General aviation monitoring and early warning method and system based on artificial intelligence

Abstract

The invention relates to a general aviation monitoring and early warning method and system based on artificial intelligence, and belongs to the technical field of general aviation monitoring. The method comprises the steps that point cloud data are collected through an airborne laser radar, a point cloud data sequence is obtained through hardware timestamp alignment, a point cloud cluster set sequence is obtained through point cloud preprocessing, and a target point cloud cluster sequence is obtained through a matching algorithm; intercepting a target point cloud cluster sequence in a preset time window to obtain a test point cloud cluster sequence, calculating a state vector of each point cloud cluster and constructing a sequence, calculating a dynamic probability through a classification model, comparing the dynamic probability with a preset comparison threshold, and dividing the point cloud clusters into a dynamic type and a static type; and obtaining a static point cloud cluster centroid coordinate sequence and a dynamic point cloud cluster centroid coordinate sequence, calculating a prediction point coordinate by using the prediction model, and calculating an obstacle avoidance path through an avoidance algorithm in combination with the starting and ending point coordinates of the aircraft. Monitoring and early warning of general aviation are realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of general aviation monitoring, and in particular relates to a general aviation monitoring and early warning method and system based on artificial intelligence. Background Art

[0002] With the rapid development of economy and the continuous advancement of aviation technology, general aviation plays an increasingly important role in emergency rescue, agricultural and forestry operations, aerial surveying and mapping, etc. When performing these tasks, general aviation aircraft often need to operate in low-altitude environments to meet specific mission requirements.

[0003] At present, Automatic Dependent Surveillance-Broadcast (ADS-B) technology is widely used in the field of general aviation as an important aviation surveillance technology. It automatically broadcasts its own position, speed, altitude and other information to ground base stations and other aircraft through the electronic equipment on the aircraft, realizing real-time monitoring of the aircraft. This technology has greatly improved the efficiency and accuracy of aviation surveillance and enhanced flight safety.

[0004] However, when general aviation aircraft perform emergency missions in dangerous conditions such as low visibility, problems arise one after another. The low-altitude operating environment is complex, and there are various external signal interference sources, such as topography, meteorological conditions, and electromagnetic interference from other electronic equipment. These interferences can easily cause ADS-B signal distortion, making it impossible for ground monitoring systems and other aircraft to obtain true signal information from general aviation aircraft. In this case, the aircraft needs to rely on the pilot's experience to make adjustments when performing obstacle avoidance operations, which will face great difficulties and is likely to cause serious accidents such as collisions, seriously threatening flight safety and the smooth execution of missions. Therefore, how to solve the problem of ADS-B signal distortion in complex environments and achieve accurate monitoring and early warning of general aviation aircraft has become a key technical problem that needs to be solved urgently in the current general aviation field. Summary of the invention

[0005] In order to solve the above problems existing in the prior art, the present invention provides a general aviation monitoring and early warning method and system based on artificial intelligence.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] Collecting point cloud data through an airborne laser radar, aligning the point cloud data through hardware timestamps to obtain a point cloud data sequence, performing point cloud preprocessing on the point cloud data sequence to obtain a point cloud cluster sequence, and labeling the point cloud cluster sequence through a matching algorithm to obtain a labeled point cloud cluster sequence;

[0008] A preset time window is used to intercept the target point cloud cluster sequence according to the time window to obtain a test point cloud cluster sequence. The state vector of each point cloud cluster in the test point cloud cluster sequence is calculated, a state vector sequence is constructed according to the state vector, a dynamic probability is calculated through a classification model according to the state vector sequence, a comparison threshold is preset, a judgment result is obtained by judging the dynamic probability according to the comparison threshold, and the point cloud clusters are classified according to the judgment result to obtain dynamic point cloud clusters and static point cloud clusters;

[0009] The centroid coordinates of the static point cloud clusters and the centroid coordinate sequence of the dynamic point cloud clusters are obtained. A predicted point coordinate is calculated through a prediction model according to the centroid coordinate sequence. The starting coordinate and the ending coordinate of the aircraft are obtained, and an obstacle avoidance path is calculated through an avoidance algorithm according to the starting coordinate, the ending coordinate, the centroid coordinate, and the predicted point coordinate.

[0010] Specifically, the obtaining of the point cloud cluster sequence through point cloud preprocessing based on the point cloud data sequence specifically includes:

[0011] The point cloud data sequence is traversed to obtain point cloud frames. The noise points are removed from the point cloud frames through a noise reduction algorithm to obtain denoised point cloud frames. The point cloud frames are clustered through a segmentation algorithm to obtain a point cloud cluster set, and each frame of the point cloud cluster set is integrated to obtain the point cloud cluster set sequence.

[0012] Specifically, the specific calculation steps of the matching algorithm include:

[0013] The morphological parameter vectors of each point cloud cluster in the point cloud cluster set sequence are obtained, two empty vector lists are established, and the morphological parameter vectors in two consecutive frames of the point cloud cluster set are respectively stored in the empty vector lists to obtain a first vector list and a second vector list;

[0014] The point cloud cluster similarity is obtained through a similarity calculation formula according to the first vector list and the second vector list. A similarity matrix is constructed according to the point cloud cluster similarity. The similarity matrix is traversed to obtain the maximum similarity value. The point cloud cluster matching pairs are determined according to the maximum similarity value. The point cloud cluster matching pairs are marked to obtain marked matching pairs, and the marked matching pairs are stored to obtain the target point cloud cluster sequence.

[0015] Specifically, the expression of the state vector is:

[0016] A = (speed t , a t , b t , c t , w t , e t ),

[0017] Among them, A is the state vector, speed t represents the moving speed of the point cloud cluster at time t, a t , b t , c t represent the centroid coordinates of the point cloud cluster at time t, w t represents the moving direction of the point cloud cluster at time t, e t represents the acceleration of the point cloud cluster at time t.

[0018] Specifically, the classification model includes an input layer, an attention layer, and a fully connected layer. The specific calculation steps are as follows:

[0019] The standardized feature vector sequence is obtained by standardizing the state vector sequence. The input layer transmits the standardized feature vector sequence to the attention layer to calculate the context vector, and the dynamic probability is calculated through the fully connected layer according to the context vector.

[0020] Specifically, the specific calculation steps of the prediction model are as follows:

[0021] The coordinate coding information is obtained by mapping the centroid coordinate sequence. The hidden variable vector is calculated through the LSTM encoder according to the coordinate coding information, and the prediction coding vector is calculated through the fully connected layer according to the hidden variable vector;

[0022] The candidate trajectory end point set is obtained. The prediction coding vector and the candidate trajectory end point set are projected to obtain the query matrix, the key matrix, and the value matrix. The weighted fusion result is obtained by performing weighted summation on the query matrix, the key matrix, and the value matrix. The probability value is obtained through calculation according to the weighted fusion result, the probability value sequence is stored according to the probability value, the maximum probability value is obtained by traversing the probability value sequence, and the predicted point coordinates are determined according to the maximum probability value.

[0023] Specifically, the specific calculation steps of the avoidance algorithm are as follows:

[0024] The point cloud coordinates of the static point cloud cluster are obtained, and the static obstacle interval radius is determined according to the centroid coordinates and the point cloud coordinates of the static point cloud cluster;

[0025] The initial point cloud coordinates and the initial centroid coordinates of the dynamic point cloud cluster are obtained, and the dynamic obstacle interval radius is determined according to the initial point cloud coordinates and the initial centroid coordinates of the dynamic point cloud cluster;

[0026] The moving interval radius and the center point are determined according to the dynamic obstacle interval radius, the initial centroid coordinates, and the predicted point coordinates;

[0027] Obtain a raster map, and label the raster map according to the centroid coordinates, the static obstacle interval radius, the moving interval radius, and the center point to obtain a cost map. Calculate the obstacle avoidance path according to the cost map, the starting coordinate, and the ending coordinate through a path planning algorithm.

[0028] A general aviation monitoring and warning system based on artificial intelligence, comprising:

[0029] A collection module, configured to collect point cloud data through an airborne lidar, align the point cloud data according to the hardware timestamp to obtain a point cloud data sequence, perform point cloud preprocessing on the point cloud data sequence to obtain a point cloud cluster sequence, and label the point cloud cluster sequence through a matching algorithm to obtain a labeled point cloud cluster sequence;

[0030] A processing module, configured to preset a time window, intercept the labeled point cloud cluster sequence according to the time window to obtain a test point cloud cluster sequence, calculate the state vector of each point cloud cluster in the test point cloud cluster sequence, construct a state vector sequence according to the state vector, calculate the dynamic probability according to the state vector sequence through a classification model, preset a comparison threshold, judge the dynamic probability according to the comparison threshold to obtain a judgment result, and classify the point cloud cluster according to the judgment result to obtain a dynamic point cloud cluster and a static point cloud cluster;

[0031] An avoidance module, configured to obtain the centroid coordinates of the static point cloud cluster and the centroid coordinate sequence of the dynamic point cloud cluster, calculate the predicted point coordinates according to the centroid coordinate sequence through a prediction model, obtain the starting coordinate and the ending coordinate of the aircraft, and calculate the obstacle avoidance path according to the starting coordinate, the ending coordinate, the centroid coordinates, and the predicted point coordinates through an avoidance algorithm.

[0032] An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the program, it implements the method for general aviation monitoring and warning based on artificial intelligence as described above.

[0033] A storage medium containing computer-executable instructions, wherein the computer-executable instructions are used to execute the method for general aviation monitoring and warning based on artificial intelligence as described above when executed by a computer processor.

[0034] The beneficial effects of the present invention are:

[0035] (1) Point cloud data is collected by airborne lidar. By combining hardware timestamp alignment and various noise reduction algorithms, the signal interference problem of traditional GPS-dependent systems in complex environments (such as urban high-rise buildings and bad weather) is effectively overcome. Compared with existing ADS-B systems, this technology does not rely on GNSS positioning, significantly improving the reliability and anti-interference ability of data collection. At the same time, high-precision segmentation of obstacles is achieved through segmentation algorithms in point cloud preprocessing.

[0036] (2) Through state vector sequence modeling (including parameters such as speed, direction, and acceleration) and classification models (attention mechanism + fully connected layer), the dynamic probability classification accuracy is significantly improved. The attention mechanism enhances the ability to capture key features of time series, reducing the misjudgment rate of dynamic probability threshold judgment and adapting to sudden changes in target motion patterns. The LSTM prediction model combined with the weighted fusion strategy of the trajectory end point candidate set reduces the average displacement error of dynamic obstacle trajectory prediction by more than 20%, and has stronger generalization ability.

[0037] (3) Through the joint avoidance algorithm of centroid coordinates and predicted point coordinates, the static obstacle avoidance accuracy reaches ±0.01 meters, and the dynamic obstacle avoidance response time is less than 0.5 seconds. Combined with path planning using a cost map, the basic obstacle avoidance function can still be maintained through historical trajectory prediction in case of signal loss or sensor failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] For the convenience of those skilled in the art to understand, the present invention will be further described below with reference to the accompanying drawings.

[0039] Figure 1 It is a schematic flowchart of a general aviation monitoring and warning method based on artificial intelligence according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will describe in detail the specific embodiments, structures, features, and their effects of the present invention with reference to the accompanying drawings and preferred embodiments.

[0041] Please refer to Figure 1 , a general aviation monitoring and warning method and system based on artificial intelligence;

[0042] Point cloud data is collected by airborne lidar, the point cloud data sequence is obtained by aligning according to the point cloud data through a hardware timestamp, the point cloud cluster sequence is obtained by point cloud preprocessing according to the point cloud data sequence, and the target point cloud cluster sequence is obtained by performing a target through a matching algorithm according to the point cloud cluster sequence;

[0043] A preset time window is used to intercept the target point cloud cluster sequence according to the time window to obtain a test point cloud cluster sequence. The state vector of each point cloud cluster in the test point cloud cluster sequence is calculated. A state vector sequence is constructed according to the state vector. A dynamic probability is calculated through a classification model according to the state vector sequence. A comparison threshold is preset. A judgment result is obtained by judging the dynamic probability according to the comparison threshold. The point cloud clusters are classified according to the judgment result to obtain dynamic point cloud clusters and static point cloud clusters;

[0044] The centroid coordinates of the static point cloud clusters and the centroid coordinate sequence of the dynamic point cloud clusters are obtained. A predicted point coordinate is calculated through a prediction model according to the centroid coordinate sequence. The starting coordinate and the ending coordinate of the aircraft are obtained. An obstacle avoidance path is calculated through an avoidance algorithm according to the starting coordinate, the ending coordinate, the centroid coordinate, and the predicted point coordinate.

[0045] In this embodiment, if the dynamic probability is less than the comparison threshold, the judgment result is output as a dynamic point cloud cluster; if the dynamic probability is not less than the comparison threshold, the judgment result is output as a static point cloud cluster.

[0046] Specifically, the obtaining of the point cloud cluster sequence by point cloud preprocessing according to the point cloud data sequence specifically includes:

[0047] The point cloud data sequence is traversed to obtain point cloud frames. The noise points are removed from the point cloud frames through a noise reduction algorithm to obtain denoised point cloud frames. The point cloud frames are clustered through a segmentation algorithm to obtain a point cloud cluster set. Each frame of the point cloud cluster set is integrated to obtain the point cloud cluster set sequence.

[0048] In this embodiment, the noise reduction algorithm includes but is not limited to a statistical filtering algorithm, a voxel grid filtering algorithm, a Gaussian filtering algorithm, a shape model-based denoising algorithm, and a deep learning algorithm;

[0049] The segmentation algorithm includes but is not limited to: a RANSAC algorithm, a Cyliner Model Segmentation algorithm, Euclidean clustering, and point cloud supervoxel clustering.

[0050] Specifically, the specific calculation steps of the matching algorithm include:

[0051] The morphological parameter vectors of each point cloud cluster in the point cloud cluster set sequence are obtained. Two empty vector lists are established. The morphological parameter vectors in two consecutive frames of the point cloud cluster set are respectively stored in the empty vector lists to obtain a first vector list and a second vector list;

[0052] The point cloud cluster similarity is obtained through a similarity calculation formula based on the first vector list and the second vector list. A similarity matrix is constructed according to the point cloud cluster similarity. The similarity matrix is traversed to obtain the maximum similarity value. The point cloud cluster matching pairs are determined according to the maximum similarity value. The point cloud cluster matching pairs are labeled to obtain the labeled matching pairs. The labeled matching pairs are stored to obtain the labeled point cloud cluster sequence;

[0053] The expression of the similarity calculation formula is:

[0054]

[0055] where, f ij represents the point cloud cluster similarity between the i-th point cloud cluster in the first vector list and the j-th point cloud cluster in the second vector list. k is a variable parameter representing the frame sequence. (X i , Y i , Z i ) represents the coordinates of the i-th point cloud cluster in the first vector list, and (X j , Y j , Z j ) represents the coordinates of the j-th point cloud cluster in the second vector list. S i is the average laser intensity of the i-th point cloud cluster, and S j represents the average laser intensity of the j-th point cloud cluster. l i , d i , h i represent the length, width, and height of the i-th point cloud cluster, and l j , d j , h j represent the length, width, and height of the j-th point cloud cluster.

[0056] Specifically, the expression of the state vector is:

[0057] A = (speed t , a t , b t , c t , w t , e t ),

[0058] where, A is the state vector, speed t represents the motion speed of the point cloud cluster at time t, a t , b t , c t represent the centroid coordinates of the point cloud cluster at time t, w t represents the motion direction of the point cloud cluster at time t, and e t represents the acceleration of the point cloud cluster at time t.

[0059] In this embodiment, the centroid coordinates of the point cloud cluster are calculated by calling the PCL function, and then the motion speed, motion direction, and acceleration of the point cloud cluster are calculated based on the centroid coordinates of the point cloud cluster.

[0060] Specifically, the classification model includes an input layer, an attention layer, and a fully connected layer. The specific calculation steps are as follows:

[0061] The standardized feature vector sequence is obtained by standardizing the state vector sequence. The input layer transmits the standardized feature vector sequence to the attention layer to calculate the context vector, and the dynamic probability is calculated based on the context vector through the fully connected layer.

[0062] The calculation expression of the context vector is:

[0063]

[0064] where G is the context vector, t is a variable constant, N is the length of the test point cloud cluster sequence, W a and b a are learning parameters, h t is the t-th standardized feature vector in the standardized feature vector sequence;

[0065] The calculation expression of the dynamic probability is:

[0066] P = sigmoid(W o * LeakyReLu(W d * G + b d ) + b o ),

[0067] where P is the dynamic probability, W o , W d , b d , b o are learning parameters.

[0068] Specifically, the specific calculation steps of the prediction model are as follows:

[0069] The coordinate coding information is obtained by mapping the centroid coordinate sequence. The hidden variable vector is calculated based on the coordinate coding information through the LSTM encoder, and the prediction coding vector is calculated based on the hidden variable vector through the fully connected layer.

[0070] The expression of the prediction coding vector is:

[0071]

[0072] where e t is the coordinate coding information at time t, (*) represents the fully connected layer when embedding coordinates, (*) represents the fully connected layer when outputting the predicted coding vector, D t represents the centroid coordinates at time t, W3 is the parameter to be trained for the fully connected layer when embedding coordinates, LSTM represents the LSTM encoder, h t is the latent variable vector at time t, W4 is the model parameter of the LSTM encoder, F a is the predicted coding vector, W5 is the parameter to be trained for the fully connected layer when outputting the predicted coding vector;

[0073] Obtain the set of candidate trajectory end points, project the predicted coding vector and the set of candidate trajectory end points to obtain a query matrix, a key matrix, and a value matrix, perform weighted summation according to the query matrix, the key matrix, and the value matrix to obtain a weighted fusion result, calculate a probability value based on the weighted fusion result, store the probability value to obtain a sequence of probability values, traverse the sequence of probability values to obtain the maximum probability value, and determine the predicted point coordinates according to the maximum probability value;

[0074] The calculation formula for the maximum probability value is:

[0075]

[0076] where, Q is the query matrix, K is the key matrix, V is the value matrix, A is the weighted fusion result, Ti is the probability value of the i-th candidate trajectory end point, G is the set of candidate trajectory end points, W Q 、W k 、W v are the first projection matrix, the second projection matrix, and the third projection matrix respectively, d is the number of matrix columns, Gi is the i-th candidate trajectory end point in the set of candidate trajectory end points, n is a variable parameter, and N is the total number of candidate trajectory end points in the set of candidate trajectory end points.

[0077] Specifically, the calculation of the obstacle avoidance path according to the starting point coordinates, the ending point coordinates, the centroid coordinates, and the predicted point coordinates through the avoidance algorithm specifically includes:

[0078] Obtain the point cloud coordinates of the static point cloud cluster, and determine the static obstacle interval radius according to the centroid coordinates and the point cloud coordinates of the static point cloud cluster;

[0079] The calculation formula for the static obstacle interval radius is:

[0080]

[0081] where, R is the static obstacle interval coordinate, (X i , Y i , Z i(X c , Y c , Z c ) is the centroid coordinate, and i is a dynamic variable;

[0082] Obtain the initial point cloud coordinates and initial centroid coordinates of the dynamic point cloud cluster, and determine the dynamic obstacle interval radius according to the initial point cloud coordinates of the dynamic point cloud cluster and the initial centroid coordinates;

[0083] Determine the moving interval radius and the center point according to the dynamic obstacle interval radius, the initial centroid coordinates, and the predicted point coordinates;

[0084] The calculation formula of the moving interval radius is:

[0085]

[0086] where R1 is the moving interval radius, r is the dynamic obstacle interval radius, (X1, Y1, Z1) is the initial centroid coordinate, and (X2, Y2, Z2) is the predicted point coordinate;

[0087] Obtain the grid map, label the grid map according to the centroid coordinates, the static obstacle interval radius, the moving interval radius, and the center point to obtain the cost map, and calculate the obstacle avoidance path through the path planning algorithm according to the cost map, the starting point coordinates, and the ending point coordinates.

[0088] In this embodiment, the path planning algorithm used is the DBO-GWO algorithm.

[0089] It should be noted that the calculation formulas of the dynamic obstacle interval radius and the static obstacle interval radius are the same.

[0090] A general aviation monitoring and warning system based on artificial intelligence includes:

[0091] An acquisition module, configured to collect point cloud data through an airborne lidar, align the point cloud data according to the hardware timestamp to obtain a point cloud data sequence, perform point cloud preprocessing on the point cloud data sequence to obtain a point cloud cluster sequence, and label the point cloud cluster sequence through a matching algorithm to obtain a labeled point cloud cluster sequence;

[0092] A processing module is configured to preset a time window, intercept a sequence of target point cloud clusters according to the time window to obtain a sequence of test point cloud clusters, calculate a state vector of each point cloud cluster in the sequence of test point cloud clusters, construct a sequence of state vectors according to the state vectors, calculate a dynamic probability according to the sequence of state vectors through a classification model, preset a comparison threshold, judge the dynamic probability according to the comparison threshold to obtain a judgment result, and classify the point cloud clusters according to the judgment result to obtain dynamic point cloud clusters and static point cloud clusters;

[0093] An avoidance module is configured to obtain the centroid coordinates of the static point cloud clusters and a sequence of centroid coordinates of the dynamic point cloud clusters, calculate predicted point coordinates according to the sequence of centroid coordinates through a prediction model, obtain the starting coordinates and ending coordinates of the aircraft, and calculate an obstacle avoidance path according to the starting coordinates, the ending coordinates, the centroid coordinates, and the predicted point coordinates through an avoidance algorithm.

[0094] An electronic device includes a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method for general aviation monitoring and early warning based on artificial intelligence as described above is implemented.

[0095] A storage medium containing computer-executable instructions, the computer-executable instructions are used to execute the method for general aviation monitoring and early warning based on artificial intelligence as described above when executed by a computer processor.

[0096] The computer storage medium of the embodiments of the present invention may adopt any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (a non-exhaustive list) of the computer-readable storage medium include: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the computer-readable storage medium may be any tangible medium that contains or stores a program, and the program may be used by or in combination with an instruction execution system, apparatus, or device.

[0097] A computer-readable signal medium can include a data signal propagated in a baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device.

[0098] The program code contained on a computer-readable medium can be transmitted using any appropriate medium, including but not limited to wireless, wire, optical fiber cable, RF, etc., or any suitable combination of the above. The computer program code for performing the operations of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., by using an Internet service provider to connect through the Internet).

[0099] As described above, it is only the preferred embodiment of the present invention, and there is no limitation in any form to the present invention. Although the present invention has been disclosed as above with the preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments with equivalent changes within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A general aviation monitoring and early warning method based on artificial intelligence, characterized in that, Including: Collecting point cloud data through airborne lidar, aligning the point cloud data according to the hardware timestamp to obtain a point cloud data sequence, obtaining a point cloud cluster sequence through point cloud preprocessing based on the point cloud data sequence, and obtaining a target point cloud cluster sequence through a matching algorithm based on the point cloud cluster sequence; A preset time window, intercepting the target point cloud cluster sequence according to the time window to obtain a test point cloud cluster sequence, calculating the state vector of each point cloud cluster in the test point cloud cluster sequence, constructing a state vector sequence based on the state vector, calculating a dynamic probability through a classification model based on the state vector sequence, presetting a comparison threshold, judging the dynamic probability according to the comparison threshold to obtain a judgment result, and classifying the point cloud cluster according to the judgment result to obtain a dynamic point cloud cluster and a static point cloud cluster; Obtaining the centroid coordinates of the static point cloud cluster and the centroid coordinate sequence of the dynamic point cloud cluster, calculating the predicted point coordinates through a prediction model based on the centroid coordinate sequence, obtaining the starting coordinates and ending coordinates of the aircraft, and calculating an obstacle avoidance path through an avoidance algorithm based on the starting coordinates, the ending coordinates, the centroid coordinates, and the predicted point coordinates.

2. The general aviation monitoring and early warning method based on artificial intelligence according to claim 1, characterized in that The specific steps of obtaining the point cloud cluster sequence through point cloud preprocessing based on the point cloud data sequence include: Traversing the point cloud data sequence to obtain a point cloud frame, removing noise points from the point cloud frame through a noise reduction algorithm to obtain a noise-reduced point cloud frame, performing point cloud clustering on the noise-reduced point cloud frame through a segmentation algorithm to obtain a point cloud cluster set, and integrating each frame of the point cloud cluster set to obtain the point cloud cluster sequence.

3. The general aviation monitoring and early warning method based on artificial intelligence according to claim 1, characterized in that, The specific calculation steps of the matching algorithm include: Obtaining the morphological parameter vector of each point cloud cluster in the point cloud cluster sequence, establishing two empty vector lists, and storing the morphological parameter vectors in two consecutive frames of the point cloud cluster set into the empty vector lists respectively to obtain a first vector list and a second vector list; Obtaining the point cloud cluster similarity through a similarity calculation formula based on the first vector list and the second vector list, constructing a similarity matrix based on the point cloud cluster similarity, traversing the similarity matrix to obtain the maximum similarity value, determining a point cloud cluster matching pair according to the maximum similarity value, obtaining a target matching pair by targeting the point cloud cluster matching pair, and storing the target matching pair to obtain a target point cloud cluster sequence.

4. The general aviation monitoring and early warning method based on artificial intelligence according to claim 1, characterized in that The expression of the state vector is: A = (speed t , a t , b t , c t , w t , e t ), Among them, A is the state vector, speed t represents the motion speed of the point cloud cluster at time t, a t , b t , c t represent the centroid coordinates of the point cloud cluster at time t, w t represents the motion direction of the point cloud cluster at time t, e t represents the acceleration of the point cloud cluster at time t.

5. The general aviation monitoring and warning method based on artificial intelligence according to claim 1, wherein The classification model includes an input layer, an attention layer, and a fully connected layer. The specific calculation steps include: Obtaining a standardized feature vector sequence through standardization based on the state vector sequence, the input layer transmitting the standardized feature vector sequence to the attention layer for calculation to obtain a context vector, and calculating the dynamic probability through the fully connected layer based on the context vector.

6. The general aviation monitoring and early warning method based on artificial intelligence according to claim 1, characterized in that The specific calculation steps of the prediction model include: Performing mapping on the centroid coordinate sequence to obtain coordinate coding information, calculating a hidden variable vector through an LSTM encoder based on the coordinate coding information, and calculating a predicted coding vector through a fully connected layer based on the hidden variable vector. Obtain the set of candidate trajectory endpoints, project the predicted encoding vector and the set of candidate trajectory endpoints to obtain a query matrix, a key matrix, and a value matrix, perform weighted summation according to the query matrix, the key matrix, and the value matrix to obtain a weighted fusion result, calculate a probability value based on the weighted fusion result, store the probability value to obtain a sequence of probability values, traverse the sequence of probability values to obtain the maximum probability value, and determine the predicted point coordinates according to the maximum probability value.

7. The general aviation monitoring and early warning method based on artificial intelligence according to claim 1, wherein, The specific calculation steps of the avoidance algorithm include: Obtain the point cloud coordinates of the static point cloud cluster, and determine the static obstacle interval radius according to the centroid coordinates and the point cloud coordinates of the static point cloud cluster; Obtain the initial point cloud coordinates and the initial centroid coordinates of the dynamic point cloud cluster, and determine the dynamic obstacle interval radius according to the initial point cloud coordinates and the initial centroid coordinates of the dynamic point cloud cluster; Determine the moving interval radius and the center point according to the dynamic obstacle interval radius, the initial centroid coordinates, and the predicted point coordinates; Obtain a grid map, label the grid map according to the centroid coordinates, the static obstacle interval radius, the moving interval radius, and the center point to obtain a cost map, and calculate the obstacle avoidance path through a path planning algorithm according to the cost map, the starting point coordinates, and the ending point coordinates.

8. A general aviation monitoring and early warning system based on artificial intelligence, characterized in that, Include: A collection module for collecting point cloud data through an airborne lidar, aligning the point cloud data according to the hardware timestamp to obtain a sequence of point cloud data, performing point cloud preprocessing on the sequence of point cloud data to obtain a sequence of point cloud clusters, and labeling the sequence of point cloud clusters through a matching algorithm to obtain a sequence of labeled point cloud clusters; A processing module for presetting a time window, intercepting the sequence of labeled point cloud clusters according to the time window to obtain a sequence of test point cloud clusters, calculating the state vector of each point cloud cluster in the sequence of test point cloud clusters, constructing a sequence of state vectors according to the state vectors, calculating a dynamic probability through a classification model according to the sequence of state vectors, presetting a comparison threshold, judging the dynamic probability according to the comparison threshold to obtain a judgment result, and classifying the point cloud clusters according to the judgment result to obtain dynamic point cloud clusters and static point cloud clusters; An avoidance module for obtaining the centroid coordinates of the static point cloud cluster and the sequence of centroid coordinates of the dynamic point cloud cluster, calculating the predicted point coordinates through a prediction model according to the sequence of centroid coordinates, obtaining the starting coordinates and the ending coordinates of the aircraft, and calculating the obstacle avoidance path through an avoidance algorithm according to the starting point coordinates, the ending point coordinates, the centroid coordinates, and the predicted point coordinates.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the general aviation monitoring and early warning method based on artificial intelligence according to any one of claims 1-7.

10. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions are used to execute the general aviation monitoring and early warning method based on artificial intelligence according to any one of claims 1-7 when executed by a computer processor.

Images