Low-altitude unmanned aerial vehicle intelligent supervision system and method based on multi-source fusion

Abstract

The invention discloses a low-altitude unmanned aerial vehicle intelligent supervision system and method based on multi-source fusion, and relates to the technical field of unmanned aerial vehicle supervision, and the supervision method comprises the following steps: capturing each flight path of an unmanned aerial vehicle, and generating a corresponding flight record; collecting flight data, confirming a flight scene and performing anomaly evaluation; acquiring an expected flight route of any flight record, and analyzing the deviation condition of the flight record; carrying out abnormal feature extraction on flight data of any flight record, and analyzing the influence condition of each abnormal feature on the deviation condition; setting corresponding deviation alarm thresholds for different flight scenes according to the abnormal feature inclusion condition of any flight record; generating a real-time flight record whenever the unmanned aerial vehicle flies in real time, and confirming a deviation alarm threshold value of the real-time flight record; and analyzing the real-time deviation condition of the flight route, and carrying out anomaly identification on the flight route of the real-time flight record.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) supervision, and in particular to a low-altitude UAV intelligent supervision system and method based on multi-source fusion. Background Art

[0002] With the rapid development of the low-altitude economy, the application scale of drones in various fields continues to expand, but the accompanying flight safety issues are also becoming increasingly prominent. The existing regulatory system generally uses fixed thresholds for flight deviation warnings, which has significant limitations.

[0003] Fixed thresholds cannot adapt to different flight environments. For example, a drone's brief deviation from its route when avoiding obstacles may be misjudged as an anomaly, while a small deviation at the edge of a no-fly zone may be ignored. At the same time, the existing system lacks intelligent analysis of historical flight data, making it difficult to identify long-standing systematic deviations or reasonable deviation patterns in specific scenarios. It is unable to dynamically adjust the alarm threshold according to the real-time environment, resulting in rigid regulatory strategies. Summary of the Invention

[0004] The purpose of the present invention is to provide a low-altitude UAV intelligent supervision system and method based on multi-source fusion to solve the problems raised in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a low-altitude UAV intelligent supervision method based on multi-source fusion, the supervision method comprising the following steps:

[0006] Step S100: The monitoring device installed on the UAV captures each flight route of the UAV and generates a corresponding flight record; the flight data of each flight record is collected, the flight scene of any flight record is confirmed, and any flight record is evaluated for abnormalities;

[0007] Step S200: Obtain the expected flight path of any flight record, compare the expected flight path with the actual flight path, and analyze the deviation of the flight record; extract abnormal features from the flight data of any flight record based on the flight abnormality assessment result of the flight record;

[0008] Step S300: Based on the deviations in any flight record, analyze the impact of each abnormal feature on the deviations to obtain the impact value of each abnormal feature; and set corresponding deviation alarm thresholds for different flight scenarios based on the inclusion of abnormal features in any flight record;

[0009] Step S400: Whenever the UAV flies in real time, a real-time flight record is generated, the flight scene of the real-time flight record is analyzed, and the deviation alarm threshold of the real-time flight record is confirmed; the real-time flight route of the UAV is captured, the real-time deviation of the flight route is analyzed, and the flight route of the real-time flight record is identified as abnormal.

[0010] Furthermore, step S100 includes the following steps:

[0011] Step S101: A plurality of monitoring devices are installed on the drone, and a positioning device included therein is selected. The coordinate data of the drone is collected at each unit time interval, and the coordinate data of the drone during flight are connected to obtain the flight path of the drone during flight;

[0012] Step S102: The monitoring data monitored by the remaining monitoring devices during the flight are collected, and the monitoring data of each monitoring device is set as dimensional data of a dimension. The dimensional data of each dimension are aggregated to obtain the flight data of the UAV; the flight route of the UAV is combined with the flight data to generate a flight record corresponding to the flight process; the dimensional data of different dimensions include flight speed fluctuation frequency, attitude angle change, signal stability, flight route completion, etc., which can reflect the flight status of the UAV;

[0013] Step S103: arbitrarily select a flight record, arbitrarily select dimension data of a dimension from the flight data of the selected flight record, perform feature extraction on the dimension data to obtain a plurality of features of the selected dimension; preset a flight scene database and store a plurality of flight scenes, wherein each flight scene is matched with a plurality of scene features; arbitrarily select a flight scene, compare the plurality of scene features of the selected flight scene with a plurality of features of each dimension; if any of the scene features have the same features as the selected flight scene, then it is determined that the UAV in the selected flight record is in the selected flight scene; the flight scenes include open airspace such as grasslands and deserts, complex urban environments, confined spaces, and adverse weather conditions;

[0014] Step S104: Obtain the dimension where the scene feature is located in the selected flight record, preset corresponding evaluation rules for each dimension except the dimension, evaluate the dimensional data of each dimension to obtain a corresponding evaluation value, and take the average of the evaluation values ​​of each dimension to obtain the evaluation value of the selected flight record; preset an abnormal evaluation threshold, if the evaluation value of the selected flight record is lower than the abnormal evaluation threshold, the selected flight record is set as an abnormal flight record; since the dimensions of the UAV include flight speed fluctuation frequency, attitude angle change, signal stability, flight route completion, etc., taking the flight speed fluctuation frequency as an example, the flight quality is reflected by calculating the deviation degree between the actual fluctuation frequency and the expected fluctuation frequency. Similarly, the deviation degree of the attitude angle change and the deviation degree of the signal stability value can also be used.

[0015] Furthermore, step S200 includes the following steps:

[0016] Step S201: randomly selecting a flight record, setting the flight route captured in the selected flight record as the actual flight route, and constructing a three-dimensional coordinate system to present the expected flight route and the actual flight route of the selected flight record in the three-dimensional coordinate system respectively;

[0017] Step S202: Set the initial time point of the two flight routes to t0, obtain the unit time interval between two adjacent time points in the actual flight route as Δt, and then obtain the i-th time point as t(i) = t0 + i × Δt; divide the expected flight route according to the time point to obtain the expected position coordinate A of the i-th time point ’ t(i) , get the actual position coordinate A of the i-th time point in the actual flight route t(i) , the distance difference between the two position coordinates at the i-th time point is calculated to be ΔA t(i) ;

[0018] Step S203: Set the instantaneous flight direction at each time point in the expected flight path to 0°, obtain the instantaneous flight directions of the two flight paths at the i-th time point, and use the initial position coordinates as the vertex to obtain the angle Δθ between the actual flight path and the expected flight path at the i-th time point. t(i) , if the position coordinates of the actual flight path at the i-th time point are in the clockwise direction of the expected flight path, then Δθ t(i) is a positive number, otherwise it is a negative number; according to the formula:

[0019]

[0020] Where i is a positive integer and i∈[1,m], m is the number of time points of the actual flight route; the deviation value P of the actual flight route in the selected flight record is calculated; if the selected flight record is an abnormal flight record, the deviation value P is set as the abnormal deviation value; the deviation of the drone's flight route is reflected by the difference between the actual position and the expected position at each time point, as well as the deviation of the flight angle. However, considering that the drone is allowed to return to the expected route in time after deviation, the deviation value will not continue to increase after the drone deviates from the route. Therefore, it is necessary to set the positive and negative values ​​of the angle to make immediate corrections to the deviation value. The absolute value after accumulation reflects the comprehensive deviation of the drone.

[0021] Step S204: Obtain the deviation value of each flight record, divide the deviation value of each flight into a normal deviation value set and an abnormal deviation value set according to whether it is an abnormal deviation value, and obtain the minimum abnormal deviation value P in the abnormal deviation value set. min ; If the deviation value of the selected flight record P≥P min , then several features of each dimension in the selected flight record are set as abnormal features, and the abnormal feature set of the selected flight record is obtained; due to different flight scenarios, the allowed deviations will be different, but using a single threshold to reflect the abnormality will result in misjudgment. Therefore, in each flight record that exceeds the minimum abnormal deviation value, although the deviation value is high, there is a normal flight situation. Therefore, it is necessary to further analyze the abnormal features that affect the deviation value to facilitate the subsequent feature analysis.

[0022] Furthermore, step S300 includes the following steps:

[0023] Step S301: arbitrarily selecting a flight scenario, obtaining all flight records in the selected flight scenario to obtain a target flight record set, arbitrarily selecting an abnormal flight record from the target flight record set and setting it as a target abnormal record, and extracting the abnormal feature set and abnormal deviation value of the target abnormal record;

[0024] Step S302: except for the target abnormal record, a target flight record is reselected from the target flight record set. If the target flight record is not an abnormal flight record, the deviation value and several features of the target flight record are extracted. If the deviation value of the target flight record is an abnormal deviation value, each abnormal feature in the abnormal feature set of the target abnormal record is compared with several features of the target flight record to obtain several identical features and eliminate each identical feature in the abnormal feature set. If the deviation value of the target flight record is not an abnormal deviation value, each identical feature in the abnormal feature set is retained and the remaining abnormal features are eliminated. If the target flight record is an abnormal flight record, the target abnormal record is merged with the two abnormal feature sets of the target flight record. If there is an abnormal feature of numerical type, the abnormal feature values ​​under the same dimension are merged to obtain an abnormal numerical interval and the interval is matched with the remaining abnormal features of the same dimension.

[0025] Compare normal flight records with abnormal flight records. If the deviation value of the normal flight record is an abnormal deviation value, it means that the same feature is not the abnormal feature that causes the difference in results and needs to be eliminated. Similarly, if the deviation value of the normal flight record is a normal deviation value, the same feature is actually the abnormal feature that affects the results.

[0026] Step S303: Correct the abnormal feature set after comparing any two target flight records in the target flight record set, and summarize the corrected abnormal feature sets to obtain the abnormal feature set of the selected flight scene; arbitrarily select an abnormal feature from the abnormal feature set of the selected flight scene, obtain the abnormal numerical range corresponding to the selected abnormal feature, preset an expected numerical range (x1, x2) for the dimension where the selected abnormal feature is located, set the abnormal numerical range to (y1, y2), and calculate the deviation amplitude of the selected abnormal feature to be f according to the formula:

[0027]

[0028] Among them, Max() is a maximum value function; by taking the average value of the abnormal interval and the normal interval respectively, we can make a fair judgment on the deviation amplitude of the abnormal feature and obtain a more accurate result;

[0029] Step S304: Obtain each abnormal deviation value in the target flight record set and take the average value to obtain an average deviation value P ave , set the selected abnormal feature to be the kth abnormal feature in the abnormal feature set, and obtain the number of remaining abnormal features in the dimension where the kth abnormal feature is located, excluding the abnormal value interval, as b k , according to the formula:

[0030]

[0031] Where d is a positive integer and d∈[1,s], s is the number of abnormal features contained in the abnormal feature set, f k is the deviation amplitude of the kth abnormal feature, f d is the deviation amplitude of the dth abnormal feature; the influence value Y of the kth abnormal feature is calculated k Different abnormal features have different dimensions. The deviation amplitude can eliminate the dimension difference and effectively identify the impact of each abnormal feature.

[0032] Step S305: randomly select a target flight record from the target flight record set of the selected flight scene, obtain the deviation value of the selected target flight record at each time point, and set the maximum deviation value P at the i-th time point. t(i) ; Extract the abnormal features contained in the selected target flight record, and set the impact value of the j-th abnormal feature to Y j , according to the formula:

[0033]

[0034] Where j is a positive integer and j∈[1,r], r is the number of abnormal features of the selected target flight record; the comprehensive deviation value P of the selected target flight record is calculated ’ ;

[0035] Step S306: Obtain the comprehensive deviation value of each target flight record in the target flight record set, summarize the comprehensive deviation values ​​of abnormal flight records and extract a minimum comprehensive deviation value P ’ min , and then extract the maximum comprehensive deviation value P from the comprehensive deviation values ​​of the remaining target flight records ’ max , if P ’ min >P ’ max , then set the deviation alarm threshold P of the selected flight scene th =P ’ max Otherwise, P th =P ’ min .

[0036] Furthermore, step S400 includes the following steps:

[0037] Step S401: collecting flight data of the UAV during real-time flight, capturing the flight route of the UAV in real time, generating a real-time flight record, and updating the flight route in the real-time flight record in real time;

[0038] Step S402: Extract features from the flight data of the real-time flight record, compare features with each flight scene in the flight scene database, determine the flight scene of the real-time flight record, and obtain the deviation alarm threshold of the real-time flight record as P. th ;

[0039] Step S403: Obtain a set of abnormal features in the flight scene of the real-time flight record; compare each feature of the real-time flight record with the abnormal features to obtain several actual abnormal features of the real-time flight record; obtain the real-time position coordinates of the real-time flight record at the current moment, and obtain the deviation value of the drone at the current moment as P now , and accumulate the impact values ​​of each actual abnormal feature to obtain the comprehensive deviation value P of the real-time flight record ’ now If P ’ now >P th , then send an abnormal reminder for the current drone's flight status.

[0040] In order to better implement the above method, a low-altitude UAV intelligent supervision system is also proposed. The supervision system includes a historical route collection module, a route deviation analysis module, an expected deviation adjustment module and a real-time anomaly analysis module;

[0041] The historical route collection module is used to capture each flight route of the drone by installing monitoring equipment on the drone and generate corresponding flight records; collect flight data for each flight record, confirm the flight scene of any flight record and perform anomaly assessment on any flight record;

[0042] The route deviation analysis module is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; and extract abnormal features from the flight data of any flight record based on the flight anomaly assessment results of any flight record;

[0043] The expected deviation adjustment module is used to analyze the impact of various abnormal features on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; according to the inclusion of abnormal features in any flight record, corresponding deviation alarm thresholds are set for different flight scenarios;

[0044] The real-time anomaly analysis module is used to generate a real-time flight record every time the drone flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; capture the real-time flight route of the drone, analyze the real-time deviation of the flight route, and identify anomalies in the flight route of the real-time flight record.

[0045] Furthermore, the historical route collection module includes a flight route collection unit and a flight quality assessment unit;

[0046] The flight route collection unit is used to capture each flight route of the drone by installing monitoring equipment on the drone and generate corresponding flight records; the flight quality assessment unit is used to collect flight data of each flight record, confirm the flight scene of any flight record and perform abnormality assessment on any flight record.

[0047] Furthermore, the route deviation analysis module includes a route deviation identification unit and an abnormal feature extraction unit;

[0048] The route deviation identification unit is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; the abnormal feature extraction unit is used to extract abnormal features of the flight data of any flight record based on the flight abnormality assessment results of any flight record.

[0049] Furthermore, the expected deviation adjustment module includes a feature deviation association unit and a deviation threshold division unit;

[0050] The feature deviation association unit is used to analyze the impact of each abnormal feature on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; the deviation threshold division unit is used to set corresponding deviation alarm thresholds for different flight scenarios according to the inclusion of abnormal features in any flight record.

[0051] Furthermore, the real-time anomaly analysis module includes a real-time deviation confirmation unit and an abnormal flight identification unit;

[0052] The real-time deviation confirmation unit is used to generate a real-time flight record every time the UAV flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; the abnormal flight identification unit is used to capture the real-time flight route of the UAV, analyze the real-time deviation of the flight route, and identify abnormalities in the flight route of the real-time flight record.

[0053] Compared with the prior art, the present invention has the following beneficial effects:

[0054] 1. This invention dynamically adjusts the abnormality alarm thresholds of drones in different flight scenarios by analyzing the abnormal characteristics and deviation effects in historical flight records. This solves the problem of misjudgment caused by the fixed thresholds used in traditional abnormality supervision, significantly reduces the probability of false detection, and improves supervision efficiency.

[0055] 2. The present invention installs a variety of monitoring devices on the drone to comprehensively analyze the drone's flight status. Through feature recognition and scene matching, it can accurately identify the drone's flight scene and any abnormal features at any time, and accurately grasp the drone's flight status in different environments.

[0056] 3. The present invention makes real-time judgments on the abnormal alarm thresholds of real-time flight data and the flight scene, calculates the comprehensive deviation value based on the impact value of the abnormal characteristics, and promptly identifies abnormal route deviations, which can effectively prevent the occurrence of flight accidents and improve the overall safety of low-altitude drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of the steps of a low-altitude UAV intelligent supervision method based on multi-source fusion;

[0058] Figure 2 This is a structural diagram of a low-altitude UAV intelligent supervision system based on multi-source fusion. DETAILED DESCRIPTION

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0060] Example: Figures 1 to 2 As shown, the present invention provides a low-altitude UAV intelligent supervision method based on multi-source fusion, and the supervision method includes the following steps:

[0061] Step S100: The monitoring device installed on the UAV captures each flight route of the UAV and generates a corresponding flight record; the flight data of each flight record is collected, the flight scene of any flight record is confirmed, and any flight record is evaluated for abnormalities;

[0062] Wherein, step S100 includes the following steps:

[0063] Step S101: A plurality of monitoring devices are installed on the drone, and a positioning device included therein is selected. The coordinate data of the drone is collected at each unit time interval, and the coordinate data of the drone during flight are connected to obtain the flight path of the drone during flight;

[0064] Step S102: Collect the monitoring data collected by the remaining monitoring devices during the flight process, set the monitoring data of each monitoring device as dimensional data of one dimension, and aggregate the dimensional data of each dimension to obtain the flight data of the drone; combine the flight route of the drone with the flight data to generate a flight record corresponding to the flight process;

[0065] Step S103: arbitrarily select a flight record, arbitrarily select dimension data of a dimension from the flight data of the selected flight record, perform feature extraction on the dimension data to obtain a plurality of features of the selected dimension; preset a flight scene database and store a plurality of flight scenes, wherein each flight scene is matched with a plurality of scene features; arbitrarily select a flight scene, compare the plurality of scene features of the selected flight scene with the plurality of features of each dimension; if any scene feature has the same feature as the selected scene feature, then it is determined that the UAV in the selected flight record is in the selected flight scene;

[0066] Step S104: Obtain the dimension in which the scene feature is located in the selected flight record, preset corresponding evaluation rules for each dimension except the dimension, evaluate the dimensional data of each dimension to obtain a corresponding evaluation value, and take the average of the evaluation values ​​of each dimension to obtain the evaluation value of the selected flight record; preset an abnormal evaluation threshold, if the evaluation value of the selected flight record is lower than the abnormal evaluation threshold, the selected flight record will be set as an abnormal flight record.

[0067] Step S200: Obtain the expected flight path of any flight record, compare the expected flight path with the actual flight path, and analyze the deviation of the flight record; extract abnormal features from the flight data of any flight record based on the flight abnormality assessment result of the flight record;

[0068] Wherein, step S200 includes the following steps:

[0069] Step S201: randomly selecting a flight record, setting the flight route captured in the selected flight record as the actual flight route, and constructing a three-dimensional coordinate system to present the expected flight route and the actual flight route of the selected flight record in the three-dimensional coordinate system respectively;

[0070] Step S202: Set the initial time point of the two flight routes to t0, obtain the unit time interval between two adjacent time points in the actual flight route as Δt, and then obtain the i-th time point as t(i) = t0 + i × Δt; divide the expected flight route according to the time point to obtain the expected position coordinate A of the i-th time point ’ t(i) , get the actual position coordinate A of the i-th time point in the actual flight route t(i) , the distance difference between the two position coordinates at the i-th time point is calculated to be ΔA t((i) ;

[0071] Step S203: Set the instantaneous flight direction at each time point in the expected flight path to 0°, obtain the instantaneous flight directions of the two flight paths at the i-th time point, and use the initial position coordinates as the vertex to obtain the angle Δθ between the actual flight path and the expected flight path at the i-th time point. t(i) , if the position coordinates of the actual flight path at the i-th time point are in the clockwise direction of the expected flight path, then Δθ t(i) is a positive number, otherwise it is a negative number; according to the formula:

[0072]

[0073] Where i is a positive integer and i∈[1,m], m is the number of time points of the actual flight route; the deviation value P of the actual flight route in the selected flight record is calculated; if the selected flight record is an abnormal flight record, the deviation value P is set as the abnormal deviation value;

[0074] Example 1: Assume that a desired flight path of a drone is divided into four time points, and the actual flight path is divided into four time points. The distance differences between the position coordinates of the two paths at each time point are 1, 2, 3, and 4, respectively, and the angles of the two sides at each time point are 60, 30, -30, and -45, respectively. The deviation value of the actual flight path is calculated to be P = |60×1+30×2-30×3-45×4| = 90;

[0075] Step S204: Obtain the deviation value of each flight record, divide the deviation value of each flight into a normal deviation value set and an abnormal deviation value set according to whether it is an abnormal deviation value, and obtain the minimum abnormal deviation value P in the abnormal deviation value set. min ; If the deviation value of the selected flight record P≥P min , then several features of each dimension in the selected flight record are set as abnormal features to obtain the abnormal feature set of the selected flight record.

[0076] Step S300: Based on the deviations in any flight record, analyze the impact of each abnormal feature on the deviations to obtain the impact value of each abnormal feature; and set corresponding deviation alarm thresholds for different flight scenarios based on the inclusion of abnormal features in any flight record;

[0077] Wherein, step S300 includes the following steps:

[0078] Step S301: arbitrarily selecting a flight scenario, obtaining all flight records in the selected flight scenario to obtain a target flight record set, arbitrarily selecting an abnormal flight record from the target flight record set and setting it as a target abnormal record, and extracting the abnormal feature set and abnormal deviation value of the target abnormal record;

[0079] Step S302: except for the target abnormal record, a target flight record is reselected from the target flight record set. If the target flight record is not an abnormal flight record, the deviation value and several features of the target flight record are extracted. If the deviation value of the target flight record is an abnormal deviation value, each abnormal feature in the abnormal feature set of the target abnormal record is compared with several features of the target flight record to obtain several identical features and eliminate each identical feature in the abnormal feature set. If the deviation value of the target flight record is not an abnormal deviation value, each identical feature in the abnormal feature set is retained and the remaining abnormal features are eliminated. If the target flight record is an abnormal flight record, the target abnormal record is merged with the two abnormal feature sets of the target flight record. If there is an abnormal feature of numerical type, the abnormal feature values ​​under the same dimension are merged to obtain an abnormal numerical interval and the interval is matched with the remaining abnormal features of the same dimension.

[0080] Step S303: Correct the abnormal feature set after comparing any two target flight records in the target flight record set, and summarize the corrected abnormal feature sets to obtain the abnormal feature set of the selected flight scene; arbitrarily select an abnormal feature from the abnormal feature set of the selected flight scene, obtain the abnormal numerical range corresponding to the selected abnormal feature, preset an expected numerical range (x1, x2) for the dimension where the selected abnormal feature is located, set the abnormal numerical range to (y1, y2), and calculate the deviation amplitude of the selected abnormal feature to be f according to the formula:

[0081]

[0082] Among them, Max() is the maximum value function;

[0083] Example 2: Assume that the expected value interval corresponding to an abnormal feature is (14, 20). If the abnormal value interval is (24, 30), then f = |17-27| / (30-14) = 0.625. If the abnormal value interval is (10, 12), then f = |11-17| / (20-10) = 0.6.

[0084] Step S304: Obtain each abnormal deviation value in the target flight record set and take the average value to obtain an average deviation value P ave , set the selected abnormal feature to be the kth abnormal feature in the abnormal feature set, and obtain the number of remaining abnormal features in the dimension where the kth abnormal feature is located, excluding the abnormal value interval, as b k , according to the formula:

[0085]

[0086] Where d is a positive integer and d∈[1,s], s is the number of abnormal features contained in the abnormal feature set, f k is the deviation amplitude of the kth abnormal feature, f d is the deviation amplitude of the dth abnormal feature; the influence value Y of the kth abnormal feature is calculated k ;

[0087] Step S305: randomly select a target flight record from the target flight record set of the selected flight scene, obtain the deviation value of the selected target flight record at each time point, and set the maximum deviation value P at the i-th time point. t(i) ; Extract the abnormal features contained in the selected target flight record, and set the impact value of the j-th abnormal feature to Y j , according to the formula:

[0088]

[0089] Where j is a positive integer and j∈[1,r], r is the number of abnormal features of the selected target flight record; the comprehensive deviation value P of the selected target flight record is calculated ’ ;

[0090] Step S306: Obtain the comprehensive deviation value of each target flight record in the target flight record set, summarize the comprehensive deviation values ​​of abnormal flight records and extract a minimum comprehensive deviation value P ’ min , and then extract the maximum comprehensive deviation value P from the comprehensive deviation values ​​of the remaining target flight records ’ max , if P ’ min >P ’ max, then set the deviation alarm threshold P of the selected flight scene th =P ’ max Otherwise, P th =P ’ min .

[0091] Step S400: Whenever the UAV flies in real time, a real-time flight record is generated, the flight scene of the real-time flight record is analyzed, and the deviation alarm threshold of the real-time flight record is determined; the real-time flight route of the UAV is captured, the real-time deviation of the flight route is analyzed, and anomalies are identified in the flight route of the real-time flight record;

[0092] Step S400 includes the following steps:

[0093] Step S401: collecting flight data of the UAV during real-time flight, capturing the flight route of the UAV in real time, generating a real-time flight record, and updating the flight route in the real-time flight record in real time;

[0094] Step S402: Extract features from the flight data of the real-time flight record, compare features with each flight scene in the flight scene database, determine the flight scene of the real-time flight record, and obtain the deviation alarm threshold of the real-time flight record as P. th ;

[0095] Step S403: Obtain a set of abnormal features in the flight scene of the real-time flight record; compare each feature of the real-time flight record with the abnormal features to obtain several actual abnormal features of the real-time flight record; obtain the real-time position coordinates of the real-time flight record at the current moment, and obtain the deviation value of the drone at the current moment as P now , and accumulate the impact values ​​of each actual abnormal feature to obtain the comprehensive deviation value P of the real-time flight record ’ now If P ’ now >P th , then send an abnormal reminder for the current drone's flight status.

[0096] An intelligent monitoring system for low-altitude UAVs, comprising a historical route collection module, a route deviation analysis module, an expected deviation adjustment module, and a real-time anomaly analysis module;

[0097] The historical route collection module is used to capture each flight route of the drone by installing monitoring equipment on the drone and generate corresponding flight records; collect flight data for each flight record, confirm the flight scene of any flight record and perform anomaly assessment on any flight record;

[0098] The route deviation analysis module is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; and extract abnormal features from the flight data of any flight record based on the flight anomaly assessment results of any flight record;

[0099] The expected deviation adjustment module is used to analyze the impact of various abnormal features on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; according to the inclusion of abnormal features in any flight record, corresponding deviation alarm thresholds are set for different flight scenarios;

[0100] The real-time anomaly analysis module is used to generate a real-time flight record every time the drone flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; capture the real-time flight route of the drone, analyze the real-time deviation of the flight route, and identify anomalies in the flight route of the real-time flight record.

[0101] Among them, the historical route collection module includes a flight route collection unit and a flight quality assessment unit;

[0102] The flight route collection unit is used to capture each flight route of the drone by installing monitoring equipment on the drone and generate corresponding flight records; the flight quality assessment unit is used to collect flight data of each flight record, confirm the flight scene of any flight record and perform abnormality assessment on any flight record.

[0103] Among them, the route deviation analysis module includes a route deviation identification unit and an abnormal feature extraction unit;

[0104] The route deviation identification unit is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; the abnormal feature extraction unit is used to extract abnormal features of the flight data of any flight record based on the flight abnormality assessment results of any flight record.

[0105] The expected deviation adjustment module includes a feature deviation association unit and a deviation threshold division unit;

[0106] The feature deviation association unit is used to analyze the impact of each abnormal feature on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; the deviation threshold division unit is used to set corresponding deviation alarm thresholds for different flight scenarios according to the inclusion of abnormal features in any flight record.

[0107] Among them, the real-time anomaly analysis module includes a real-time deviation confirmation unit and an abnormal flight identification unit;

[0108] The real-time deviation confirmation unit is used to generate a real-time flight record every time the UAV flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; the abnormal flight identification unit is used to capture the real-time flight route of the UAV, analyze the real-time deviation of the flight route, and identify abnormalities in the flight route of the real-time flight record.

[0109] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above and that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Therefore, the embodiments should be considered in all respects as illustrative and non-restrictive, and the scope of the invention is defined by the appended claims, not the foregoing description, and all variations within the meaning and range of equivalents of the claims are intended to be included therein. Any reference sign in a claim should not be construed as limiting the claim to which it relates.

Claims

1. A low-altitude UAV intelligent monitoring method based on multi-source fusion, characterized by: The supervision method comprises the following steps: Step S100: The monitoring device installed on the UAV captures each flight route of the UAV and generates a corresponding flight record; the flight data of each flight record is collected, the flight scene of any flight record is confirmed, and any flight record is evaluated for abnormalities; Step S200: Obtain the expected flight path of any flight record, compare the expected flight path with the actual flight path, and analyze the deviation of the flight record; extract abnormal features from the flight data of any flight record based on the flight abnormality assessment result of the flight record; Step S300: Based on the deviations in any flight record, analyze the impact of each abnormal feature on the deviations to obtain the impact value of each abnormal feature; and set corresponding deviation alarm thresholds for different flight scenarios based on the inclusion of abnormal features in any flight record; Step S400: Whenever the UAV flies in real time, a real-time flight record is generated, the flight scene of the real-time flight record is analyzed, and the deviation alarm threshold of the real-time flight record is confirmed; the real-time flight route of the UAV is captured, the real-time deviation of the flight route is analyzed, and the flight route of the real-time flight record is identified as abnormal.

2. The low-altitude UAV intelligent monitoring method based on multi-source fusion according to claim 1 is characterized by: The step S100 includes the following steps: Step S101: A plurality of monitoring devices are installed on the drone, and a positioning device included therein is selected. The coordinate data of the drone is collected at each unit time interval, and the coordinate data of the drone during flight are connected to obtain the flight path of the drone during flight; Step S102: Collect the monitoring data collected by the remaining monitoring devices during the flight process, set the monitoring data of each monitoring device as dimensional data of one dimension, and aggregate the dimensional data of each dimension to obtain the flight data of the drone; combine the flight route of the drone with the flight data to generate a flight record corresponding to the flight process; Step S103: arbitrarily select a flight record, arbitrarily select dimension data of a dimension from the flight data of the selected flight record, perform feature extraction on the dimension data to obtain a plurality of features of the selected dimension; preset a flight scene database and store a plurality of flight scenes, wherein each flight scene is matched with a plurality of scene features; arbitrarily select a flight scene, compare the plurality of scene features of the selected flight scene with the plurality of features of each dimension; if any scene feature has the same feature as the selected scene feature, then it is determined that the UAV in the selected flight record is in the selected flight scene; Step S104: Obtain the dimension in which the scene feature is located in the selected flight record, preset corresponding evaluation rules for each dimension except the dimension, evaluate the dimensional data of each dimension to obtain a corresponding evaluation value, and take the average of the evaluation values ​​of each dimension to obtain the evaluation value of the selected flight record; preset an abnormal evaluation threshold, if the evaluation value of the selected flight record is lower than the abnormal evaluation threshold, the selected flight record will be set as an abnormal flight record.

3. The low-altitude UAV intelligent monitoring method based on multi-source fusion according to claim 2 is characterized by: The step S200 includes the following steps: Step S201: randomly selecting a flight record, setting the flight route captured in the selected flight record as the actual flight route, and constructing a three-dimensional coordinate system to present the expected flight route and the actual flight route of the selected flight record in the three-dimensional coordinate system respectively; Step S202: Set the initial time point of the two flight routes to t0, obtain the unit time interval between two adjacent time points in the actual flight route as Δt, and then obtain the i-th time point as t(i) = t0 + i × Δt; divide the expected flight route according to the time point to obtain the expected position coordinate A of the i-th time point ’ t(i) , get the actual position coordinate A of the i-th time point in the actual flight route t(i) , the distance difference between the two position coordinates at the i-th time point is calculated to be ΔA t(i) ; Step S203: Set the instantaneous flight direction at each time point in the expected flight path to 0°, obtain the instantaneous flight directions of the two flight paths at the i-th time point, and use the initial position coordinates as the vertex to obtain the angle Δθ between the actual flight path and the expected flight path at the i-th time point. t(i) , if the position coordinates of the actual flight path at the i-th time point are in the clockwise direction of the expected flight path, then Δθ t(i) is a positive number, otherwise it is a negative number; according to the formula: Where i is a positive integer and i∈[1,m], m is the number of time points of the actual flight route; the deviation value P of the actual flight route in the selected flight record is calculated; if the selected flight record is an abnormal flight record, the deviation value P is set as the abnormal deviation value; Step S204: Obtain the deviation value of each flight record, divide the deviation value of each flight into a normal deviation value set and an abnormal deviation value set according to whether it is an abnormal deviation value, and obtain the minimum abnormal deviation value P in the abnormal deviation value set. min ; If the deviation value of the selected flight record P≥P min , then several features of each dimension in the selected flight record are set as abnormal features to obtain the abnormal feature set of the selected flight record.

4. The method for intelligent monitoring of low-altitude UAVs based on multi-source fusion according to claim 3 is characterized by: The step S300 includes the following steps: Step S301: arbitrarily selecting a flight scenario, obtaining all flight records in the selected flight scenario to obtain a target flight record set, arbitrarily selecting an abnormal flight record from the target flight record set and setting it as a target abnormal record, and extracting the abnormal feature set and abnormal deviation value of the target abnormal record; Step S302: except for the target abnormal record, a target flight record is reselected from the target flight record set. If the target flight record is not an abnormal flight record, the deviation value and several features of the target flight record are extracted. If the deviation value of the target flight record is an abnormal deviation value, each abnormal feature in the abnormal feature set of the target abnormal record is compared with several features of the target flight record to obtain several identical features and eliminate each identical feature in the abnormal feature set. If the deviation value of the target flight record is not an abnormal deviation value, each identical feature in the abnormal feature set is retained and the remaining abnormal features are eliminated. If the target flight record is an abnormal flight record, the target abnormal record is merged with the two abnormal feature sets of the target flight record. If there is an abnormal feature of numerical type, the abnormal feature values ​​under the same dimension are merged to obtain an abnormal numerical interval and the interval is matched with the remaining abnormal features of the same dimension. Step S303: Correct the abnormal feature set after comparing any two target flight records in the target flight record set, and summarize the corrected abnormal feature sets to obtain the abnormal feature set of the selected flight scene; arbitrarily select an abnormal feature from the abnormal feature set of the selected flight scene, obtain the abnormal numerical range corresponding to the selected abnormal feature, preset an expected numerical range (x1, x2) for the dimension where the selected abnormal feature is located, set the abnormal numerical range to (y1, y2), and calculate the deviation amplitude of the selected abnormal feature to be f according to the formula: Among them, Max() is the maximum value function; Step S304: Obtain each abnormal deviation value in the target flight record set and take the average value to obtain an average deviation value P ave , set the selected abnormal feature to be the kth abnormal feature in the abnormal feature set, and obtain the number of remaining abnormal features in the dimension where the kth abnormal feature is located, excluding the abnormal value interval, as b k , according to the formula: Where d is a positive integer and d∈[1,s], s is the number of abnormal features contained in the abnormal feature set, f k is the deviation amplitude of the kth abnormal feature, f d is the deviation amplitude of the dth abnormal feature; the influence value Y of the kth abnormal feature is calculated k ; Step S305: randomly select a target flight record from the target flight record set of the selected flight scene, obtain the deviation value of the selected target flight record at each time point, and set the maximum deviation value P at the i-th time point. t(i) ; Extract the abnormal features contained in the selected target flight record, and set the impact value of the j-th abnormal feature to Y j , according to the formula: Where j is a positive integer and j∈[1,r], r is the number of abnormal features of the selected target flight record; the comprehensive deviation value P of the selected target flight record is calculated ’ ; Step S306: Obtain the comprehensive deviation value of each target flight record in the target flight record set, summarize the comprehensive deviation values ​​of abnormal flight records and extract a minimum comprehensive deviation value P ’ min , and then extract the maximum comprehensive deviation value P from the comprehensive deviation values ​​of the remaining target flight records ’ max , if P ’ min >P ’ max , then set the deviation alarm threshold P of the selected flight scene th =P ’ max Otherwise, P th =P ’ min .

5. The low-altitude UAV intelligent monitoring method based on multi-source fusion according to claim 4 is characterized by: The step S400 includes the following steps: Step S401: collecting flight data of the UAV during real-time flight, capturing the flight route of the UAV in real time, generating a real-time flight record, and updating the flight route in the real-time flight record in real time; Step S402: Extract features from the flight data of the real-time flight record, compare features with each flight scene in the flight scene database, determine the flight scene of the real-time flight record, and obtain the deviation alarm threshold of the real-time flight record as P. th ; Step S403: Obtain a set of abnormal features in the flight scene of the real-time flight record; compare each feature of the real-time flight record with the abnormal features to obtain several actual abnormal features of the real-time flight record; obtain the real-time position coordinates of the real-time flight record at the current moment, and obtain the deviation value of the drone at the current moment as P now , and accumulate the impact values ​​of each actual abnormal feature to obtain the comprehensive deviation value P of the real-time flight record ’ now If P ’ now >P th , then send an abnormal reminder for the current drone's flight status.

6. A low-altitude UAV intelligent monitoring system, configured to implement the low-altitude UAV intelligent monitoring method based on multi-source fusion as claimed in any one of claims 1 to 5, characterized in that: The supervision system includes a historical route collection module, a route deviation analysis module, an expected deviation adjustment module and a real-time anomaly analysis module; The historical route collection module is used to capture each flight route of the drone by installing a monitoring device on the drone and generate corresponding flight records; collect flight data for each flight record, confirm the flight scene of any flight record and perform anomaly assessment on any flight record; The route deviation analysis module is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; and extract abnormal features from the flight data of any flight record based on the flight anomaly assessment result of any flight record; The expected deviation adjustment module is used to analyze the impact of each abnormal feature on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; according to the inclusion of abnormal features in any flight record, set corresponding deviation warning thresholds for different flight scenarios; The real-time anomaly analysis module is used to generate a real-time flight record every time the drone flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; capture the real-time flight route of the drone, analyze the real-time deviation of the flight route, and identify anomalies in the flight route of the real-time flight record.

7. The low-altitude UAV intelligent monitoring system according to claim 6 is characterized by: The historical route collection module includes a flight route collection unit and a flight quality assessment unit; The flight route collection unit is used to capture each flight route of the drone by installing monitoring equipment on the drone and generate corresponding flight records; the flight quality assessment unit is used to collect flight data of each flight record, confirm the flight scene of any flight record and perform abnormality assessment on any flight record.

8. The low-altitude UAV intelligent monitoring system according to claim 6 is characterized by: The route deviation analysis module includes a route deviation identification unit and an abnormal feature extraction unit; The route deviation identification unit is used to obtain the expected flight route of any flight record, analyze the deviation of the flight record by comparing the difference with the actual flight route; the abnormal feature extraction unit is used to extract abnormal features from the flight data of any flight record based on the flight abnormality assessment result of any flight record.

9. The low-altitude UAV intelligent monitoring system according to claim 6 is characterized by: The expected deviation adjustment module includes a feature deviation association unit and a deviation threshold division unit; The feature deviation association unit is used to analyze the impact of each abnormal feature on the deviation situation based on the deviation situation in any flight record, and obtain the impact value of each abnormal feature; the deviation threshold division unit is used to set corresponding deviation alarm thresholds for different flight scenarios according to the inclusion of abnormal features in any flight record.

10. The low-altitude UAV intelligent monitoring system according to claim 6, characterized in that: The real-time anomaly analysis module includes a real-time deviation confirmation unit and an abnormal flight identification unit; The real-time deviation confirmation unit is used to generate a real-time flight record whenever the UAV flies in real time, analyze the flight scene of the real-time flight record, and confirm the deviation alarm threshold of the real-time flight record; the abnormal flight identification unit is used to capture the real-time flight route of the UAV, analyze the real-time deviation of the flight route, and identify abnormalities in the flight route of the real-time flight record.

Images