Multi-source data fusion positioning monitoring method and system for low-altitude aircraft

Abstract

The invention discloses a multi-source data fusion positioning monitoring method and system for a low-altitude aircraft. The method comprises the following steps: collecting multi-source data, carrying out abnormal data processing, and then carrying out space-time alignment; dividing the Beidou absolute position data into different scene regions, preliminarily fusing the aligned multi-source data in the scene regions, configuring basic weights of the multi-source data in different flight scenes, and calculating dynamic weights of the multi-source data in real time based on equivalent variances of the multi-source data to obtain real-time comprehensive weights of the multi-source data; and performing state space fusion based on the preliminary fusion positioning information and the real-time comprehensive weight of each piece of multi-source data. According to the method, the pain point that a single data source is easily interfered, drifted or invalid is solved, and meanwhile, the real-time weight of each multi-source data can be intelligently optimized according to a task environment, real-time data quality and a task instruction, so that a fusion positioning result is corrected, and reliable monitoring in dynamic tasks such as emergency search and rescue, logistics transportation and the like is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of speech synthesis technology, and more specifically, to a multi-source data fusion positioning and monitoring method and system for low-altitude aircraft. Background Technology

[0002] In recent years, the low-altitude economy has been developing rapidly as a national strategic emerging industry, with an explosive growth in the number of low-altitude aircraft. Therefore, the safety monitoring and precise tracking of low-altitude airspace have become an unavoidable and urgent need. However, due to the complex environment of low-altitude airspace, with numerous clutter interferences and ground object obstructions, obtaining accurate and reliable monitoring and positioning of maneuvering aerial targets has become challenging.

[0003] Currently, low-altitude positioning and surveillance of aircraft primarily rely on single or dual-source data. Automatic Dependent Surveillance-Broadcast (ADS-B), a representative of single-source data surveillance, achieves collaborative airspace surveillance by broadcasting aircraft position and speed information acquired from GPS, and is the mainstream technology for low-altitude traffic management. However, this technology is entirely dependent on satellite navigation output position data. When satellite signals fail due to geographical obstruction, the broadcast position data becomes distorted. Furthermore, it is limited by base station locations and signal blockage, leading to blind spots in airspace surveillance and hindering effective monitoring. In contrast, a combined satellite navigation and inertial guidance positioning technology uses BeiDou or GPS satellite navigation data as its core, combined with an inertial measurement system (INS) for short-term motion prediction to compensate for short-term satellite signal interruptions. However, this approach relies heavily on inertial measurement units, leading to error accumulation and failing to meet the high-precision requirements of low-altitude operations. Additionally, positioning schemes using pure vision or a combination of vision and lidar achieve localized precise positioning by identifying ground reference points. The scheme lacks global coordinate references, making it impossible to achieve large-scale airspace collaborative management. It is also greatly affected by environmental characteristics and is prone to failure when there are no reference points, poor lighting, or bad weather.

[0004] In view of this, there is an urgent need for a technology that can break through the limitations of existing positioning and surveillance technologies on monitoring scenarios and achieve high-precision and high-continuity positioning and surveillance of low-altitude economic aircraft in complex environments. Summary of the Invention

[0005] In view of this, the main objective of the present invention is to propose a multi-source data fusion positioning and monitoring method and system for low-altitude aircraft, in order to at least partially solve at least one of the above-mentioned technical problems.

[0006] To address the aforementioned technical problems, the first aspect of this invention proposes a multi-source data fusion positioning and monitoring method for low-altitude aircraft, comprising: Collect multi-source data, including: BeiDou absolute position data of low-altitude aircraft, pose data relative to local maps, and broadcast data from surrounding aircraft or ground stations; After performing anomaly processing on the multi-source data, spatiotemporal alignment is then performed. Based on the absolute position data of BeiDou, different scene areas are divided. The aligned multi-source data in each scene area are initially fused to obtain preliminary fused positioning information. Configure the basic weights of each multi-source data under different flight scenarios, and calculate the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data. Then, merge the basic weights and dynamic weights of each multi-source data to obtain the real-time comprehensive weights of each multi-source data. Based on the preliminary fusion positioning information and the real-time comprehensive weights of various multi-source data, state space fusion is performed to obtain the fusion positioning result of the low-altitude aircraft.

[0007] According to a preferred embodiment of the present invention, the abnormal data processing of the multi-source data includes: For BeiDou absolute position data, the severely obstructed BeiDou absolute position data is filtered by the geometric precision factor threshold. At the same time, pseudorange change rate is detected based on sliding window statistics, and BeiDou absolute position data corresponding to pseudorange jumps exceeding the preset threshold are removed. For the pose data of the relative local map, the pose data of the relative local map with pose jumps are determined and removed based on the relationship between the pose change of adjacent frames and the second threshold, and the pose data of the relative local map with confidence less than the threshold are filtered. For broadcast data from surrounding aircraft or ground stations, verify the integrity of the broadcast data, and remove broadcast data with sudden changes in position, broadcast data with speeds exceeding the maximum maneuvering speed threshold of the aircraft, and abnormal broadcasts with timestamp conflicts. According to a preferred embodiment of the present invention, the spatiotemporal alignment of the multi-source data after anomaly processing includes: Using the time of BeiDou absolute position data reception as the reference time axis, the pose data relative to the local map and the broadcast data of surrounding aircraft or ground stations are aligned to the time of BeiDou absolute position data reception through linear interpolation. Using the center of the global map as the origin, the BeiDou absolute position data and the broadcast data of surrounding aircraft or ground stations are transformed to the northeast celestial coordinates under the same global center origin through Gauss-Kruger projection; the aircraft pose data in the pose data relative to the local map are transformed to the northeast celestial coordinates under the same global center origin through origin transformation; and the ground feature point data in the pose data relative to the local map are used to calculate the northeast celestial coordinates of the ground feature points through inverse perspective transformation of the camera intrinsic parameter matrix and the aircraft attitude angle.

[0008] According to a preferred embodiment of the present invention, the step of dividing the BeiDou absolute position data into different scene areas, and performing preliminary fusion of the aligned multi-source data in each scene area to obtain preliminary fused positioning information includes: Based on the parameter thresholds that characterize data quality, BeiDou absolute position data are divided into stable scene area, first occlusion scene area and second occlusion scene area; For the first obstruction scenario area, the drift error of the BeiDou absolute position data is corrected by using the reference position weighted average of the ground station broadcast data to generate preliminary fused positioning information; For stable scene areas, the pose data relative to the local map is aligned with the BeiDou absolute position data through initial extrinsic parameter calibration to construct a local map and generate preliminary fused positioning information; For the second occlusion scene area, preliminary fusion positioning information is generated by combining broadcast data from surrounding aircraft or ground stations and pose data relative to the local map.

[0009] According to a preferred embodiment of the present invention, the basic weights for configuring various multi-source data under different flight scenarios include: Based on the geometric precision factor of BeiDou absolute position data, the confidence level of pose data relative to local maps, and the loss rate of broadcast data from surrounding aircraft or ground stations, multi-source data are divided into open scene, slightly occluded scene, and heavily occluded scene. In open environments: the basic weights of BeiDou absolute position data, pose data relative to the local map, and broadcast data from surrounding aircraft or ground stations decrease in that order; in slightly obstructed environments: the basic weights of pose data relative to the local map, BeiDou absolute position data, and broadcast data from surrounding aircraft or ground stations decrease in that order; in heavily obstructed environments: the basic weights of pose data relative to the local map, broadcast data from surrounding aircraft or ground stations, and BeiDou absolute position data decrease in that order. Real-time parsing of voice commands, determination of task categories based on voice commands, and adjustment of basic weights based on a mapping table between task categories and weight adjustment amounts; If a certain multi-source data is not received within a preset time or if feature matching of a certain multi-source data fails for multiple consecutive frames, the basic weight of the certain multi-source data will be temporarily frozen until the certain multi-source data is received again or feature matching of the certain multi-source data is successful, at which point the freeze will be lifted.

[0010] According to a preferred embodiment of the present invention, the real-time calculation of the dynamic weights of each multi-source data based on the equivalent variance of each multi-source data includes: Calculate the equivalent variance of each multi-source data; The dynamic weights of each multi-source data are obtained by taking the reciprocal of the equivalent variance of each multi-source data and normalizing it.

[0011] According to a preferred embodiment of the present invention, the step of performing state-space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft includes: Update the state vector at time K based on the Kalman gain and observation residuals. Covariance Matrix The fused position, velocity, and attitude estimates are obtained; where: State vector ; To initially fuse the vectors corresponding to the positioning information, h( ) is the observation function; For Kalman gain, For the observation function h ( The Jacobian matrix is ​​represented by h( )exist The result is obtained by first-order Taylor expansion at the location; for The transpose of the matrix; To measure the noise covariance matrix: in , , The weighted averages are the absolute position data from BeiDou, the pose data relative to the local map, and the broadcast data from nearby aircraft or ground stations. , , These are the base noise levels for BeiDou absolute position data, pose data relative to a local map, and broadcast data from nearby aircraft or ground stations, calibrated according to industry standards and equipment accuracy parameters; diag ( ) represents a diagonal matrix; covariance matrix It is an identity matrix with the same dimensions as the state vector.

[0012] To address the aforementioned technical problems, a second aspect of the present invention provides a multi-source data fusion positioning and monitoring system for low-altitude aircraft, comprising: The acquisition module is used to acquire multi-source data, which includes: BeiDou absolute position data of low-altitude aircraft, pose data relative to a local map, and broadcast data from surrounding aircraft or ground stations. The alignment module is used to perform spatiotemporal alignment on the multi-source data after anomaly processing. The preliminary fusion module is used to divide the BeiDou absolute position data into different scene areas. The aligned multi-source data in each scene area is initially fused to obtain preliminary fused positioning information. The weight adjustment module is used to configure the basic weights of various multi-source data under different flight scenarios, and to calculate the dynamic weights of various multi-source data in real time based on the equivalent variance of each multi-source data. The basic weights and dynamic weights of each multi-source data are then integrated to obtain the real-time comprehensive weights of each multi-source data. The state fusion module is used to perform state space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft.

[0013] To solve the above-mentioned technical problems, a third aspect of the present invention provides an electronic device, comprising: A processor; and a memory storing computer-executable instructions, which, when executed, cause the processor to perform the method according to any one of the preceding claims.

[0014] To solve the above-mentioned technical problems, the fourth aspect of the present invention provides a computer program product, including a computer program, characterized in that the computer program, when executed by a processor, implements the method described in any of the above-mentioned embodiments.

[0015] In summary, this invention performs preliminary fusion of BeiDou absolute position data of low-altitude aircraft, pose data relative to local maps, and broadcast data from surrounding aircraft or ground stations in different scenario areas to obtain preliminary fused positioning information; determines the basic weights of each multi-source data based on different flight scenarios, and calculates the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data; and fuses the basic weights and dynamic weights of each multi-source data to obtain the real-time comprehensive weight of each multi-source data. Finally, state-space fusion is performed based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft. This solves the pain point of single data sources being susceptible to interference, drift, or failure, and improves the positioning accuracy and continuity of aircraft in complex low-altitude environments such as urban obstruction and multi-path conditions in mountainous areas. Simultaneously, it can intelligently optimize the real-time weights of various multi-source data according to the mission environment, real-time data quality, and mission instructions, thereby correcting the fused positioning result. This ensures reliable monitoring in dynamic tasks such as emergency search and rescue and logistics transportation, and provides highly reliable aircraft position and trajectory prediction for airspace traffic management, effectively supporting efficient resource allocation and collision warning in low-altitude airspace, and achieving robust low-altitude flight safety assurance in diverse environments. Compared with existing technologies, this invention has at least the following beneficial effects: 1. Multi-source data fusion architecture: Based on the local high-precision correction characteristics of Beidou absolute position data of low-altitude aircraft, pose data relative to local maps, and broadcast data from surrounding aircraft or ground stations, this invention combines preliminary fusion and state fusion to solve the reliability problem of single data sources in complex scenarios such as obstruction and signal failure.

[0016] 2. Dynamic Adaptive Weight Adjustment: This invention dynamically allocates the comprehensive weights of various multi-source data based on scene classification (open / slight / heavy occlusion) and real-time data reliability. It also introduces language command assistance and a sensor failure freeze mechanism to further improve the localization adaptability under complex tasks.

[0017] 3. Adaptability to complex low-altitude environments: In response to scenarios such as obstruction by tall buildings in cities and multipath reflection in mountainous areas, the system fills the gaps in BeiDou absolute position data by using pose data from local maps, and coordinates monitoring with broadcast data from surrounding aircraft or ground stations to achieve high-precision positioning and airspace surveillance in diverse environments. Attached Figure Description

[0018] To make the technical problems solved by this invention, the technical means adopted, and the technical effects achieved clearer, specific embodiments of this invention will be described in detail below with reference to the accompanying drawings. However, it should be noted that the drawings described below are merely drawings of exemplary embodiments of this invention. Those skilled in the art can obtain drawings of other embodiments based on these drawings without any creative effort.

[0019] Figure 1 This is a flowchart illustrating a multi-source data fusion positioning and monitoring method for low-altitude aircraft provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of another multi-source data fusion positioning and monitoring method for low-altitude aircraft provided in an embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the coordinate alignment of multi-source data according to an embodiment of the present invention; Figure 4 This is a schematic diagram of the structural framework of the multi-source data fusion positioning and monitoring system for low-altitude aircraft provided in an embodiment of the present invention; Figure 5 This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention. Detailed Implementation

[0020] Subject to the inventive concept, the structures, performance, effects or other features described in a particular embodiment may be combined in any suitable manner with one or more other embodiments.

[0021] In the description of specific embodiments, detailed descriptions of structures, performance, effects, or other features are provided to enable those skilled in the art to fully understand the embodiments. However, it is not excluded that those skilled in the art can implement the present invention under certain circumstances with technical solutions that do not contain the above-described structures, performance, effects, or other features. The figures in the accompanying drawings are merely illustrative examples and do not imply that the solutions of the present invention must include all the contents, operations, and steps shown in the figures, nor do they imply that they must be performed in the order shown in the figures.

[0022] refer to Figure 1 , Figure 1 This is a flowchart illustrating a multi-source data fusion positioning and monitoring method for low-altitude aircraft provided in an embodiment of the present invention, as shown below. Figure 1 The method includes: S1. Collect data from multiple sources. In this embodiment, the multi-source data may include: BeiDou absolute position data of the low-altitude aircraft, pose data relative to a local map, and broadcast data from surrounding aircraft or ground stations. Specifically: the BeiDou absolute position data can be collected via a BeiDou positioning module; the broadcast data from surrounding aircraft or ground stations can be collected via an Automatic Dependent Surveillance-Broadcast (ADS-B) receiver / transmitter module; and the pose data relative to a local map can be collected via a Vision-and-Language Navigation (VLN) module. Figure 2 The BeiDou positioning module collects BeiDou data (i.e., BeiDou absolute position data), the ADS-B receiving / transmitting module collects ADS-B data (i.e., broadcast data from surrounding aircraft or ground stations), and the VLN visual positioning module collects VLN data (i.e., pose data relative to a local map).

[0023] For example, the BeiDou positioning module is equipped with a multi-frequency satellite navigation receiver to collect the aircraft's raw observation data and navigation messages, and outputs the low-altitude aircraft's BeiDou absolute position data, including longitude, latitude, altitude, velocity, and Coordinated Universal Time (UTC) timestamp. In this embodiment, the BeiDou positioning module outputs a set of position / velocity / timestamps every TBDS seconds. The BeiDou position data is based on the WGS-84 coordinate system, and the altitude H is the geodetic height.

[0024] ADS-B Receiver / Transmitter Module: Receives broadcast data from surrounding aircraft or ground stations via an onboard ADS-B receiver, including target longitude, latitude, speed, heading, timestamp, and International Civil Aviation Organization (ICAO) address. In this embodiment, the ADS-B receiver / transmitter module receives broadcast data from surrounding aircraft or ground stations every T...ADS_B Receives its own location data every second, with the timestamp being the broadcast time in UTC. ADS_B . In this embodiment, the Vision-and-Language Navigation (VLN) module integrates a visible light / infrared camera and an Inertial Measurement Unit (IMU). Through Simultaneous Localization and Mapping (SLAM) or a matching algorithm based on feature points / semantic maps, it outputs in real-time pose data (position + attitude) and confidence level of the low-altitude aircraft relative to a local map, including reprojection error and map matching score. Figure 2 The pose data of a low-altitude aircraft relative to a local map can include: 1) Visual data: RGB images captured by dual downward / forward cameras, used to extract visual features of ground reference objects, such as manhole covers, streetlights, and building edges. 2) Inertial navigation unit (IMU) data: Raw angular velocity and acceleration data provided by the IMU, fused with visual data via visual inertial odometry, outputting the aircraft's position, attitude, and confidence level relative to the local map, including reprojection error and map matching score. 3) Semantic data: Voice commands issued by the user during flight. In this embodiment, the visual positioning module camera captures one frame of image every 1 / fcam second, and the IMU captures images at f frames per second. IMU The Hz outputs acceleration and angular velocity; the vision processing unit outputs the aircraft's pose relative to the local map in real time through visual SLAM, including position [x,y,z] and attitude [θ, φ, ψ] (pitch, roll, yaw) data based on the ENU coordinate system, as well as data confidence Conf_VLN; for semantic data, VLN semantic command keywords are extracted based on the BERT lightweight model, including "precise hovering" and "avoiding obstacles".

[0025] S2. Perform spatiotemporal alignment on the multi-source data after anomaly data processing. After collecting multi-source data, such as Figure 2 First, anomaly processing is performed on the multi-source data to remove abnormal or low-precision multi-source data. For example, the anomaly processing of the multi-source data may include: For BeiDou absolute position data, severely obstructed BeiDou absolute position data is filtered out using a geometric precision factor (PDOP) threshold. Specifically, a PDOP threshold representing extremely poor accuracy is set; when the value exceeds the PDOP threshold, the BeiDou absolute position data is discarded. Simultaneously, pseudorange change rate is detected statistically using a sliding window, and BeiDou absolute position data corresponding to pseudorange jumps exceeding a preset threshold are removed. For pose data of relative local maps, pose data of relative local maps with pose jumps are determined and removed based on the relationship between pose change in adjacent frames and a second threshold. For example, pose data of relative local maps with pose change in adjacent frames less than the second threshold are removed as jump data. Furthermore, extremely low-precision observations are filtered through the confidence level Conf_VLN threshold. That is, a Conf_VLN threshold with extremely low representation accuracy is set. When the value is lower than the Conf_VLN threshold, the pose data of the relative local map is discarded.

[0026] For broadcast data from surrounding aircraft or ground stations, verify the integrity of the broadcast data, including ICAO address validity and CRC checksum verification. Remove broadcast data with abrupt position changes, broadcast data exceeding the aircraft's maximum maneuvering speed threshold, and abnormal broadcasts with timestamp conflicts. Abrupt position changes refer to position changes exceeding a preset threshold. In this embodiment, multi-source data undergoes anomaly processing before spatiotemporal alignment to achieve temporal and coordinate uniformity across various multi-source data, facilitating subsequent fusion processing of the multi-source data. For example, the spatiotemporal alignment includes: S21. Using the time of BeiDou absolute position data reception as the reference time axis, the pose data relative to the local map and the broadcast data of surrounding aircraft or ground stations are aligned to the time of BeiDou absolute position data reception through linear interpolation. Specifically, let the sampling period of data source A be TA, and the sampling period of data source B be TB, where TA:TB = n. The last time data source A updates the target state is (k-1)TA. Then the time [(k-1)TA+n]... [TB] represents the next target state update time of data source A, i.e., kTA, and vice versa. That is, when the target state of data source A is updated twice, assuming the next target update time is tA=kTA, the target state of data source B is updated n times, and its measurement time is tB,1=(k−1)TA+TB, tB,2=(k−1)TA+2TB, ..., tB,n=(k−1)TA+nTB=kTA.

[0027] Subsequently, linear interpolation is used to synthesize the n measurements from data source B into an equivalent measurement value zsync. The merging method is as follows: Step 1) Time sorting and indexing: Sort the n measurement times tB,i and corresponding measurement values ​​zB,i of data source B in ascending order of time (tB,1≤tB,2≤…≤tB,n).

[0028] Step 2) Linear interpolation: First, if there exists a tB,j = tA, then directly take that measurement value as the equivalent measurement value: zsync = zB,j. Second, if there is no precisely matched tB,j for tA, find two adjacent measurement points tB,jpre and tB,jpost before and after time tA, satisfying: tB,jpre ≤ tA ≤ tB,jpost. Calculate the time interpolation weight α: α=(tA−tB,jpre) / tB,jpost−tB,jpre) The equivalent measurement zsync is: zsync = zB,jpre×α + zB,jpost×(1−α) At this point, the n equivalent measurements from data source B are aligned with data source A in chronological order. In this example, data source A is BeiDou absolute position data, and data source B is pose data relative to a local map, as well as broadcast data from surrounding aircraft or ground stations.

[0029] S22. Using the center of the global map as the origin, transform the BeiDou absolute position data and the broadcast data of surrounding aircraft or ground stations to the northeast celestial coordinates under the same global center origin through Gauss-Kruger projection; transform the aircraft pose data in the pose data relative to the local map to the northeast celestial coordinates under the same global center origin through origin transformation; calculate the northeast celestial coordinates of the ground feature points in the pose data relative to the local map through inverse perspective transformation of the camera intrinsic parameter matrix and the aircraft attitude angle.

[0030] In this embodiment, the coordinates of the multi-source data are uniformly transformed to the Northeast Sky ENU local coordinate system, with reference to... Figure 3 The coordinates are determined by the global map center origin, with the E-axis pointing east, the N-axis pointing north, and the U-axis pointing upward. The coordinates of BeiDou data (geocentric coordinate system) and ADS-B data (geocentric coordinate system) are transformed using Gauss-Kruger projection. The aircraft pose data acquired from VLN data is transformed from the local center ENU coordinates to ENU coordinates under the same global center origin through origin transformation. For ground feature point data in the visual image, the ENU coordinates of the ground feature points are calculated through inverse perspective transformation using the camera intrinsic parameter matrix and the aircraft attitude angles, including pitch angle θ, roll angle φ, and yaw angle ψ.

[0031] S3. Divide the BeiDou absolute positioning data into different scene areas, and perform preliminary fusion of the aligned multi-source data in each scene area to obtain preliminary fused positioning information; For example, this step may include: S31. Based on the parameter thresholds that characterize data quality, the BeiDou absolute position data is divided into stable scene area, first occlusion scene area and second occlusion scene area. For absolute positional data, PDOP is used as a parameter to characterize data quality; the smaller the PDOP, the better the data quality. For example, PDOP thresholds PDOPgood and PDOPmid can be set to determine whether PDOP is below PDOPgood, corresponding to good data quality; above PDOPmid, corresponding to poor data quality; and between PDOPgood and PDOPmid, corresponding to stable data quality.

[0032] Based on this, the PDOP of BeiDou absolute positioning data is PDOP≤PDOP good At times, such as Figure 2 When the BeiDou signal is stable, i.e., when the quality of the BeiDou absolute position data is good, this BeiDou absolute position data is assigned to the stable scene area. The PDOP of the BeiDou absolute position data is 2. When PDOPmid ≥ PDOP > PDOPmid, such as Figure 2 When the BeiDou signal is weak, low-altitude aircraft may be in environments with obstructed signals, such as urban high-rise buildings or mountainous areas. In such cases, the BeiDou absolute position data should be classified into the first obstruction scenario area. The PDOP of BeiDou absolute position data is greater than 2. PDOPmid, such as Figure 2 Since BeiDou is almost unusable, it is classified as the second obstruction scenario area.

[0033] S32. For the first obstructed scene area, based on the absolute position of BeiDou, the drift error of the absolute position data is corrected by using the reference position weighted average of the ground station broadcast data to generate preliminary fused positioning information. For the first obstruction scenario area, using the BeiDou absolute position as the reference, and leveraging the ground station's broadcast data's own reference position, a weighted average is applied to correct the drift error of the BeiDou absolute position data, thereby revising the initial position estimate and achieving loosely coupled fusion of real-time observation positions. The final result is: Preliminary location = w_pre_BDS Beidou absolute position + w_pre_ADS_B ADS_B location; Among them, the weight of the BeiDou absolute position data is w_pre_BDS=1-(PDOP-PDOPmid) / PDOPmid, and w_pre_ADS_B=1-w_pre_BDS.

[0034] S33. For stable scene areas, the initial extrinsic parameters of the pose data relative to the local map are calibrated based on the BeiDou absolute position data, a local map is constructed, and preliminary fused positioning information is generated. Since this embodiment predetermines the absolute position of the origin relative to the pose data of the local map, performs initial extrinsic parameter calibration, and then constructs the local map to obtain the coordinates of ground feature points, the initial absolute position of the origin may contain errors. Therefore, in this embodiment, when the BeiDou absolute position data is in a stable scene area, it further calibrates based on the initial extrinsic parameters of the BeiDou absolute position data relative to the pose data of the local map, so as to use the BeiDou stabilization signal to correct the origin position of the local map.

[0035] When there is no BeiDou signal, the initial extrinsic parameters remain valid and can be used as a reference for the next spatiotemporal alignment.

[0036] S34. For the second occlusion scene area, generate preliminary fused positioning information by integrating broadcast data from surrounding aircraft or ground stations and pose data relative to the local map.

[0037] For the second occlusion scene area, PDOP>2 PDOPmid indicates that the BeiDou positioning module has failed.

[0038] If valid broadcast data from nearby aircraft or ground stations exists, the ground station broadcast data is used as alternative data to replace the BeiDou absolute position data to generate preliminary fused positioning information.

[0039] If there is no valid broadcast data from nearby aircraft or ground stations at the same time, the location of the last valid BeiDou absolute position data will be used as the starting point for determining the position of the BeiDou absolute position data. equal Add the accumulated VLN visual inertial odometry position increment value from the last valid BeiDou absolute position data time to the current time period relative to the local map pose data. This allows the VLN's visual inertial odometry to output short-term, high-precision pose, filling the signal gap caused by the failure of the BeiDou positioning module.

[0040] S4. Configure the basic weights of each multi-source data under different flight scenarios, and calculate the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data. Then, merge the basic weights and dynamic weights of each multi-source data to obtain the real-time comprehensive weights of each multi-source data. To address the dynamic changes in the complex low-altitude environment, this embodiment calculates the dynamic weights of each data source in real time based on the equivalent variance of each data source, and combines them with the basic weights under different flight scenarios to obtain a comprehensive weight. This comprehensive weight serves as the basis for calculating the measurement noise covariance matrix during each multi-source data state space fusion, thereby correcting the final fusion positioning result.

[0041] For example, the basic weights for configuring various multi-source data under different flight scenarios include: like Figure 2Scene determination: Based on the geometric precision factor of BeiDou absolute position data, the confidence level of pose data relative to local maps, and the loss rate of broadcast data from surrounding aircraft or ground stations, multi-source data are divided into open scenes, slightly occluded scenes, and heavily occluded scenes. like Figure 2 Classification configuration: In open scenarios: the basic weights of BeiDou absolute position data, pose data relative to the local map, and broadcast data from surrounding aircraft or ground stations decrease in that order; in slightly obscured environments: the basic weights of pose data relative to the local map, BeiDou absolute position data, and broadcast data from surrounding aircraft or ground stations decrease in that order; in heavily obscured environments: the basic weights of pose data relative to the local map, broadcast data from surrounding aircraft or ground stations, and BeiDou absolute position data decrease in that order.

[0042] For example, in one basic weight configuration method: in an open scene, the BeiDou absolute position data PDOP≤PDOPgood, the loss rate of broadcast data from surrounding aircraft or ground stations Lost≤Lostgood, and the confidence of pose data relative to the local map ≥Conf_VLNgood. The basic weights are: basic weight of BeiDou absolute position data > basic weight of pose data relative to the local map > basic weight of broadcast data from surrounding aircraft or ground stations, and the sum of the weights of these three data is 1. For example, the basic weight of BeiDou absolute position data w_baseBDS=0.6, the basic weight of pose data relative to the local map w_baseVLN=0.3, and the basic weight of broadcast data from surrounding aircraft or ground stations w_baseADS_B=0.1.

[0043] In lightly obstructed scenarios, the BeiDou absolute position data PDOPgood < PDOP ≤ PDOPmid, Lostgood < loss rate of broadcast data from surrounding aircraft or ground stations ≤ Lostmid, and Conf_VLNmid ≤ confidence of pose data relative to the local map < Conf_VLNgood. The base weights are: base weight of pose data relative to the local map > base weight of BeiDou absolute position data > base weight of broadcast data from surrounding aircraft or ground stations, and the sum of the weights of these three data is 1. For example, the base weight of BeiDou absolute position data w_baseBDS = 0.4, the base weight of VLN w_baseVLN = 0.5, and the base weight of broadcast data from surrounding aircraft or ground stations w_baseADS_B = 0.1.

[0044] In heavily obstructed scenarios, the BeiDou absolute position data PDOP > PDOPmid, the loss rate of broadcast data from surrounding aircraft or ground stations > Lostmid or no new data, and the confidence level of pose data relative to the local map < Conf_VLNmid: Basic weights: the weight of pose data relative to the local map > the weight of broadcast data from surrounding aircraft or ground stations > the weight of BeiDou absolute position data, and the sum of the weights of these three data is 1. When a data signal fails, the corresponding weight is set to 0.

[0045] Furthermore, the aforementioned basic weights can be modified to allow for real-time parsing of voice commands, determination of task categories based on these commands, and correction of the basic weights according to the mapping table between task categories and weight correction amounts. Specifically, task categories can be obtained by extracting semantic features from voice commands, and a mapping relationship between semantic features and task categories can be pre-set to determine the corresponding task category based on the semantic command. Table 1 shows the mapping relationship between semantic features, task categories, and weight correction amounts from different multi-source data. In addition, when semantic features conflict, dynamic obstacle avoidance localization is prioritized for correction, followed by static high-precision localization, and finally environmental adaptation correction.

[0046] Table 1 shows the mapping relationship between semantic features, task categories, and weight correction values ​​for different multi-source data. In addition, in this step, if a certain multi-source data is not received within a preset time or if the feature matching of a certain multi-source data fails for multiple consecutive frames, the basic weight of the certain multi-source data is temporarily frozen until the certain multi-source data is received again or the feature matching of the certain multi-source data is successful, and then the freeze is lifted.

[0047] During the dynamic weight calculation phase, the equivalent variance of each data source is used. The real-time fusion weights are determined using the inverse variance weighting method, thereby ensuring that high-reliability data receives higher fusion weights while suppressing the contribution of low-reliability data, achieving adaptive optimal estimation in a dynamic environment. For example, the real-time calculation of the dynamic weights of each multi-source data based on the equivalent variance of each multi-source data includes: S41. Calculate the equivalent variance of each multi-source data; Equivalent variance of BeiDou absolute positioning data First, obtain the basic variance of the BeiDou absolute positioning data. In engineering, the square of the root mean square error (RMSE) is generally used as the basic variance. For example, if the equipment accuracy is "horizontal RMSE = 0.15 meters, vertical RMSE = 0.2 meters", then... =0.152 + 0.152 + 0.22 ≈ 0.085m 2Secondly, the PDOP impact factor fPDOP(PDOP) is calculated. Assuming a non-linear positive correlation between PDOP and variance, typically an increase of 1 in PDOP leads to a 10%–30% increase in observation error. This invention uses a linear approximation function to calculate the PDOP impact factor fPDOP(PDOP) = 1 + kPDOP × (PDOP - PDOPref), where PDOPref is the reference PDOP threshold corresponding to different flight scenarios, used to adjust the weights of BeiDou absolute position data under different flight scenarios; kPDOP is the adjustment coefficient. Finally, the equivalent variance of the BeiDou absolute position data is obtained. . = ×f PDOP(PDOP).

[0048] Equivalent variance of broadcast data from surrounding aircraft or ground stations : First, the basic variance of broadcast data from surrounding aircraft or ground stations is obtained based on the equipment's accuracy. Secondly, the loss rate impact factor f_lost(Lost) is calculated using a linear function: f_lost(Lost) = 1 + kLost × (Lost - Lostref). Here, Lostref is the reference Lost threshold for different flight scenarios, used to adjust the BeiDou weights under different flight scenarios; kLost is the adjustment coefficient. Finally, the equivalent variance of the broadcast data from surrounding aircraft or ground stations is obtained. .

[0049] = ×f lost(Lost).

[0050] Equivalent variance of pose data relative to local map First, the fundamental variance of the pose data relative to the local map is calculated based on the device's accuracy. Secondly, the influence of confidence level is corrected by using a quadratic penalty function to amplify the negative impact of low confidence level, resulting in the confidence level influence factor fConf_VLN(Conf_VLN) = (1-Conf_VLN). 2 Finally, the equivalent variance of the pose data relative to the local map is obtained. .

[0051] = f Conf_VLN (Conf_VLN) S42. Take the reciprocal of the equivalent variance of each multi-source data and normalize it to obtain the dynamic weight of each multi-source data.

[0052] Specifically, the inverse of the equivalent variance for each multi-source dataset is taken and normalized: For example: BeiDou's dynamic weights .

[0053] After obtaining the basic and dynamic weights of each multi-source data, the proportions of the basic and dynamic weights can be set as follows: The combined weights of the various multi-source data are obtained.

[0054] Taking BeiDou absolute positioning data as an example, the comprehensive weight is: _BDS= _baseBDS+ S5. Based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data, state space fusion is performed to obtain the fused positioning result of the low-altitude aircraft.

[0055] For example, this step updates the state vector at time K based on the Kalman gain and the observation residual. Covariance Matrix The fused position, velocity, and attitude estimates are obtained; where: State vector ; To initially fuse the vectors corresponding to the positioning information, h( ) is the observation function; For Kalman gain, For the observation function h ( The Jacobian matrix is ​​represented by h( )exist The result is obtained by first-order Taylor expansion at the location; for The transpose of the matrix; To measure the noise covariance matrix: in , , The weighted averages are the absolute position data from BeiDou, the pose data relative to the local map, and the broadcast data from nearby aircraft or ground stations. , , These are the base noise levels for BeiDou absolute position data, pose data relative to a local map, and broadcast data from nearby aircraft or ground stations, calibrated according to industry standards and equipment accuracy parameters; diag ( ) represents a diagonal matrix; covariance matrix It is an identity matrix with the same dimensions as the state vector.

[0056] In one specific implementation, step S5 may include: S51, such as Figure 2 Constructing a state model: state vector = [x, y, z, v_x, v_y, v_z, θ, φ,ψ, , ]ᵀ, containing the position of the spacecraft in the ENU coordinate system p=[x,y,z] T Speed ​​v[vx,vy,vz] T Attitude Euler angles [θ,φ,ψ] T IMU accelerometer zero bias and gyroscope zero bias .

[0057] S52, Prediction Phase: like Figure 2 Prior state estimation: State prediction is performed using nonlinear state equations, primarily based on acceleration a_k and angular velocity ω_k data collected by the IMU. The inertial navigation model is used to predict the position, velocity, and attitude at the next moment. The state transition function is... in This is process noise. f( is the Kalman fusion time step, consistent with the time step after time unification above). The function is a nonlinear dynamic function that predicts the state value at time k from the state value at time k-1. The core idea here is to utilize the motion integral equation of the IMU. Considering the high-frequency update characteristics of the IMU, the high-frequency step size of the IMU is used. The IMU continuously iterates to obtain the state prediction value for the next time step. The change in state over time consists of multiple The state of the IMU time is obtained through recursion.

[0058] The main recursive sub-function equations are as follows: Angular velocity correction: ; This is Gaussian white noise for the gyroscope, with a typical value of 0.001 rad / s. The raw angular velocity is measured by the IMU gyroscope. The posterior bias of the gyroscope at time k-1 is the state vector from the previous time step.

[0059] Attitude update: using attitude Euler angles [θ,φ,ψ] T Quaternions obtained through quaternion transformation Calculate attitude update. + .

[0060] Assuming the rotation order of Euler angles is ZYX, that is, φ (yaw angle) rotates around the Z-axis; θ (pitch angle) rotates around the Y-axis; and ψ (roll angle) rotates around the X-axis, the conversion formula is as follows: =[q0,q1,q2,q3] T q0 = cos(φ / 2)·cos(θ / 2)·cos(ψ / 2) + sin(φ / 2)·sin(θ / 2)·sin(ψ / 2) q1 = cos(φ / 2)·cos(θ / 2)·sin(ψ / 2) - sin(φ / 2)·sin(θ / 2)·cos(ψ / 2) q2 = cos(φ / 2)·sin(θ / 2)·cos(ψ / 2) + sin(φ / 2)·cos(θ / 2)·sin(ψ / 2) q3 = sin(φ / 2)·cos(θ / 2)·cos(ψ / 2) - cos(φ / 2)·sin(θ / 2)·sin(ψ / 2) Linear acceleration correction: This is the attitude matrix of the machine body to the global coordinate system. These are the raw measurements from the IMU accelerometer. It is gravitational acceleration. This is accelerometer noise. It is an accelerometer zero bias that estimates the system state at time (k-1) based on all observation data up to time (k-1).

[0061] = [[q0²+q1²−q2²−q3², 2(q1q2−q0q3), 2(q1q3+q0q2) ], [2(q1q2+q0q3), q0²−q1²+q2²−q3², 2(q2q3−q0q1) ], [2(q1q3−q0q2), 2(q2q3+q0q1), q0²−q1²−q2²+q3²]] Location update: + Zero bias update: Considering the slow change of the zero bias, it can be assumed to remain approximately constant during the prediction phase. The initial value is calibrated using equipment parameters.

[0062] ; like Figure 2 Calculation of the prior covariance matrix The covariance matrix P describes the uncertainty relationship between states. The prior covariance matrix is ​​calculated as follows: in: For the dynamic function f( )exist The Jacobian matrix can be obtained by calculating the partial derivatives using numerical differentiation. for The transpose of . Let be the posterior covariance matrix at time k-1. Evidence for poor process noise reporting was obtained through sensor calibration.

[0063] S53, such as Figure 2 Observation model: The observation function is as follows: in: This represents the multi-source fusion observation vector. h( ) is the observation function. The noise used for measurement is Gaussian white noise, and its covariance matrix is... .

[0064] ; ; The observation function is obtained using the corresponding values ​​in the state vector.

[0065] S54, such as Figure 2 Kalman gain calculation For the observation function h ( The Jacobian matrix is ​​represented by h( )exist The result is obtained by first-order Taylor expansion at the given location. For example: for The transpose of . The noise covariance matrix is ​​calculated based on the basic weights and comprehensive weights of multiple source data to measure the noise covariance matrix.

[0066] in , , Assuming a comprehensive weight for all multi-source data, , , The basic weights for each multi-source data.

[0067] S55, such as Figure 2 Update phase Update the target state vector at each time step based on the Kalman gain and observation residuals. Covariance Matrix The fused position, velocity, and attitude estimates are obtained as follows: Posterior state estimation: To observe the residuals.

[0068] Calculation of the posterior covariance matrix: It is an identity matrix with the same dimensions as the state vector.

[0069] S56. The posterior state at time k and posterior covariance By repeatedly using steps S52-55 as the prior input at time k+1, the real-time dynamic estimation of the target state can be achieved.

[0070] Based on the above-mentioned multi-source data fusion positioning and monitoring method for low-altitude aircraft, this invention also provides a multi-source data fusion positioning and monitoring system for low-altitude aircraft, such as... Figure 4 As shown, the system includes: The acquisition module 41 is used to acquire multi-source data, including: BeiDou absolute position data of the low-altitude aircraft, pose data relative to the local map, and broadcast data from surrounding aircraft or ground stations. Alignment module 42 is used to perform spatiotemporal alignment on the multi-source data after abnormal data processing; The preliminary fusion module 43 is used to divide the BeiDou absolute position data into different scene areas. The aligned multi-source data in each scene area is initially fused to obtain preliminary fused positioning information. The weight adjustment module 44 is used to configure the basic weights of each multi-source data under different flight scenarios, and to calculate the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data. The basic weights and dynamic weights of each multi-source data are then integrated to obtain the real-time comprehensive weights of each multi-source data. The state fusion module 45 is used to perform state space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft.

[0071] In one example, the acquisition module 41 may include: a BeiDou positioning module for the BeiDou absolute position data of the low-altitude aircraft. Specifically, it may employ a dual-frequency multi-system GNSS receiver (supporting BDS B1 / B2), integrating a high-precision RTC clock with an output frequency of 10Hz, providing the aircraft's longitude (λ), latitude (φ), altitude (h, ellipsoidal height), and three-dimensional velocity (v). x The module includes the BeiDou absolute position reference (VZ, VY, VZ) and UTC timestamps (accuracy ±ΔtBDS). However, it is prone to signal attenuation or drift in complex environments such as tall buildings blocking the signal or multiple paths in mountainous areas.

[0072] The ADS-B receiver / transmitter module is used to collect broadcast data from surrounding aircraft or ground stations. Specifically, the receiver can support the 1090ES data link. The received data includes the target ICAO address, location (including latitude, longitude, and altitude), velocity vector, heading angle, and UTC timestamp (accuracy ±ΔtADS). ADS-B provides surrounding airspace coordination information, but the data update rate is low, the coverage is limited, it depends on the base station location, and it is easily lost in scenarios such as urban canyons.

[0073] VLN visual positioning module: used to acquire pose data relative to a local map, may include a binocular camera: global shutter, resolution Rcam, frame rate fcam frames per second, field of view θcam; IMU: MEMS six-axis, accelerometer range ±Amax, gyroscope range ±Gmax, output frequency fIMUHz per second; visual processing unit: equipped with a computing chip. The camera is mounted on the belly of the aircraft, with a downward pitch angle αcam to acquire ground texture, such as roads, building edges, etc. The IMU and camera are synchronized in time and space through rigid calibration. VLN uses visual SLAM based on feature point matching algorithm, i.e., SLAM technology, to output the pose data of the aircraft relative to the local map in real time, including position [x, y, z] in ENU coordinate system, attitude [θ, φ, ψ] (pitch, roll, yaw) and confidence score (Conf_VLN, range 0~1, high value when reprojection error < δreproj pixels and feature point matching number > Nfeat, linearly decreasing to low value as the error increases).

[0074] Furthermore, the acquisition module 41 may also include: a VLN semantic level adaptation module: receiving natural language commands from ground stations or airborne mission systems via the aircraft data link.

[0075] The alignment module 42 may include: The preprocessing unit filters out severely obstructed BeiDou absolute position data by using a geometric precision factor threshold, and detects pseudorange change rate based on sliding window statistics, removing BeiDou absolute position data corresponding to pseudorange jumps that exceed a preset threshold. For pose data of relative local maps, the unit determines and removes pose data of relative local maps with pose jumps based on the relationship between pose change in adjacent frames and a second threshold, and filters pose data of relative local maps with confidence levels less than a threshold. For broadcast data from surrounding aircraft or ground stations, verify the integrity of the broadcast data, and remove broadcast data with sudden changes in position, broadcast data with speeds exceeding the maximum maneuvering speed threshold of the aircraft, and abnormal broadcasts with timestamp conflicts. The sub-alignment unit is used to align the pose data relative to the local map and the broadcast data of surrounding aircraft or ground stations to the time of BeiDou absolute position data reception using linear interpolation, with the time of BeiDou absolute position data reception as the reference time axis. Using the center of the global map as the origin, the BeiDou absolute position data and the broadcast data of surrounding aircraft or ground stations are transformed to the northeast celestial coordinates under the same global center origin through Gauss-Kruger projection. The aircraft pose data in the pose data relative to the local map are transformed to the northeast celestial coordinates under the same global center origin through origin transformation. The ground feature point data in the pose data relative to the local map are used to calculate the northeast celestial coordinates of the ground feature points through inverse perspective transformation of the camera intrinsic parameter matrix and the aircraft attitude angle.

[0076] The initial fusion module 43 includes: The first division unit is used to divide the BeiDou absolute position data into the stable scene area, the first occlusion scene area and the second occlusion scene area according to the parameter threshold that characterizes the data quality. The first sub-fusion unit is used to correct the drift error of the BeiDou absolute position data by using the reference position weighted average of the ground station broadcast data for the first occluded scene area, and to generate preliminary fused positioning information. For stable scene areas, the pose data relative to the local map is aligned with the BeiDou absolute position data through initial extrinsic parameter calibration to construct a local map and generate preliminary fused positioning information; for second occlusion scene areas, preliminary fused positioning information is generated by integrating broadcast data from surrounding aircraft or ground stations and pose data relative to the local map.

[0077] The weight adjustment module 44 includes: The second division unit is used to divide multi-source data into open scene, slightly occluded scene and heavily occluded scene based on the geometric precision factor of Beidou absolute position data, the confidence of pose data relative to local map, and the loss rate of broadcast data from surrounding aircraft or ground stations. The sub-configuration unit is used to configure the following in open scenarios: the basic weights of BeiDou absolute position data, pose data relative to the local map, and broadcast data from surrounding aircraft or ground stations decrease sequentially; in slightly occluded environments: the basic weights of pose data relative to the local map, BeiDou absolute position data, and broadcast data from surrounding aircraft or ground stations decrease sequentially; in heavily occluded environments: the basic weights of pose data relative to the local map, broadcast data from surrounding aircraft or ground stations, and BeiDou absolute position data decrease sequentially. The parsing and correction unit is used to parse voice commands in real time, determine the task category based on the voice commands, and correct the basic weights according to the mapping table between the task category and the weight correction amount. Other correction units are used to temporarily freeze the basic weight of a certain multi-source data when a certain multi-source data is not received for more than a preset time or when the feature matching of a certain multi-source data fails for multiple consecutive frames, until the freeze is lifted when the certain multi-source data is received again or when the feature matching of a certain multi-source data is successful.

[0078] The computational unit is used to calculate the equivalent variance of each multi-source data. The normalization processing unit is used to take the reciprocal of the equivalent variance of each multi-source data and normalize it to obtain the dynamic weight of each multi-source data.

[0079] State fusion module 45 is used to update the state vector at time K based on Kalman gain and observation residual. Covariance Matrix The fused position, velocity, and attitude estimates are obtained; where: State vector ; To initially fuse the vectors corresponding to the positioning information, h( ) is the observation function; For Kalman gain, For the observation function h ( The Jacobian matrix is ​​represented by h( )exist The result is obtained by first-order Taylor expansion at the location; for The transpose of the matrix; To measure the noise covariance matrix: in , , The weighted averages are the absolute position data from BeiDou, the pose data relative to the local map, and the broadcast data from nearby aircraft or ground stations. , , These are the base noise levels for BeiDou absolute position data, pose data relative to a local map, and broadcast data from nearby aircraft or ground stations, calibrated according to industry standards and equipment accuracy parameters; diag ( ) represents a diagonal matrix; covariance matrix It is an identity matrix with the same dimensions as the state vector.

[0080] In summary, the multi-source data fusion positioning and monitoring method and system for low-altitude aircraft provided by this invention have at least the following beneficial effects compared to the prior art: 1. Multi-source data fusion architecture: Based on the local high-precision correction characteristics of Beidou absolute position data of low-altitude aircraft, pose data relative to local maps, and broadcast data from surrounding aircraft or ground stations, this invention combines preliminary fusion and state fusion to solve the reliability problem of single data sources in complex scenarios such as obstruction and signal failure.

[0081] 2. Dynamic Adaptive Weight Adjustment: This invention dynamically allocates the comprehensive weights of various multi-source data based on scene classification (open / slight / heavy occlusion) and real-time data reliability. It also introduces language command assistance and a sensor failure freeze mechanism to further improve the localization adaptability under complex tasks.

[0082] 3. Adaptability to complex low-altitude environments: In response to scenarios such as obstruction by tall buildings in cities and multipath reflection in mountainous areas, the system fills the gaps in BeiDou absolute position data by using pose data from local maps, and coordinates monitoring with broadcast data from surrounding aircraft or ground stations to achieve high-precision positioning and airspace surveillance in diverse environments.

[0083] Those skilled in the art will understand that the modules in the above system embodiments can be distributed in the device as described, or they can be modified accordingly and distributed in one or more devices different from the above embodiments. The modules in the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0084] The following describes embodiments of the electronic device of the present invention, which can be considered as implementations of the physical form of the methods and apparatus embodiments of the present invention described above. Details described in the embodiments of the electronic device of the present invention should be considered as supplements to the methods or apparatus embodiments described above; details not disclosed in the embodiments of the electronic device of the present invention can be implemented with reference to the methods or systems embodiments described above.

[0085] Figure 5 This is a structural block diagram of an exemplary embodiment of an electronic device according to the present invention. Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0086] like Figure 5 As shown, the electronic device 500 of this exemplary embodiment is manifested in the form of a general data processing device. The components of the electronic device 500 may include, but are not limited to: at least one processing unit 510, at least one storage unit 520, a bus 530 connecting different electronic device components (including storage unit 520 and processing unit 510), a display unit 540, etc.

[0087] The storage unit 520 stores a computer-readable program, which may be source code or read-only code. The program can be executed by the processing unit 510, causing the processing unit 510 to perform the steps of various embodiments of the present invention. For example, the processing unit 510 can perform actions such as... Figure 1 The steps are shown.

[0088] Bus 530 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0089] Electronic device 500 can also communicate with one or more external devices 100 (e.g., keyboard, monitor, network device, Bluetooth device, etc.), enabling users to interact with electronic device 500 via these external devices 100, and / or enabling electronic device 500 to communicate with one or more other data processing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 550, and also via network adapter 560 to one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public network). Network adapter 560 can communicate with other modules of electronic device 500 via bus 530.

[0090] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the method described in any of the above-mentioned embodiments. Specifically: It collects multi-source data, including: BeiDou absolute position data of the low-altitude aircraft, pose data relative to a local map, and broadcast data from surrounding aircraft or ground stations; performs spatiotemporal alignment on the multi-source data after anomaly processing; divides the BeiDou absolute position data into different scene zones, and performs preliminary fusion of the aligned multi-source data within each scene zone to obtain preliminary fused positioning information; configures basic weights for each multi-source data under different flight scenarios, and calculates the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data; fuses the basic weights and dynamic weights of each multi-source data to obtain the real-time comprehensive weight of each multi-source data; and performs state-space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of each multi-source data to obtain the fused positioning result of the low-altitude aircraft.

[0091] In summary, the present invention can be implemented by methods, systems, electronic devices, or computer program products that execute computer programs. In practice, some or all of the functions of the present invention can be implemented using general-purpose data processing devices such as microprocessors or digital signal processors (DSPs).

[0092] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the present invention is not inherently related to any specific computer, virtual device, or electronic device, and various general-purpose devices can also implement the present invention. The above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-source data fusion positioning and monitoring method for low-altitude aircraft, characterized in that, include: Collect multi-source data, including: BeiDou absolute position data of low-altitude aircraft, pose data relative to local maps, and broadcast data from surrounding aircraft or ground stations; After performing anomaly processing on the multi-source data, spatiotemporal alignment is then performed. The BeiDou absolute positioning data is divided into different scene areas. The aligned multi-source data in each scene area is initially fused to obtain preliminary fused positioning information. Configure the basic weights of each multi-source data under different flight scenarios, and calculate the dynamic weights of each multi-source data in real time based on the equivalent variance of each multi-source data. Then, merge the basic weights and dynamic weights of each multi-source data to obtain the real-time comprehensive weights of each multi-source data. Based on the preliminary fusion positioning information and the real-time comprehensive weights of various multi-source data, state space fusion is performed to obtain the fusion positioning result of the low-altitude aircraft.

2. The method according to claim 1, characterized in that, The abnormal data processing of the multi-source data includes: For BeiDou absolute position data, the severely obstructed BeiDou absolute position data is filtered by the geometric precision factor threshold. At the same time, pseudorange change rate is detected based on sliding window statistics, and BeiDou absolute position data corresponding to pseudorange jumps exceeding the preset threshold are removed. For the pose data of the relative local map, the pose data of the relative local map with pose jumps are determined and removed based on the relationship between the pose change of adjacent frames and the second threshold, and the pose data of the relative local map with confidence less than the threshold are filtered. For broadcast data from surrounding aircraft or ground stations, verify the integrity of the broadcast data, and remove broadcast data with sudden changes in position, broadcast data with speeds exceeding the maximum maneuvering speed threshold of the aircraft, and abnormal broadcasts with timestamp conflicts.

3. The method according to claim 2, characterized in that, The spatiotemporal alignment of the multi-source data after anomaly processing includes: Using the time of BeiDou absolute position data reception as the reference time axis, the pose data relative to the local map and the broadcast data of surrounding aircraft or ground stations are aligned to the time of BeiDou absolute position data reception through linear interpolation. Using the center of the global map as the origin, the BeiDou absolute position data and the broadcast data of surrounding aircraft or ground stations are transformed to the northeast celestial coordinates under the same global center origin through Gauss-Kruger projection; the aircraft pose data in the pose data relative to the local map are transformed to the northeast celestial coordinates under the same global center origin through origin transformation; and the ground feature point data in the pose data relative to the local map are used to calculate the northeast celestial coordinates of the ground feature points through inverse perspective transformation of the camera intrinsic parameter matrix and the aircraft attitude angle.

4. The method according to claim 1, characterized in that, The process involves dividing the data into different scene zones based on BeiDou absolute positioning data, and then performing preliminary fusion of the aligned multi-source data within each scene zone to obtain preliminary fused positioning information, including: Based on the parameter thresholds that characterize data quality, BeiDou absolute position data are divided into stable scene area, first occlusion scene area and second occlusion scene area; For the first obstruction scenario area, the drift error of the BeiDou absolute position data is corrected by using the reference position weighted average of the ground station broadcast data to generate preliminary fused positioning information; For stable scene areas, the pose data relative to the local map is aligned with the BeiDou absolute position data through initial extrinsic parameter calibration to construct a local map and generate preliminary fused positioning information; For the second occlusion scene area, preliminary fusion positioning information is generated by combining broadcast data from surrounding aircraft or ground stations and pose data relative to the local map.

5. The method according to claim 4, characterized in that, The basic weights for configuring various multi-source data under different flight scenarios include: Based on the geometric precision factor of BeiDou absolute position data, the confidence level of pose data relative to local maps, and the loss rate of broadcast data from surrounding aircraft or ground stations, multi-source data are divided into open scene, slightly occluded scene, and heavily occluded scene. In open environments: the basic weights of BeiDou absolute position data, pose data relative to the local map, and broadcast data from surrounding aircraft or ground stations decrease in that order; in slightly obstructed environments: the basic weights of pose data relative to the local map, BeiDou absolute position data, and broadcast data from surrounding aircraft or ground stations decrease in that order; in heavily obstructed environments: the basic weights of pose data relative to the local map, broadcast data from surrounding aircraft or ground stations, and BeiDou absolute position data decrease in that order. Real-time parsing of voice commands, determination of task categories based on voice commands, and adjustment of basic weights based on a mapping table between task categories and weight adjustment amounts; If a certain multi-source data is not received within a preset time or if feature matching of a certain multi-source data fails for multiple consecutive frames, the basic weight of the certain multi-source data will be temporarily frozen until the certain multi-source data is received again or feature matching of the certain multi-source data is successful, at which point the freeze will be lifted.

6. The system according to claim 1, characterized in that, The real-time calculation of the dynamic weights of each multi-source data based on the equivalent variance of each multi-source data includes: Calculate the equivalent variance of each multi-source data; The dynamic weights of each multi-source data are obtained by taking the reciprocal of the equivalent variance of each multi-source data and normalizing it.

7. The system according to claim 1, characterized in that, The process of performing state-space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft includes: Update the state vector at time K based on the Kalman gain and observation residuals. Covariance Matrix The fused position, velocity, and attitude estimates are obtained; where: State vector ; To initially fuse the vectors corresponding to the positioning information, h( ) is the observation function; For Kalman gain, For the observation function h ( The Jacobian matrix is ​​represented by h( )exist The result is obtained by first-order Taylor expansion at the location; for The transpose of the matrix; To measure the noise covariance matrix: in , , The weighted averages are the absolute position data from BeiDou, the pose data relative to the local map, and the broadcast data from nearby aircraft or ground stations. , , These are the base noise levels for BeiDou absolute position data, pose data relative to a local map, and broadcast data from nearby aircraft or ground stations, calibrated according to industry standards and equipment accuracy parameters; diag ( ) represents a diagonal matrix; covariance matrix It is an identity matrix with the same dimensions as the state vector.

8. A multi-source data fusion positioning and monitoring system for low-altitude aircraft, characterized in that, include: The acquisition module is used to acquire multi-source data, which includes: BeiDou absolute position data of low-altitude aircraft, pose data relative to a local map, and broadcast data from surrounding aircraft or ground stations. The alignment module is used to perform spatiotemporal alignment on the multi-source data after anomaly processing. The preliminary fusion module is used to divide the BeiDou absolute position data into different scene areas. The aligned multi-source data in each scene area is initially fused to obtain preliminary fused positioning information. The weight adjustment module is used to configure the basic weights of various multi-source data under different flight scenarios, and to calculate the dynamic weights of various multi-source data in real time based on the equivalent variance of each multi-source data. The basic weights and dynamic weights of each multi-source data are then integrated to obtain the real-time comprehensive weights of each multi-source data. The state fusion module is used to perform state space fusion based on the preliminary fused positioning information and the real-time comprehensive weights of various multi-source data to obtain the fused positioning result of the low-altitude aircraft.

9. An electronic device, comprising: processor; And a memory storing computer-executable instructions, which, when executed, cause the processor to perform the method according to any one of claims 9 to 16.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 9 to 16.

Images