Vehicle accurate positioning data fusion processing system based on Beidou satellite navigation communication

Abstract

The invention relates to the technical field of computer data processing, in particular to a vehicle accurate positioning data fusion processing system based on Beidou satellite navigation communication. A vehicle accurate positioning data fusion processing system based on Beidou satellite navigation communication comprises the following parts: a data acquisition and state subset extraction module for acquiring vehicle positioning data and executing action positioning data, and acquiring a vehicle state subset in an action period according to the vehicle positioning data; the action positioning sequence generation module is used for determining a navigation coordinate system offset vector according to the action period vehicle state subset; and synthesizing the absolute position of the vehicle and the offset of the navigation coordinate system to obtain an operation end absolute coordinate of the operation end of the execution mechanism at the current moment, and obtaining an execution action positioning time sequence according to the operation end absolute coordinate. According to the invention, Beidou and inertial data are fused, a positioning fusion strategy is dynamically optimized, and the operation precision and robustness are improved.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, and in particular to a vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication. Background Technology

[0002] Vehicle precision positioning and collaborative operation systems are core supporting technologies for modern agricultural and industrial automation. The accuracy of their data processing directly determines the execution efficiency and resource utilization of the machinery. Existing technologies, by integrating BeiDou satellite navigation and positioning modules with inertial measurement units, have achieved centimeter-level position perception for vehicle platforms. Furthermore, based on a fixed-weight Kalman filter algorithm, significant results have been achieved in path tracking control under open field conditions, laying the foundation for the widespread application of automated operations.

[0003] However, existing methods face severe challenges when dealing with complex and dynamic operational scenarios. Traditional data fusion strategies often employ static parameter configurations, making it difficult to adapt to the switching of positioning sources and fluctuations in data quality caused by the coupling of multiple factors such as signal obstruction, mechanical vibration, and carrier movement. They also lack a quantitative evaluation mechanism for the temporal stability and source consistency of positioning data streams, which leads to positioning jumps and motion deviations during continuous operations such as precision spraying and automatic harvesting, failing to meet the actual data reliability requirements of high-precision operational tasks.

[0004] Therefore, there is an urgent need to study an intelligent positioning data processing method that can adapt to complex working conditions. By performing temporal correlation analysis and dynamic strategy adjustment on multi-source positioning data, the positioning robustness and motion execution accuracy of vehicles in dynamic working environments can be improved. Summary of the Invention

[0005] To overcome the drawback of unstable positioning data sources, this invention provides a vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication.

[0006] The technical implementation scheme of the present invention is as follows: a vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication, comprising the following parts: The data acquisition and state subset extraction module acquires vehicle positioning data and execution action positioning data, and obtains a vehicle state subset for the action period based on the vehicle positioning data. The action positioning sequence generation module determines the navigation coordinate system offset vector based on the vehicle state subset during the action period; it synthesizes the absolute position of the vehicle with the navigation coordinate system offset to obtain the absolute coordinates of the actuator's working end at the current moment, and obtains the action positioning time sequence based on the absolute coordinates of the working end. The association construction and directed graph generation module determines the first association and the second association based on the execution action location time sequence; and constructs a directed graph based on the first association and the second association. The action cluster analysis and feature extraction module obtains a set of continuous action clusters based on the directed graph, and determines the first sequence traceability rate, the second sequence dependency, and the BeiDou positioning integrity index based on the set of continuous action clusters. The data fusion strategy decision module determines the vehicle precise positioning data fusion processing method based on the first sequence traceability rate, the second sequence dependency degree, and the Beidou positioning integrity index.

[0007] Preferably, the data acquisition and state subset extraction module acquires vehicle positioning data and execution action positioning data, including: The vehicle positioning data refers to the measurement information of the vehicle platform's own position and attitude in the global coordinate system; The execution action positioning data refers to the calculated location information of the working end of the vehicle's actuator in the actual geographical space; The vehicle positioning data is reorganized into a first vehicle positioning time sequence according to a fixed update frequency; The vehicle positioning data is reorganized into a second vehicle positioning time sequence according to the actual update frequency.

[0008] Preferably, obtaining the vehicle status subset during the action period based on the vehicle positioning data includes: Based on the start and end timestamps of the actions of the actuator's working end, each data point of the actuator's working end from the start to the end of the action is extracted from the vehicle positioning data to form a subset of the vehicle status during the action period.

[0009] Preferably, the action positioning sequence generation module determines the navigation coordinate system offset vector based on the vehicle state subset during the action period, including: For each data point in the vehicle state subset during the aforementioned action period, perform the following calculations: Obtain the platform roll angle, platform pitch angle, and platform yaw angle of the data points; A three-dimensional rotation matrix is ​​constructed based on the platform roll angle, platform pitch angle, and platform yaw angle. Multiply the fixed actuator working end lever parameters by the three-dimensional rotation matrix to calculate the navigation coordinate system offset vector in the ENU coordinate system; The offset vector of the navigation coordinate system is obtained based on the spatial coordinate transformation formula of the actuator's working end. The spatial coordinate transformation formula of the actuator's working end is: ; in, , and For the lever parameters at the working end of the actuator, For the platform roll angle, The platform's pitch angle, For the platform's heading angle, It is a three-dimensional rotation matrix. , and This is the offset vector of the navigation coordinate system.

[0010] Preferably, the step of synthesizing the absolute position of the vehicle with the offset of the navigation coordinate system to obtain the absolute coordinates of the actuator's working end at the current moment, and obtaining the execution action positioning time sequence based on the absolute coordinates of the working end, includes: the synthesis formula is: ; ; ; in, , and For the absolute coordinates of the vehicle platform, , This is the parameter for the Earth's radius of curvature. , and The coordinates are the absolute coordinates of the working end; The absolute coordinates of the working end are reorganized in chronological order to form a timing sequence of the execution action positioning. The execution action positioning timing sequence refers to the data sequence of the continuous motion trajectory of the working end of the actuator in the global coordinate system during the execution of the action.

[0011] Preferably, the association construction and directed graph generation module determines the first association and the second association based on the execution action location time sequence, including: A tolerance threshold is set. For each data point in the execution action positioning time sequence, a first difference is calculated between the timestamp of the data point and the timestamp corresponding to the first vehicle positioning time sequence. If the first difference is less than the tolerance threshold, the data point is associated with the first vehicle positioning time sequence to obtain a first association relationship. If the first difference is greater than or equal to the tolerance threshold, then the second difference between the timestamp of the data point and the timestamp corresponding to the second vehicle positioning time sequence is calculated. If the second difference is less than the tolerance threshold, then the data point is associated with the second vehicle positioning time sequence to obtain a second association relationship.

[0012] Preferably, the step of constructing a directed graph based on the first association relationship and the second association relationship includes: Each data point in the first vehicle positioning time sequence is taken as the first sequence node; Each data point in the second vehicle positioning time sequence is used as a second sequence node; Each data point in the action location time sequence is used as an action sequence node; The first sequence node and the second sequence node are collectively referred to as the source sequence node. For each pair of action sequence nodes and source sequence nodes that are associated through the first association relationship or the second association relationship, a directed edge is created from the action sequence node to the source sequence node, and a directed graph is obtained based on the directed edge.

[0013] Preferably, the action cluster analysis and feature extraction module obtains a set of continuous action clusters based on the directed graph, including: Set a maximum time interval threshold, extract all action sequence nodes from the directed graph, and sort and traverse them based on the timestamps of the action sequence nodes; merge action sequence nodes with consecutive timestamps and adjacent intervals less than the maximum time interval threshold into the same cluster to obtain a set of continuous action clusters.

[0014] Preferably, determining the first sequence tracing rate, the second sequence dependency, and the BeiDou positioning integrity index based on the set of continuous action clusters includes: For each continuous action cluster in the set of continuous action clusters, calculate the first sequence traceability rate and the second sequence dependency. The first sequence traceability rate is used to measure the degree to which the location data of a continuous action cluster originates from an ideal location sequence; The second sequence dependency is used to measure the degree to which the localization data of a continuous action cluster depends on the actual localization sequence; The formula for calculating the first sequence traceability rate is: ;in, This represents the number of action points within a continuous action cluster that are associated with the first sequence node. The total number of action points within a continuous action cluster. The first sequence traceability rate; The formula for calculating the second sequence dependency is: ;in, This represents the second sequence dependency. For each action point within a continuous action cluster, the location source node is found through the first association relationship and the second association relationship; Extract the carrier-to-noise ratio, number of satellites, and accuracy factor from the BeiDou satellite navigation and positioning data corresponding to the positioning source node; The average values ​​of the carrier-to-noise ratio, number of satellites, and accuracy factor for all action points within a continuous action cluster are calculated and normalized to obtain the BeiDou positioning integrity index.

[0015] Preferably, the data fusion strategy decision module determines the vehicle precise positioning data fusion processing method based on the first sequence traceability rate, the second sequence dependency, and the BeiDou positioning integrity index, including: Weighting coefficients are assigned to BeiDou satellite navigation and positioning data and attitude data based on a weighted least squares estimation model. The optimal estimate of the system state is obtained by solving for the system state that minimizes the sum of squared weighted errors. The allocation of the weighting coefficients is dynamically determined based on the traceability consistency coefficient, the BeiDou positioning integrity index, and the first sequence traceability rate. The dynamic determination includes: For the current continuous action cluster, the source consistency coefficient is calculated based on the first and second association relationships of each action sequence node within the continuous action cluster; The source consistency coefficient is used to quantify the stability of the location data source within a continuous action cluster; the formula for calculating the source consistency coefficient is: ; in, Let be the traceability consistency coefficient, when or When the logarithm is 0, the corresponding logarithmic term is defined as 0; If the traceability consistency coefficient is greater than or equal to the preset traceability consistency threshold, then the continuous action cluster is determined to be a high inconsistency cluster. If it is a highly inconsistent cluster, increase the weight of attitude data and decrease the weight of BeiDou satellite navigation and positioning data; If the traceability consistency coefficient is less than the preset traceability consistency threshold, the continuous action cluster is determined to be a low inconsistency cluster. If it is a low-inconsistency cluster, when the BeiDou positioning integrity index is less than the BeiDou positioning integrity index threshold, the weight of BeiDou satellite navigation and positioning data is reduced and the weight of attitude data is increased; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is greater than or equal to the first sequence traceability rate threshold, BeiDou satellite navigation and positioning data is used as the core observation source; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is less than the first sequence traceability rate threshold, a high-frequency attitude priority fusion strategy is executed; the high-frequency attitude priority fusion strategy is a fusion mode that takes high-frequency attitude data as the main factor, constructs a smooth trajectory benchmark through its integration results, and then uses BeiDou satellite navigation and positioning data for global correction.

[0016] Beneficial Effects: This invention utilizes deep fusion of BeiDou satellite navigation communication and inertial measurement data to construct a time-series sequence of vehicle positioning data and executed action positioning data. Based on a subset of vehicle states during the action period, it dynamically generates a navigation coordinate system offset vector, achieving precise synthesis of the absolute coordinates of the actuator's operating end. By establishing a directed graph topology of first and second association relationships, a set of continuous action clusters is formed. The traceability rate of the first sequence, the dependency of the second sequence, and the BeiDou positioning integrity index are quantified to comprehensively evaluate the stability and quality fluctuations of the positioning data source. Furthermore, a traceability consistency coefficient is used to dynamically assess the degree of data source disorder, and a multi-threshold discrimination mechanism adaptively adjusts the fusion weights of BeiDou positioning data and attitude data, achieving intelligent strategy switching from high-inconsistency clusters to low-inconsistency clusters. This effectively overcomes the insufficient adaptability of traditional static parameter fusion in complex dynamic operating scenarios, significantly improving the robustness of vehicle precise positioning and action execution accuracy, and enhancing the system's intelligent identification and adaptation capabilities to multi-factor coupling interference from signal obstruction, mechanical vibration, and carrier maneuvering. Attached Figure Description

[0017] Figure 1 This is a structural diagram of the vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication according to the present invention; Figure 2 The flowchart for constructing the first and second association relationships of this invention is shown below. Detailed Implementation

[0018] The present invention will be further described below with reference to specific embodiments. The illustrative embodiments and descriptions herein are used to explain the present invention, but are not intended to limit the present invention.

[0019] A vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication, such as Figure 1 and Figure 2 As shown, it includes the following parts: The data acquisition and state subset extraction module acquires vehicle positioning data and execution action positioning data. The vehicle positioning data refers to the measurement information of the vehicle platform's own position and attitude in the global coordinate system; The execution action positioning data refers to the calculated location information of the working end of the vehicle's actuator in the actual geographical space; The vehicle positioning data is reorganized into a first vehicle positioning time sequence according to a fixed update frequency; The vehicle positioning data is reorganized into a second vehicle positioning time sequence according to the actual update frequency.

[0020] It should be noted that, to address the trajectory deviation issue caused by unstable data sources in traditional vehicle positioning systems, this solution acquires the vehicle's longitude, latitude, and elevation data in the WGS-84 coordinate system through a BeiDou satellite navigation module. Simultaneously, it collects the platform's roll, pitch, and yaw angles through an inertial measurement unit, collectively forming the vehicle's positioning data. Based on the mechanical parameters of the actuator's operating end, the absolute position of the operating end is calculated through spatial coordinate transformation, such as the three-dimensional coordinates of the harvesting head of an automatic harvester, forming the positioning data for the executed actions.

[0021] To address the issue of inconsistent data timing in dynamic environments, the original data is constructed into a dual-sequence structure: a fixed-frequency sequence is organized with strictly equal intervals, forming an ideal sequence [08:00:00.000, 08:00:00.010, 08:00:00.020...]; the actual frequency sequence retains the original acquisition characteristics, constructing a non-uniform sequence [08:00:00.005, 08:00:00.016, 08:00:00.027...]. This dual-sequence construction method effectively solves the data mismatch problem of traditional single-time-sequence streams under complex conditions by establishing a comparison between the ideal timing benchmark and the actual acquisition characteristics, providing a processing foundation with timing alignment and quality classification capabilities for subsequent data fusion.

[0022] A subset of vehicle status during the action period is obtained based on the vehicle positioning data; Based on the start and end timestamps of the actions of the actuator's working end, each data point of the actuator's working end from the start to the end of the action is extracted from the vehicle positioning data to form a subset of the vehicle status during the action period.

[0023] It should be noted that, addressing the issue of trajectory deviation caused by the mismatch between positioning data and the timing of operational actions during traditional vehicle operations, this solution achieves precise data correlation by defining a subset of vehicle states during the action period. The actuator refers to the specific operational device, such as the spraying arm of an automatic sprayer. The start and end timestamps record the start time 08:00:00.000 and the end time 08:00:15.000. Instead of extracting data from a single raw data stream, this solution simultaneously extracts all data points within the specified action period from both the constructed "First Vehicle Positioning Timing Sequence" (fixed-frequency ideal sequence) and the "Second Vehicle Positioning Timing Sequence" (actual-frequency real sequence). Each data point contains complete status information, including a timestamp, latitude and longitude, and attitude angle. Subsequently, the data points from the two sequences are merged and aggregated to form a unified subset of vehicle states during the action period. This subset can be represented as a mixed time-series sequence containing data points from different sources: [08:00:00.100 (from the first sequence), 08:00:00.203 (from the second sequence), 08:00:00.210 (from the first sequence), ..., 08:00:14.900]. This data extraction and aggregation method based on dual-series sources and timestamp windows not only filters out redundant data unrelated to the action, ensuring the temporal consistency of the basic data for analysis, but also lays a reliable foundation for subsequent modules to analyze the data source quality of each action point (whether it originates from an ideal sequence or the actual sequence) and perform intelligent data fusion accordingly. This effectively solves the control error problem caused by mixed data sources and asynchronous temporal sequences.

[0024] The action positioning sequence generation module determines the navigation coordinate system offset vector based on the vehicle state subset during the action period. For each data point in the vehicle state subset during the aforementioned action period, perform the following calculations: Obtain the platform roll angle, platform pitch angle, and platform yaw angle of the data points; A three-dimensional rotation matrix is ​​constructed based on the platform roll angle, platform pitch angle, and platform yaw angle. Multiply the fixed actuator working end lever parameters by the three-dimensional rotation matrix to calculate the navigation coordinate system offset vector in the ENU coordinate system; The offset vector of the navigation coordinate system is obtained based on the spatial coordinate transformation formula of the actuator's working end. The spatial coordinate transformation formula of the actuator's working end is: ; in, , and For the lever parameters at the working end of the actuator, For the platform roll angle, The platform's pitch angle, For the platform's heading angle, It is a three-dimensional rotation matrix. , and This is the offset vector of the navigation coordinate system.

[0025] It should be noted that this step achieves accurate calculation of the navigation coordinate system offset vector through the motion positioning sequence generation module. The platform roll angle, pitch angle, and yaw angle are acquired in real time by an embedded inertial measurement unit. The roll angle represents the vehicle's lateral tilt, the pitch angle reflects the longitudinal pitch state, and the yaw angle defines the driving direction reference. The three-dimensional rotation matrix constructed based on the platform roll angle, pitch angle, and yaw angle is essentially a spatial transformation operator describing the rigid body attitude, used to map the vehicle's body coordinate system to the Northeast-East-South navigation coordinate system. The actuator's working end lever parameters are fixed geometric properties. Taking the automatic sprinkler boom as an example, its spatial relationship with the vehicle's center of mass is expressed as […]. =1.2m, =0.5m, =0.8m]. By multiplying the lever arm parameters by the rotation matrix using the coordinate transformation formula, the resulting navigation coordinate system offset vector is [ , , Essentially, this is the projected displacement of the operating end in the navigation coordinate system. This calculation process effectively overcomes the adaptability defects of traditional static parameter fusion under complex attitudes through a dynamic coordinate transformation mechanism, and establishes a geometric consistency guarantee for subsequent data fusion. , and It refers to the eastward, northward, and celestial offsets of the actuator's working end relative to the vehicle's center of gravity in the Northeast-Sky (ENU) navigation coordinate system. , and It is the fixed three-dimensional offset of the actuator's working end relative to the vehicle's center of mass in the vehicle's coordinate system.

[0026] The absolute position of the vehicle is combined with the offset of the navigation coordinate system to obtain the absolute coordinates of the actuator at the current moment. The positioning timing sequence of the execution action is obtained based on the absolute coordinates of the actuator. The synthesis formula is: ; ; ; in, , and For the absolute coordinates of the vehicle platform, , This is the parameter for the Earth's radius of curvature. , and The coordinates are the absolute coordinates of the working end; The absolute coordinates of the working end are reorganized in chronological order to form a timing sequence of the execution action positioning. The execution action positioning timing sequence refers to the data sequence of the continuous motion trajectory of the working end of the actuator in the global coordinate system during the execution of the action.

[0027] It should be noted that, to overcome the problem of motion deviation caused by inaccurate position calculation in traditional machinery in complex terrain, this solution adopts a spatial coordinate synthesis method to establish a precise positioning model for the operating end. The absolute position of the vehicle is obtained through the BeiDou satellite navigation system, specifically represented by longitude 118.5°, latitude 32.8°, and elevation 50.2 meters in the WGS-84 coordinate system. Taking the harvesting head of an automatic harvester as an example, the operating end of the actuator integrates the navigation coordinate system offset vector with the vehicle's absolute position through a three-dimensional coordinate transformation formula. The longitude component is calculated by incorporating the Earth's radius of curvature parameter for radian conversion, while the elevation component is directly superimposed using algebraic superposition. The resulting absolute coordinates of the operating end form a precise positioning result such as [118.50015°, 32.80012°, 50.35 meters]. These coordinates are then reassembled into an ordered sequence according to timestamps, for example, [08:00:00.100: position 1, 08:00:00.200: position 2…], constructing a complete timing sequence for the execution motion positioning. This time series clearly records the continuous motion trajectory of the working end in the global coordinate system, effectively solving the problem of trajectory breakage caused by imperfect coordinate transformation during dynamic operations in traditional positioning methods, and providing a high-precision spatiotemporal reference for subsequent data analysis. : is the radius of curvature of the Earth's geoid. : is the radius of curvature of the Earth's meridian. , and It is the longitude, latitude, and elevation of the vehicle's center of gravity in the WGS-84 coordinate system, measured by the BeiDou satellite navigation system. , and It is the longitude, latitude, and elevation of the actuator working end in the WGS-84 coordinate system, calculated by this method.

[0028] The association relationship construction and directed graph generation module determines the first association relationship and the second association relationship based on the execution action location time sequence; A tolerance threshold is set. For each data point in the execution action positioning time sequence, a first difference is calculated between the timestamp of the data point and the timestamp corresponding to the first vehicle positioning time sequence. If the first difference is less than the tolerance threshold, the data point is associated with the first vehicle positioning time sequence to obtain a first association relationship. If the first difference is greater than or equal to the tolerance threshold, then the second difference between the timestamp of the data point and the timestamp corresponding to the second vehicle positioning time sequence is calculated. If the second difference is less than the tolerance threshold, then the data point is associated with the second vehicle positioning time sequence to obtain a second association relationship.

[0029] It should be noted that, Figure 2 As shown, this step establishes a multi-level data traceability system through the association construction module. The first association represents the baseline correspondence between the executed action data points and the fixed frequency sequence, reflecting the stability characteristics of the ideal data source; the second association reflects the dynamic mapping relationship between the executed action data points and the actual frequency sequence, demonstrating the adaptability of the actual data source. The tolerance threshold is used to define the reasonable deviation range of time synchronization.

[0030] In practice, the first difference is used as the time offset between the action point and the ideal sequence. When this value is less than the tolerance threshold, it indicates that the data point is effectively synchronized with the reference sequence, and thus the first correlation is established to ensure data quality. If the first difference exceeds the tolerance range, the timing match between the action point and the actual sequence is evaluated using the second difference. When the second difference meets the tolerance requirements, the second correlation is established, forming an auxiliary data channel. For example, the timing deviation between the action point timestamp 08:00:00.103 and the first sequence reference point 08:00:00.100 is 3 milliseconds, which is lower than the set tolerance threshold, so the first correlation is established. When an action point with timestamp 08:00:00.112 is encountered, its deviation from the reference point exceeds the tolerance range, so the deviation between it and the actual sequence point 08:00:00.110 is calculated to be 2 milliseconds, and the second correlation is established accordingly.

[0031] This multi-level association mechanism based on dynamic threshold determination effectively solves the adaptability problem of traditional single-path data association under complex working conditions by establishing a clear primary and secondary data tracing path, and provides a reliable data source identification basis for subsequent data fusion analysis.

[0032] It should be noted that the tolerance threshold, the preset traceability consistency threshold, the BeiDou positioning integrity index threshold, and the first sequence traceability rate threshold are determined by one or a combination of the following methods: determined based on percentiles from historical data statistical analysis; obtained by parameter optimization with positioning accuracy and action execution stability as optimization objectives through system simulation tests in different typical scenarios; or assigned based on engineering experience values ​​according to the physical performance indicators of the system positioning unit and the execution mechanism's working end.

[0033] Construct a directed graph based on the first and second association relationships; Each data point in the first vehicle positioning time sequence is taken as the first sequence node; Each data point in the second vehicle positioning time sequence is used as a second sequence node; Each data point in the action location time sequence is used as an action sequence node; The first sequence node and the second sequence node are collectively referred to as the source sequence node. For each pair of action sequence nodes and source sequence nodes that are associated through the first association relationship or the second association relationship, a directed edge is created from the action sequence node to the source sequence node, and a directed graph is obtained based on the directed edge.

[0034] It should be noted that, to address the insufficient fusion accuracy caused by unclear data source relationships in traditional positioning systems, this step utilizes a directed graph to achieve a structured representation of data dependencies. The first sequence node represents ideal positioning data at a fixed frequency, the second sequence node corresponds to dynamic positioning data at the actual frequency, and the action sequence node represents the operational trajectory data of the actuator. The first and second sequence nodes, which provide the data source for the action nodes, are collectively referred to as "source sequence nodes." Based on the established first and second association relationships, a directed edge is created from the action node to the source node for each pair of associated "action sequence node-source sequence node." For example, action node A1 (timestamp 08:00:00.100) is connected to the first sequence node F1 (08:00:00.100), which is the source sequence node, through a first association relationship, forming a directed edge from A1 to F1; action node A2 (08:00:00.112) is connected to the second sequence node S1 (08:00:00.110), which is the source sequence node, through a second association relationship, forming a directed edge from A2 to S1. This directed graph construction mechanism based on unified definition and association pairs effectively solves the problem of unclear source paths in traditional data fusion, and provides a data relationship topology with a rigorous structure and clear relationships for subsequent feature extraction.

[0035] The action cluster analysis and feature extraction module obtains a set of continuous action clusters based on the directed graph; Set a maximum time interval threshold, extract all action sequence nodes from the directed graph, and sort and traverse them based on the timestamps of the action sequence nodes; merge action sequence nodes with consecutive timestamps and adjacent intervals less than the maximum time interval threshold into the same cluster to obtain a set of continuous action clusters.

[0036] It should be noted that, to address the analysis difficulties caused by the temporal breaks in action sequences in traditional positioning systems, this step achieves coherence analysis of executed actions through continuous action clusters. The continuous action cluster set consists of action sequence nodes with consecutive timestamps, used to characterize the spatiotemporal characteristics of a complete task segment. In implementation, firstly, all action sequence nodes are extracted from the constructed directed graph; these nodes already carry association information with the source sequence nodes. Then, ignoring the topological structure of these nodes in the graph, sorting and traversing are performed entirely based on their own timestamp attributes. When the time interval between adjacent action nodes after sorting is less than a set maximum time interval threshold, they are merged into the same cluster. Taking the extracted action node time sequence 08:00:00.100, 08:00:00.105, 08:00:00.115 as an example, if the threshold is set to 10 milliseconds, the first two nodes are grouped into the same cluster due to a 5-millisecond interval, while the third node is grouped into a new cluster due to a 10-millisecond interval. This "extracting from the graph and clustering in order" mechanism effectively solves the problem of discontinuous segmentation of action trajectories in traditional methods, and provides complete and high-quality data units for subsequent feature extraction based on the internal relationships of each cluster.

[0037] The first sequence traceability rate, the second sequence dependency, and the BeiDou positioning integrity index are determined based on the set of continuous action clusters. For each continuous action cluster in the set of continuous action clusters, calculate the first sequence traceability rate and the second sequence dependency. The first sequence traceability rate is used to measure the degree to which the location data of a continuous action cluster originates from an ideal location sequence; The second sequence dependency is used to measure the degree to which the localization data of a continuous action cluster depends on the actual localization sequence; The formula for calculating the first sequence traceability rate is: ;in, This represents the number of action points within a continuous action cluster that are associated with the first sequence node. The total number of action points within a continuous action cluster. The first sequence traceability rate; The formula for calculating the second sequence dependency is: ;in, This represents the second sequence dependency. For each action point within a continuous action cluster, the location source node is found through the first association relationship and the second association relationship; Extract the carrier-to-noise ratio, number of satellites, and accuracy factor from the BeiDou satellite navigation and positioning data corresponding to the positioning source node; The average values ​​of the carrier-to-noise ratio, number of satellites, and accuracy factor for all action points within a continuous action cluster are calculated and normalized to obtain the BeiDou positioning integrity index.

[0038] It should be noted that in the evolution of vehicle precise positioning systems, traditional methods often lack a systematic evaluation mechanism for data source quality, leading to a lack of objective basis for fusion decisions. This step establishes a feature extraction system, constructing a three-dimensional evaluation system: the first, the sequence traceability rate, represents the degree of consistency between action cluster data and the ideal sequence; an increase in this indicator indicates enhanced data source stability. The second, the sequence dependency, reveals the data's dependence on the actual sequence; its increase reflects the cumulative effect of environmental interference factors. The calculation model is based on the proportional relationship of the number of associated nodes. When the proportion of associated points in the ideal sequence increases, the traceability rate increases linearly, indicating improved data reliability; conversely, an increase in dependency indicates that the system needs to activate a fault-tolerance mechanism. The positioning source node refers to the first or second sequence node that provides the original BeiDou satellite navigation and positioning data to the action sequence node through the associated edges in the directed graph.

[0039] Through a pre-defined directed graph topology network, each action node is reverse-positioned to the BeiDou data source node along the associated edges. The carrier-to-noise ratio characterizes the signal's anti-interference capability, the number of satellites determines the spatial resolution, and the accuracy factor reflects the level of position calculation error. By averaging multi-dimensional indicators within the cluster to smooth instantaneous fluctuations, and then normalizing to unify the dimensions, a BeiDou positioning integrity index in the 0-1 range is finally generated. The formula for calculating this index is defined as follows: ; In the formula, The BeiDou positioning integrity index, The average carrier-to-noise ratio. The average number of satellites. The denominator uses the theoretical extreme values ​​of each parameter as the mean of the accuracy factor. This multi-parameter weighted fusion method effectively avoids the one-sidedness of traditional single-indicator evaluation and provides a complete data quality profile for dynamic fusion strategies.

[0040] The data fusion strategy decision module determines the vehicle precise positioning data fusion processing method based on the first sequence traceability rate, the second sequence dependency degree, and the Beidou positioning integrity index.

[0041] Weighting coefficients are assigned to BeiDou satellite navigation and positioning data and attitude data based on a weighted least squares estimation model. The optimal estimate of the system state is obtained by solving for the system state that minimizes the sum of squared weighted errors. The allocation of the weighting coefficients is dynamically determined based on the traceability consistency coefficient, the BeiDou positioning integrity index, and the first sequence traceability rate. The dynamic determination includes: For the current continuous action cluster, the source consistency coefficient is calculated based on the first and second association relationships of each action sequence node within the continuous action cluster; The source consistency coefficient is used to quantify the stability of the location data source within a continuous action cluster; the formula for calculating the source consistency coefficient is: ; in, Let be the traceability consistency coefficient, when or When the logarithm is 0, the corresponding logarithmic term is defined as 0; If the traceability consistency coefficient is greater than or equal to the preset traceability consistency threshold, then the continuous action cluster is determined to be a high inconsistency cluster. If it is a highly inconsistent cluster, increase the weight of attitude data and decrease the weight of BeiDou satellite navigation and positioning data; If the traceability consistency coefficient is less than the preset traceability consistency threshold, the continuous action cluster is determined to be a low inconsistency cluster. If it is a low-inconsistency cluster, when the BeiDou positioning integrity index is less than the BeiDou positioning integrity index threshold, the weight of BeiDou satellite navigation and positioning data is reduced and the weight of attitude data is increased; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is greater than or equal to the first sequence traceability rate threshold, BeiDou satellite navigation and positioning data is used as the core observation source; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is less than the first sequence traceability rate threshold, a high-frequency attitude priority fusion strategy is executed; the high-frequency attitude priority fusion strategy is a fusion mode that takes high-frequency attitude data as the main factor, constructs a smooth trajectory benchmark through its integration results, and then uses BeiDou satellite navigation and positioning data for global correction.

[0042] It should be noted that in dynamic operating environments, traditional positioning systems often struggle to adapt to fluctuations in data source quality due to fixed parameter configurations, leading to significant deviations in the fusion results.

[0043] This solution achieves adaptive optimization of multi-source data by constructing an intelligent fusion architecture based on weighted least squares. The weighted least squares estimation model is expressed as follows: ; in, : Optimal estimate of system state (including position, velocity, and attitude). BeiDou positioning data weighting coefficient Attitude data weighting coefficients Raw observations from BeiDou satellites Attitude sensor measurement value BeiDou observation matrix Attitude observation matrix; The BeiDou satellite navigation and positioning data specifically refers to the absolute positioning information directly provided by the BeiDou satellite navigation system in the WGS-84 global geocentric coordinate system. Its core observation values ​​include at least longitude, latitude, and elevation, and can derive parameters reflecting signal quality integrity such as carrier-to-noise ratio, number of visible satellites, and accuracy factor.

[0044] The attitude data specifically refers to the relative measurement information provided by the inertial measurement unit (IMU) to describe the three-dimensional spatial attitude of the vehicle platform itself. Its core observations include at least the platform roll angle, platform pitch angle, and platform yaw angle, and can be further calculated by integration to obtain angular velocity, acceleration, and displacement changes.

[0045] It should be noted that the aforementioned BeiDou observation matrix It is a matrix that maps the system state vector to BeiDou satellite navigation observations. Its specific form depends on the selected system state. For example, if the system state... It includes three-dimensional position and three-dimensional velocity, while BeiDou observations If longitude, latitude, and elevation information are provided, then... It is an observation mapping matrix from the state space to the WGS-84 geographic coordinate system. Its core function is to extract the position components that directly correspond to the BeiDou observation values ​​from the system state.

[0046] The attitude observation matrix It is a matrix that maps the system state vector to attitude sensor observations. Similarly, its specific form is closely related to the definition of the system state. For example, if the system state... Includes attitude angles (roll, pitch, yaw), while attitude observations If these angles are measured directly by the IMU, then This simplifies to an identity matrix, or a transformation matrix from attitude quaternion states to Euler angle observations. If the states also include gyroscope bias, then... It also needs to include the corresponding mapping relationship.

[0047] Through these two observation matrices, the system can correlate observation data with different physical meanings and coordinate systems (absolute positioning BeiDou data and relative measurement attitude data) with the system state to be estimated under a unified weighted least squares estimation framework, laying the foundation for subsequent dynamic weight allocation and optimal state solution.

[0048] This scheme uses the traceability consistency coefficient as a quantitative indicator of data source stability. This coefficient reflects the degree of disorder in the data source based on information entropy theory. Its calculation model follows the principle of information uncertainty: when the ideal sequence traceability rate or the actual sequence dependency deviates from the equilibrium distribution, the system disorder decreases accordingly; when the two are in equilibrium, the disorder reaches its maximum value. Due to the limitations of the mathematical definition, when the ideal sequence traceability rate or the actual sequence dependency takes a zero value, the corresponding logarithmic terms are zeroed out to ensure the completeness of the formula.

[0049] The decision-making process for dynamic weights is illustrated in the following example: Handling highly inconsistent clusters: When When the value is greater than or equal to a preset consistency threshold (e.g., 0.8), it is identified as a high inconsistency cluster. In this case, the system significantly increases the weight of the attitude data (e.g., by setting...). =0.7), reducing the weight of BeiDou satellite navigation and positioning data (e.g., =0.3). This means that the state estimation mainly relies on the IMU integral trajectory, with BeiDou data only playing an auxiliary correction role.

[0050] Handling low-inconsistency clusters: When When the value is less than a preset threshold, it is identified as a low-inconsistency cluster, and a triple discrimination mechanism is activated: If the BeiDou positioning integrity index is below its threshold (e.g., 0.6), the weight of BeiDou satellite navigation and positioning data will be reduced. =0.4), increasing the weight of attitude data ( =0.6), forming a robust integration model with attitude data as the main component and BeiDou as the auxiliary component.

[0051] If the BeiDou positioning integrity index is good (≥0.6) and the first sequence traceability rate is extremely high (≥0.9), it indicates that the ideal data source is dominant and the quality is reliable. In this case, BeiDou satellite navigation and positioning data will be used as the core observation source. =1, =0), fully relying on BeiDou observation values.

[0052] If only the BeiDou positioning integrity index condition (≥0.6) is met, but the first sequence tracing rate is low, then a high-frequency attitude priority fusion strategy is implemented. This strategy uses high-frequency attitude data as the primary source to construct a smooth trajectory benchmark, and then uses BeiDou satellite navigation and positioning data for global correction, achieving fusion solution within a weighted least squares framework.

[0053] It should be noted that the weighting coefficient of the BeiDou positioning data... Weighting coefficients of attitude data The assignment satisfies the relation + =1. Its specific value is obtained by directly mapping the source consistency coefficient, Beidou positioning integrity index, and the first sequence source tracing rate to the decision interval. The correspondence between each decision interval and the weight coefficient is determined through a pre-offline system calibration experiment, which aims to minimize the root mean square error between the fused positioning trajectory and the real reference trajectory.

[0054] The above description is only a preferred embodiment of the present invention and is 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 vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication, characterized in that, Includes the following parts: The data acquisition and state subset extraction module acquires vehicle positioning data and execution action positioning data, and obtains a vehicle state subset for the action period based on the vehicle positioning data. The action positioning sequence generation module determines the navigation coordinate system offset vector based on the vehicle state subset during the action period; it synthesizes the absolute position of the vehicle with the navigation coordinate system offset to obtain the absolute coordinates of the actuator's working end at the current moment, and obtains the action positioning time sequence based on the absolute coordinates of the working end. The association construction and directed graph generation module determines the first association and the second association based on the execution action location time sequence; and constructs a directed graph based on the first association and the second association. The action cluster analysis and feature extraction module obtains a set of continuous action clusters based on the directed graph, and determines the first sequence traceability rate, the second sequence dependency, and the BeiDou positioning integrity index based on the set of continuous action clusters. The data fusion strategy decision module determines the vehicle precise positioning data fusion processing method based on the first sequence traceability rate, the second sequence dependency degree, and the Beidou positioning integrity index.

2. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The data acquisition and state subset extraction module acquires vehicle positioning data and execution action positioning data, including: The vehicle positioning data refers to the measurement information of the vehicle platform's own position and attitude in the global coordinate system; The execution action positioning data refers to the calculated location information of the working end of the vehicle's actuator in the actual geographical space; The vehicle positioning data is reorganized into a first vehicle positioning time sequence according to a fixed update frequency; The vehicle positioning data is reorganized into a second vehicle positioning time sequence according to the actual update frequency.

3. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The step of obtaining a subset of vehicle states during the action period based on the vehicle positioning data includes: Based on the start and end timestamps of the actions of the actuator's working end, each data point of the actuator's working end from the start to the end of the action is extracted from the vehicle positioning data to form a subset of the vehicle status during the action period.

4. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The action positioning sequence generation module determines the navigation coordinate system offset vector based on the vehicle state subset during the action period, including: For each data point in the vehicle state subset during the aforementioned action period, perform the following calculations: Obtain the platform roll angle, platform pitch angle, and platform yaw angle of the data points; A three-dimensional rotation matrix is ​​constructed based on the platform roll angle, platform pitch angle, and platform yaw angle. Multiply the fixed actuator working end lever parameters by the three-dimensional rotation matrix to calculate the navigation coordinate system offset vector in the ENU coordinate system; The offset vector of the navigation coordinate system is obtained based on the spatial coordinate transformation formula of the actuator's working end. The spatial coordinate transformation formula of the actuator's working end is: ； in, , and For the lever arm parameters at the working end of the actuator, For the platform's roll angle, The platform's pitch angle, For the platform's heading angle, It is a three-dimensional rotation matrix. , and This is the offset vector of the navigation coordinate system.

5. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The process of combining the vehicle's absolute position with the navigation coordinate system offset to obtain the absolute coordinates of the actuator's working end at the current moment, and obtaining the execution action positioning time sequence based on the absolute coordinates of the working end, includes: the synthesis formula is: ; ; ; in, , and For the absolute coordinates of the vehicle platform, , This is the parameter for the Earth's radius of curvature. , and The coordinates are the absolute coordinates of the working end; The absolute coordinates of the working end are reorganized in chronological order to form a timing sequence of the execution action positioning. The execution action positioning timing sequence refers to the data sequence of the continuous motion trajectory of the working end of the actuator in the global coordinate system during the execution of the action.

6. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 2, characterized in that, The association construction and directed graph generation module determines the first and second association relationships based on the execution action location time sequence, including: A tolerance threshold is set. For each data point in the execution action positioning time sequence, a first difference is calculated between the timestamp of the data point and the timestamp corresponding to the first vehicle positioning time sequence. If the first difference is less than the tolerance threshold, the data point is associated with the first vehicle positioning time sequence to obtain a first association relationship. If the first difference is greater than or equal to the tolerance threshold, then the second difference between the timestamp of the data point and the timestamp corresponding to the second vehicle positioning time sequence is calculated. If the second difference is less than the tolerance threshold, then the data point is associated with the second vehicle positioning time sequence to obtain a second association relationship.

7. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The construction of the directed graph based on the first and second association relationships includes: Each data point in the first vehicle positioning time sequence is taken as the first sequence node; Each data point in the second vehicle positioning time sequence is used as a second sequence node; Each data point in the action location time sequence is used as an action sequence node; The first sequence node and the second sequence node are collectively referred to as the source sequence node. For each pair of action sequence nodes and source sequence nodes that are associated through the first association relationship or the second association relationship, a directed edge is created from the action sequence node to the source sequence node, and a directed graph is obtained based on the directed edge.

8. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 7, characterized in that, The action cluster analysis and feature extraction module obtains a set of continuous action clusters based on the directed graph, including: Set a maximum time interval threshold, extract all action sequence nodes from the directed graph, and sort and traverse them based on the timestamps of the action sequence nodes; merge action sequence nodes with consecutive timestamps and adjacent intervals less than the maximum time interval threshold into the same cluster to obtain a set of continuous action clusters.

9. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The determination of the first sequence tracing rate, the second sequence dependency, and the BeiDou positioning integrity index based on the set of continuous action clusters includes: For each continuous action cluster in the set of continuous action clusters, calculate the first sequence traceability rate and the second sequence dependency. The first sequence traceability rate is used to measure the degree to which the location data of a continuous action cluster originates from an ideal location sequence; The second sequence dependency is used to measure the degree to which the localization data of a continuous action cluster depends on the actual localization sequence; The formula for calculating the first sequence traceability rate is: ;in, This represents the number of action points within a continuous action cluster that are associated with the first sequence node. The total number of action points within a continuous action cluster. The first sequence traceability rate; The formula for calculating the second sequence dependency is: ;in, This represents the second sequence dependency. For each action point within a continuous action cluster, the location source node is found through the first association relationship and the second association relationship; Extract the carrier-to-noise ratio, number of satellites, and accuracy factor from the BeiDou satellite navigation and positioning data corresponding to the positioning source node; The average values ​​of the carrier-to-noise ratio, number of satellites, and accuracy factor for all action points within a continuous action cluster are calculated and normalized to obtain the BeiDou positioning integrity index.

10. The vehicle precise positioning data fusion processing system based on BeiDou satellite navigation communication as described in claim 1, characterized in that, The data fusion strategy decision module determines the vehicle precise positioning data fusion processing method based on the first sequence traceability rate, the second sequence dependency, and the BeiDou positioning integrity index, including: Weighting coefficients are assigned to BeiDou satellite navigation and positioning data and attitude data based on a weighted least squares estimation model. The optimal estimate of the system state is obtained by solving for the system state that minimizes the sum of squared weighted errors. The allocation of the weighting coefficients is dynamically determined based on the traceability consistency coefficient, the BeiDou positioning integrity index, and the first sequence traceability rate. The dynamic determination includes: For the current continuous action cluster, the source consistency coefficient is calculated based on the first and second association relationships of each action sequence node within the continuous action cluster; The source consistency coefficient is used to quantify the stability of the location data source within a continuous action cluster; the formula for calculating the source consistency coefficient is: ； in, Let be the traceability consistency coefficient, when or When the logarithm is 0, the corresponding logarithmic term is defined as 0; If the traceability consistency coefficient is greater than or equal to the preset traceability consistency threshold, then the continuous action cluster is determined to be a high inconsistency cluster. If it is a highly inconsistent cluster, increase the weight of attitude data and decrease the weight of BeiDou satellite navigation and positioning data; If the traceability consistency coefficient is less than the preset traceability consistency threshold, the continuous action cluster is determined to be a low inconsistency cluster. If it is a low-inconsistency cluster, when the BeiDou positioning integrity index is less than the BeiDou positioning integrity index threshold, the weight of BeiDou satellite navigation and positioning data is reduced and the weight of attitude data is increased; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is greater than or equal to the first sequence traceability rate threshold, BeiDou satellite navigation and positioning data is used as the core observation source; when the BeiDou positioning integrity index is greater than or equal to the BeiDou positioning integrity index threshold and the first sequence traceability rate is less than the first sequence traceability rate threshold, a high-frequency attitude priority fusion strategy is executed; the high-frequency attitude priority fusion strategy is a fusion mode that takes high-frequency attitude data as the main factor, constructs a smooth trajectory benchmark through its integration results, and then uses BeiDou satellite navigation and positioning data for global correction.

Images