position coordinates from the relative position sequence according to the corresponding predicted moment, associating the conflict position coordinates with the identifiers of the two aircraft, and generating a conflict warning signal containing the aircraft identifiers and conflict position coordinates.

[0014] The beneficial effects of this invention are as follows: This invention utilizes a dynamic zoning mechanism driven by wind field and terrain to enable real-time airspace partitioning that responds to environmental changes, providing adaptive spatial units for subsequent data processing and significantly improving the targeting and resource allocation efficiency of multi-source data fusion. For sparse data areas, a low-rank matrix filling technique is employed to reconstruct the aircraft trajectory, fully leveraging the inherent correlation and motion continuity between multi-dimensional states to effectively overcome the shortcomings of traditional interpolation methods in terms of accuracy. Even under conditions of severe observation gaps, continuous and reliable trajectory data can still be recovered. Furthermore, through a hierarchical fusion strategy combining global coarse-grained screening and local fine calculation, while ensuring the real-time nature of conflict detection, historical window data is used to accurately calculate relative acceleration and heading angle, significantly improving the robustness and prediction reliability of motion parameter estimation. Finally, based on a uniformly accelerated relative motion model, a traversal search is performed on the minimum prediction distance, generating a refined early warning signal containing the conflict location and time, providing complete information support for automatic collision avoidance and control decisions. This systematically solves the problems of inaccurate situational awareness, difficulty in balancing computational efficiency and accuracy, and early warning in complex low-altitude airspace environments. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 This is a flowchart illustrating a low-altitude aircraft situational awareness and conflict early warning method based on multi-source data fusion, as shown in this invention.

[0016] Figure 2 This is a flowchart of the hierarchical partitioning fusion strategy shown in this invention. Detailed Implementation

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0018] Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort should fall within the scope of protection of this invention.

[0019] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0020] According to an embodiment of the present invention, in combination Figure 1 The flowchart shown illustrates a method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion, comprising: S1: Receives multi-source data streams, dynamically partitions the low-altitude airspace based on wind field information and terrain data, and generates multiple airspace partitions with different sensing characteristics.

[0021] It should be noted that existing technologies for monitoring and managing low-altitude airspace typically use fixed geographic grids or pre-defined static sectors as the basic airspace units. However, such static division methods are difficult to adapt to the dynamic changes in meteorological conditions and terrain features in low-altitude flight environments. For example, changes in wind field distribution directly affect the energy consumption and trajectory stability of aircraft, while terrain undulations significantly impact radar detection and data link coverage. If airspace division is not dynamically correlated with real-time environmental data, the reliability and completeness of sensing data in different areas will vary, thereby affecting the accuracy of subsequent aircraft status reconstruction and conflict detection. To address these issues, the present invention operates as follows: S1.1: Receive the multi-source data stream, extract wind field vector data from the meteorological data, extract terrain elevation data from the multi-source data stream, map the wind field vector data and the terrain elevation data to a unified geographic grid coordinate system, and generate gridded environmental data containing wind field vectors and terrain elevations.

[0022] The multi-source data stream includes broadcast automatic correlation surveillance data, radar data, and meteorological data; the meteorological data includes wind field information and topographic data. Wind field vector data includes wind speed components and wind direction angles; topographic elevation data includes gridded data from digital elevation models or lidar point cloud-derived data.

[0023] Furthermore, the wind field vector data and the terrain elevation data are spatiotemporally aligned and transformed using coordinate systems. A geographic information system spatial interpolation algorithm is then used to map them to a pre-defined unified geographic grid coordinate system. This geographic grid coordinate system is divided using equal latitude and longitude intervals or equidistant projection, generating gridded environmental data containing the mean wind field vector and terrain elevation values ​​within each grid cell. For example, the grid resolution can be set to 0.01 degrees × 0.01 degrees (approximately 1 kilometer × 1 kilometer), and Kriging interpolation is used to interpolate the discrete wind field observation point data to the center point of each grid cell. Simultaneously, the terrain elevation data is aligned to the same grid cell using bilinear interpolation.

[0024] S1.2: Calculate the wind field variance and topographic relief within each geographic grid, input the wind field variance and topographic relief as partitioning feature values ​​into a pre-established clustering model (K-means or DBSCAN), and output the initial partitioning label for each geographic grid.

[0025] Specifically, for each geographic grid in the gridded environmental data, the wind field variance and topographic relief are calculated. The wind field variance is calculated by: statistically analyzing the historical sequence of wind speed components for the grid within a preset time window (e.g., the past 10 minutes) and calculating the variance to characterize the intensity of temporal fluctuations in the wind field; or statistically analyzing the spatial differences in wind field vector space between the grid and its adjacent grids and calculating the spatial gradient variance to characterize the spatial non-uniformity of the wind field.

[0026] The method for calculating terrain undulation is as follows: based on the elevation values ​​of the grid and neighboring grids, calculate the elevation standard deviation or local terrain undulation index to reflect the impact of terrain on aircraft maneuverability and perception masking.

[0027] The wind field variance value of each grid is combined with the topographic relief to form a two-dimensional feature vector, which is then input into a pre-trained clustering model. The clustering model can use the K-means algorithm or the density-based DBSCAN algorithm. The feature vectors of the entire spatial grid are classified unsupervised by the clustering model, and the initial partition label of each geographic grid is output.

[0028] Among them, the K-means algorithm is suitable for scenarios where the feature distribution is clustered and the number of partitions can be preset. It achieves partitioning by iteratively optimizing the sum of squared intra-class distances. The DBSCAN algorithm, on the other hand, does not require a preset number of partitions and can automatically identify partitions of arbitrary shapes based on the density connectivity of the feature space. It is suitable for low-altitude environments with irregular wind field and terrain feature distributions.

[0029] It can be seen that by transforming complex multidimensional environmental features into clusterable quantitative indicators and using machine learning methods to automate and objectify spatial domain partitioning, the subjectivity and static nature of manual partitioning are avoided, and the partitioning results can dynamically respond to environmental changes.

[0030] S1.3: Based on the initial partition label, adjacent geographic grids with the same initial partition label are merged to form multiple consecutive spatial partitions, and a corresponding partition identifier is generated for each spatial partition.

[0031] In this embodiment of the invention, based on the initial partition label, a connected region labeling algorithm is used to merge adjacent geographic grids with the same initial partition label to form spatially continuous spatial partitions.

[0032] Specifically, a four-neighbor or eight-neighbor connectivity criterion can be used to scan the entire spatial grid, grouping grids with the same label and spatial adjacency into the same connected component. Each connected component corresponds to an independent spatial partition. For each generated spatial partition, a unique partition identifier is generated. This partition identifier can be encoded using a UUID or an incrementing integer and is stored in association with attribute information such as the list of grids covered by the partition, the partition center coordinates, and the partition boundary polygon.

[0033] The partition identifier is used in subsequent steps to associate data reconstruction and collision detection processing of aircraft within the partition. For example, in S2, the data filling algorithm is triggered only for regions identified as sparse partitions.

[0034] As can be seen, this invention uses gridded feature calculation and clustering algorithms to overcome the shortcomings of traditional static partitioning in reflecting dynamic environmental changes, thereby improving the scientific nature and flexibility of spatial domain division. Furthermore, by generating and associating partition identifiers, a spatial index foundation is established for multi-source data fusion and hierarchical processing, which helps to reduce computational complexity and improve the targeting of conflict detection.

[0035] S2: For the sparse data partitions in the airspace partition, the missing aircraft status data is reconstructed using low-rank matrix filling technology to form a complete airspace situation matrix.

[0036] S2.1: Based on the preset sparsity criteria, all airspace partitions are filtered to identify sparse partitions requiring data reconstruction, and a list of sparse partition identifiers is recorded. Specifically, based on the preset sparsity criteria, all airspace partitions are filtered in real-time to identify sparse partitions requiring data reconstruction, and a list of sparse partition identifiers is recorded. The sparsity criteria include two parallel conditions: First, the number of aircraft in the partition at the current moment is less than a first threshold, for example, set to 3 aircraft. When the number of aircraft in the partition is less than or equal to 3, it is determined to be a sparse partition. Second, the data missing rate of the aircraft in the partition within the historical time window is higher than a second threshold. The data missing rate is calculated as follows: for all aircraft in the partition, the observation data missing situation at each sampling moment within the historical time window (e.g., the past 60 seconds) is statistically analyzed, and the ratio of the total number of missing sampling points to the number of sampling points that should be sampled is used as the missing rate. If this missing rate exceeds a preset threshold (e.g., 30%), the partition is determined to be a sparse partition.

[0037] Reconstruction can be triggered if either of the above two conditions is met, so as to take into account both the absolute sparsity and the severe relative data missing.

[0038] S2.2: For each sparse partition, obtain the identifiers of all aircraft currently located within the sparse partition from the broadcast automatic correlation surveillance data and radar data, and extract the discrete state point sequence within the historical time window for each aircraft.

[0039] The discrete state points include timestamps, longitude, latitude, and altitude.

[0040] Specifically, for each sparsely labeled partition, based on the partition identifier, the unique identifiers (such as ICAO 24-bit address codes or flight numbers) of all aircraft currently located in that partition are extracted from the broadcast automatic dependent surveillance data stream and radar data stream.

[0041] For each aircraft, a historical time window of length T (T can be 60 seconds) is set, and a uniform time series is generated with a fixed sampling interval Δt (e.g., 1 second). The discrete state points of the aircraft at each sampling time within the time window are retrieved from the original data.

[0042] If observation data exists at a certain moment, the timestamp, longitude, latitude, and altitude of that moment are recorded; if no observation data exists at that moment, it is marked as missing. After extraction, each aircraft corresponds to a set of original observation sequences arranged in chronological order, which includes several known state points and a large number of missing points.

[0043] S2.3: For each aircraft, arrange the discrete state points in chronological order to construct a sparse data matrix.

[0044] In this sparse data matrix, the rows correspond to uniform sampling times within a historical time window, and the columns correspond to the three state parameters of longitude, latitude, and altitude. Utilizing the continuity and dynamic characteristics of the aircraft's trajectory, a low-rank matrix filling technique is used to fill in the missing elements in the sparse data matrix, generating a complete trajectory data matrix for the aircraft.

[0045] For example, for each aircraft, its time-series data at equal intervals within a historical time window is organized into a sparse data matrix. ,in The total number of sampling moments within the time window (e.g., T=60 seconds, Δt=1 second, then) =60), the three columns of the matrix correspond to the three spatial dimensions of longitude, latitude and altitude, respectively.

[0046] The observed positional elements in the matrix retain their original values, while missing elements are filled with null or zero values.

[0047] Because the motion of an aircraft is continuous over a short period of time, the curves of its longitude, latitude, and altitude change over time are usually smooth and there is an inherent correlation between the dimensions. Therefore, the matrix has an approximately low-rank characteristic (for example, the rank of the matrix does not exceed 2 under ideal uniform linear motion and does not exceed 3 under uniformly accelerated motion).

[0048] Based on this, a low-rank matrix filling technique is used to recover missing elements.

[0049] One specific implementation involves solving the following constrained nuclear norm minimization problem: in, Representation matrix The nuclear norm (sum of singular values). For the set of indices of the observed elements, For projection operators, As constraints, Given the complete data matrix to be solved, Given an original sparse data matrix containing known observations and missing values ​​(missing positions are filled with zeros or null values), this problem can be solved iteratively using the Singular Value Thresholding (SVT) algorithm: at each step, the current matrix is ​​decomposed into singular values, soft-thresholding is applied to the singular values, and then the matrix is ​​projected back to the observations. To enhance the physical plausibility of the imputation results, a time smoothing regularization term can be added to the optimization objective, for example: in, The regularization coefficient is . It is a second-order difference matrix used to penalize abrupt changes in acceleration. The Frobenius norm, which is the square root of the sum of the squares of all elements in the matrix, is used here to measure the temporal smoothness of the trajectory. This problem can be solved using the Alternating Direction Multipliers (ADMM) method or the proximal gradient algorithm. Through iterative optimization, the complete trajectory data matrix is ​​finally obtained. Each row corresponds to an estimated value of longitude, latitude, and altitude at a sampling time.

[0050] It should be noted that this invention fully utilizes the inherent low-rank nature and temporal continuity of aircraft motion to achieve high-precision missing value completion with a small amount of observation data, without requiring precise motion model assumptions, and is adaptable to various maneuvering modes.

[0051] S2.4: Merge the complete trajectory data of all aircraft in the sparse partition with the original observation data (or data that has been simply interpolated) of aircraft in the non-sparse partition to form a globally unified airspace situation dataset. The dataset contains the continuous position information of each aircraft at every moment within the historical window.

[0052] The airspace situation dataset uses the aircraft identifier as the primary key to store the continuous position information of each aircraft at every moment within the historical time window. Furthermore, it can calculate motion parameters such as velocity and acceleration using numerical difference methods (e.g., deriving the velocity vector from the position sequence using the central difference formula).

[0053] The dataset can be physically stored using a time-series database or an in-memory hash table structure to support the rapid retrieval of the historical state of a specific aircraft in subsequent step S3. Furthermore, to facilitate subsequent hierarchical fusion processing, the identifier of the current partition of the aircraft can be retained in the dataset as an auxiliary index, but this is not mandatory.

[0054] As can be seen, this invention breaks through the limitations of traditional single-objective interpolation methods by using low-rank matrix filling technology, improves the accuracy of missing data recovery by utilizing multi-dimensional intrinsic correlation, and optimizes the real-time performance of the system by focusing computing resources on the regions that really need to be filled through a sparse partitioning identification mechanism.

[0055] S3: Based on the airspace situation matrix, a hierarchical partition fusion strategy is adopted to perform coarse-grained fusion in the global scope to screen out candidate conflict aircraft pairs, and high-precision fusion is performed in the local partition where the candidate conflict aircraft pairs are located to obtain accurate relative motion parameters.

[0056] In low-altitude airspace vehicle conflict detection, traditional methods typically perform pairwise calculations on all vehicles and determine conflicts based on instantaneous states. However, as the number of vehicles in the airspace increases, full-pair calculations lead to a combination explosion, making real-time performance difficult to guarantee. Simultaneously, observation data at a single moment is susceptible to noise interference, and directly using it for conflict determination can easily result in false alarms or missed detections. Existing hierarchical fusion strategies mostly employ fixed grids or static partitions for data aggregation, failing to fully utilize the characteristics of dynamic airspace partitioning, and lacking adaptability in the connection between coarse screening and fine calculation. This invention addresses these problems by proposing a hierarchical fusion strategy based on dynamic partitioning: firstly, candidate conflict pairs are quickly screened globally using low-precision but high-efficiency indicators; then, for each candidate pair, high-precision relative motion parameters are calculated using historical trajectory data within a local area, thereby achieving a balance between computational efficiency and detection accuracy. Figure 2 As shown, the details are as follows: S3.1: Extract the position coordinates and velocity vectors of all aircraft at the current moment from the airspace situation dataset, pair all aircraft together, calculate the relative distance and relative velocity between each pair of aircraft, mark the aircraft pairs whose relative distance is less than a first distance threshold and whose relative velocity is greater than a first velocity threshold as candidate conflict aircraft pairs, and generate a list of candidate conflict aircraft pairs.

[0057] For example, the current time is extracted from the aforementioned airspace situation dataset. Spatial status information for all aircraft, including the position vector of each aircraft. and The position vector can be directly obtained from the latest latitude and longitude coordinates in the dataset, converted to Cartesian coordinates. The velocity vector can be obtained through the position difference between adjacent time points, or read from the smoothed velocity field pre-stored in the dataset. All aircraft are paired up. For any pair of aircraft... Calculate the relative distance and relative velocity magnitude : Set a first distance threshold (e.g., 5 km) and a first speed threshold (e.g., 50 knots, which needs to be converted to m / s in engineering implementation). If both conditions are met... Less than or equal to the first distance threshold and If the speed exceeds the first speed threshold, the aircraft pair is marked as a candidate conflict aircraft pair and recorded in the candidate list.

[0058] It should be noted that the velocity condition here is used to exclude pairs that are relatively stationary or approaching slowly, to avoid subsequent detailed calculations for non-threatening targets.

[0059] S3.2: For each pair of aircraft in the candidate conflict aircraft pair list, extract the historical state sequence of the two aircraft within a time window at the current moment from the airspace situation dataset according to the aircraft identifier.

[0060] For example, for each pair of aircraft in the candidate list Based on the aircraft identifiers, extract the current status of these two aircraft from the airspace situation dataset. A previous time window Internal historical state sequence, where This is the preset local observation window length (e.g., 10 seconds). The sequence contains uniformly sampled time intervals. ,in, The sampling interval, such as the position vector corresponding to 1 second. and ,in, The number of discrete sampling points within the historical time window is reduced by 1. If some moments in the dataset are missing, the complete trajectory data reconstructed in S2 can be used to ensure the continuity of the sequence. After extraction, each pair of aircraft corresponds to two sets of equally spaced position-time series.

[0061] S3.3: Perform a relative motion coordinate system transformation on the extracted historical state sequences of the two aircraft, establish a relative motion coordinate system with one of the aircraft as the origin, and calculate the relative motion parameters, which include the relative velocity vector, the relative acceleration vector, and the relative heading angle.

[0062] For example, one of the aircraft (e.g., an aircraft) Establish a relative motion coordinate system with the origin as the coordinate point. Position the aircraft... Compared to aircraft The relative position vector is defined as: Obtain the relative position sequence Assuming that the relative motion of the two aircraft within a short time window can be approximated by a uniform acceleration model, meaning their relative positions change quadratically with time, a polynomial fitting method is used. Least squares fitting is performed on each component of the relative position vector (e.g., x, y, z directions), and the fitting function is given by: in, The relative position, relative velocity, and relative acceleration vectors are respectively located in... The estimated value at time. By minimizing the sum of squared errors: in, .

[0063] Calculated That is, the relative velocity vector , That is, the relative acceleration vector .

[0064] Furthermore, relative heading angle Defined as the angle between the projection of the relative velocity vector onto the horizontal plane and the due north direction, the calculation formula is: in, This represents the horizontal component of the relative velocity vector. If the vertical direction needs to be considered, the vertical relative velocity component can be calculated separately.

[0065] The calculated relative velocity vector Relative acceleration vector and The precise relative motion parameters are output for subsequent steps to predict conflicts.

[0066] It should be noted that the above operations, through quadratic fitting of local historical data, not only obtain the instantaneous relative motion state at the current moment, but also estimate the acceleration information, making future trajectory prediction more accurate; at the same time, least squares fitting has the effect of smoothing noise and improving the robustness of the parameters.

[0067] S4: Calculate the minimum predicted distance between the aircraft based on the precise relative motion parameters. When the minimum predicted distance is less than the safety threshold, generate a conflict warning signal and associate it with the corresponding aircraft identifier.

[0068] S4.1: Obtain the relative motion parameters, assume that the two aircraft will maintain uniform acceleration during the predicted time period in the future, establish the relative motion equation, and solve it to obtain the relative position sequence at each future moment.

[0069] Assuming a future forecast period The two aircraft maintain uniformly accelerated relative motion, among which The preset prediction duration is (e.g., 30 seconds). Then, at any future time... relative position vector It can be represented as: in, The relative position vector at the current moment can be obtained directly from the airspace situation dataset or calculated from the position difference between the two aircraft. The prediction time period is divided into fixed steps. Discretize (e.g., 1 second) to obtain a series of discrete moments: Substituting into the above formula, the relative position vector at each discrete moment can be calculated. Forming a relative position sequence .

[0070] S4.2: Traverse the relative position sequence, calculate the relative distance at each moment, extract the minimum value as the minimum predicted distance, and compare the minimum predicted distance with the preset safe distance threshold.

[0071] For each discrete time Calculate the relative distance scalar (Euclidean norm). Take all The minimum value in the range is used as the minimum predicted distance of the aircraft for the future prediction period. Simultaneously, the time index or time corresponding to this minimum value can be recorded for subsequent collision location. Compare with a preset safe distance threshold (e.g., 500 meters). If If the distance is less than the preset safe distance threshold, a potential conflict risk is identified, and an early warning needs to be triggered; otherwise, the two aircraft are considered safe and no further action is required.

[0072] It can be seen that by traversing discrete points to accurately obtain the minimum distance, the missed reports caused by peak omissions are avoided; at the same time, the threshold comparison provides a clear basis for conflict determination, and its value can be dynamically adjusted according to the airspace type (such as urban airspace, mountainous airspace), thereby enhancing adaptability.

[0073] S4.3: When the minimum predicted distance is less than the preset safe distance threshold, the conflict position coordinates are extracted from the relative position sequence according to the corresponding minimum distance time, and the conflict position coordinates are associated with the identifiers of the two aircraft to generate a conflict warning signal containing the aircraft identifiers and conflict position coordinates.

[0074] Among them, by means of coordinate transformation, the relative position can be converted into the coordinates of the absolute conflict position in the geographic coordinate system.

[0075] In practical applications, the average of the extrapolated positions of the two aircraft can be used as the conflict location to reduce unilateral errors. Subsequently, the conflict location coordinates, the expected conflict time, and the identifiers of the two aircraft (such as ICAO address codes) are associated and encapsulated to generate a structured conflict warning signal. This signal can further include auxiliary information such as minimum distance values ​​and relative speeds to support decision-making by ground control systems or airborne collision avoidance systems. The warning signal can be transmitted to the relevant aircraft or display terminals via data link.

[0076] As can be seen, this invention replaces traditional linear extrapolation with a uniformly accelerated relative motion model, making future trajectory predictions more closely match the actual maneuvering characteristics of aircraft in low-altitude environments, thus significantly improving the accuracy of conflict prediction. Simultaneously, the use of a discretized traversal method to accurately search for the minimum prediction distance effectively avoids missed reports due to peak omissions, and a preset safety threshold ensures reliable conflict determination. The resulting warning signal not only contains information about the existence of a conflict but also links the specific conflict location coordinates and the expected conflict time, providing a complete spatiotemporal decision-making basis for ground control systems or airborne collision avoidance equipment. This transforms the precise relative motion parameters obtained in the preceding steps into directly executable avoidance commands.

[0077] The method also includes one or more processors and memory.

[0078] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the flow of a low-altitude aircraft situational awareness and conflict early warning method based on multi-source data fusion according to the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0079] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of a low-altitude aircraft situational awareness and conflict early warning method based on multi-source data fusion according to the foregoing embodiments, particularly... Figure 1 The flowchart of the method is shown.

[0080] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0081] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0082] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0083] In any case, the language can be either compiled or interpreted.

[0084] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0085] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0086] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0087] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0088] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0089] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion, characterized in that: include: It receives multi-source data streams and dynamically partitions the low-altitude airspace based on wind field information and terrain data, generating multiple airspace partitions with different sensing characteristics. For the sparse data partitions in the airspace partitioning, the missing aircraft status data is reconstructed using low-rank matrix filling technology to form a complete airspace situation matrix. Based on the airspace situation matrix, a hierarchical partition fusion strategy is adopted to perform coarse-grained fusion in the global scope to screen out candidate conflict aircraft pairs, and high-precision fusion is performed in the local partition where the candidate conflict aircraft pairs are located to obtain accurate relative motion parameters. The minimum predicted distance between the aircraft is calculated based on the precise relative motion parameters. When the minimum predicted distance is less than a safety threshold, a conflict warning signal is generated and associated with the corresponding aircraft identifier.

2. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 1, characterized in that: The multi-source data stream includes broadcast automatic correlation surveillance data, radar data, and meteorological data; the meteorological data includes wind field information and terrain data.

3. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 2, characterized in that: The dynamic partitioning includes: Receive the multi-source data stream, extract wind field vector data from the meteorological data, extract terrain elevation data from the multi-source data stream, map the wind field vector data and the terrain elevation data to a unified geographic grid coordinate system, and generate gridded environmental data containing wind field vectors and terrain elevations. Calculate the wind field variance and topographic relief within each geographic grid, input the wind field variance and topographic relief as partitioning feature values ​​into a pre-established clustering model, and output the initial partitioning label for each geographic grid. Based on the initial partition label, adjacent geographic grids with the same initial partition label are merged to form multiple consecutive spatial partitions, and a corresponding partition identifier is generated for each spatial partition.

4. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 3, characterized in that: The pre-established clustering model is either K-means or DBSCAN.

5. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 4, characterized in that: The formation of a complete airspace situation matrix includes: Based on the preset sparsity judgment criteria, all spatial partitions are filtered to determine the sparse partitions that need data reconstruction, and the list of sparse partition identifiers is recorded. For each sparse partition, the identifiers of all aircraft currently located within the sparse partition are obtained from the broadcast automatic correlation surveillance data and radar data, and a sequence of discrete state points within the historical time window is extracted for each aircraft. The discrete state points include timestamps, longitude, latitude, and altitude; For each aircraft, the discrete state points are arranged in chronological order to construct a sparse data matrix; In this sparse data matrix, the rows correspond to uniform sampling times within a historical time window, and the columns correspond to three state parameters: longitude, latitude, and altitude. Utilizing the continuity and dynamic characteristics of the aircraft's trajectory, a low-rank matrix filling technique is used to fill in the missing elements in the sparse data matrix, generating a complete trajectory data matrix for the aircraft. The complete trajectory data of all aircraft in the sparse partitions are merged with the original observation data of aircraft in the non-sparse partitions to form a globally unified airspace situation dataset. The dataset contains the continuous position information of each aircraft at every moment within the historical window.

6. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 5, characterized in that: The preset sparsity determination criteria include: The number of aircraft in the sparse partition at the current moment is lower than the first threshold, or the data missing rate of aircraft in the partition within the historical time window is higher than the second threshold.

7. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 6, characterized in that: The hierarchical partitioning fusion strategy includes: The position coordinates and velocity vectors of all aircraft at the current moment are extracted from the airspace situation dataset. All aircraft are paired up, and the relative distance and relative velocity between each pair of aircraft are calculated. Aircraft pairs whose relative distance is less than a first distance threshold and whose relative velocity is greater than a first velocity threshold are marked as candidate conflict aircraft pairs, and a list of candidate conflict aircraft pairs is generated. For each pair of aircraft in the candidate conflict aircraft pair list, extract the historical state sequence of the two aircraft within a time window at the current moment from the airspace situation dataset based on the aircraft identifier; The historical state sequences of the two aircraft are extracted and a relative motion coordinate system transformation is performed. A relative motion coordinate system is established with one of the aircraft as the origin, and the relative motion parameters are calculated. The relative motion parameters include the relative velocity vector, the relative acceleration vector, and the relative heading angle.

8. The method for situational awareness and conflict early warning of low-altitude aircraft based on multi-source data fusion as described in claim 7, characterized in that: The generation of conflict warning signals and association with corresponding aircraft identifiers includes: The relative motion parameters are obtained. Assuming that the two aircraft maintain uniform acceleration during the predicted time period in the future, the relative motion equations are established and solved to obtain the relative position sequence at each future moment. Traverse the relative position sequence, calculate the relative distance at each moment, extract the minimum value as the minimum predicted distance, and compare the minimum predicted distance with a preset safe distance threshold; When the minimum predicted distance is less than the preset safe distance threshold, the conflict position coordinates are extracted from the relative position sequence according to the corresponding predicted time, and the conflict position coordinates are associated with the identifiers of the two aircraft to generate a conflict warning signal containing the aircraft identifiers and conflict position coordinates.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the low-altitude aircraft situational awareness and conflict early warning method based on multi-source data fusion as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the low-altitude aircraft situational awareness and conflict early warning method based on multi-source data fusion as described in any one of claims 1 to 8.

Images