Vehicle high-precision positioning method based on rtk post-difference for open-pit coal mine truck

Abstract

The application discloses a kind of open coal mine vehicle based on rtk post-difference vehicle high-precision positioning method, comprising: first query fusion multi-path heat map information digital twin scene map to obtain environmental feature label, according to this adaptive adjustment multi-source sensor fusion strategy, generate first positioning solution, then through 5G network and edge server cooperation, using network rtk enhancement correction number correction to obtain second high-precision positioning solution, finally calculate and output fusion multi-dimensional information real-time credibility score.The method effectively solves the problem of positioning failure under the complex environment of open pit mine, realizes continuous, accurate and reliable position perception, and provides reliable guarantee for unmanned operation.

Description

Technical Field

[0001] This invention relates to the field of mining vehicle positioning technology, and more specifically, to a high-precision positioning method for open-pit coal mine vehicles based on RTK post-differential positioning. Background Technology

[0002] In the intelligent and unmanned transformation and upgrading of open-pit coal mines, high-precision and high-reliability positioning of mining, transportation and other operating vehicles is the key foundation for achieving autonomous driving, intelligent scheduling and safe collaboration. Global Navigation Satellite Systems, especially their real-time dynamic differential technology, can provide centimeter-level real-time positioning accuracy and have become the core means of vehicle positioning in open-pit mines. In addition, post-differential processing technology can further improve the accuracy and reliability of positioning results through subsequent fine calculation of observation data, and is often used to make up for the shortcomings of real-time RTK in harsh environments.

[0003] However, the production environment of open-pit coal mines is extremely complex, with severe satellite signal blockage caused by steep slopes and deep pits, strong multipath reflection effects caused by large metal equipment and dynamically changing terrain, and the impact of harsh weather conditions such as dust, rain, snow, and extreme cold. These factors pose serious challenges to traditional single RTK positioning systems in such environments, including: frequent signal loss and poor positioning continuity; difficulty in modeling and eliminating multipath errors, resulting in large fluctuations in positioning accuracy; and the system's inability to know the reliability of positioning results, posing potential safety risks.

[0004] Although existing technologies have developed solutions that combine inertial navigation, lidar, and other multi-sensor fusion to improve robustness, they still generally suffer from problems such as insufficient environmental adaptability, fixed and rigid fusion strategies, and a lack of real-time quantitative evaluation of the quality of positioning results. These issues make it difficult to meet the requirements of unmanned operations in open-pit mines that are available around the clock, throughout the entire process, and with a high level of safety.

[0005] Therefore, there is an urgent need for a high-precision positioning method and system that can intelligently sense environmental interference, adaptively fuse multi-source information, and evaluate positioning reliability online, in order to overcome the technical bottleneck of continuous, accurate, and reliable positioning of vehicles in complex open-pit mine environments and provide core technical support for the large-scale safe application of unmanned mining trucks. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art and solve the technical problems of easy failure of positioning and difficulty in ensuring continuous, accurate and reliable position perception in the complex environment of open-pit mines, the present invention provides a high-precision vehicle positioning method based on RTK post-differential for open-pit coal mine vehicles, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a high-precision vehicle positioning method for open-pit coal mine cars based on RTK post-differential positioning, comprising the following steps: S1. Through the vehicle-mounted GNSS receiving module, inertial measurement unit, and lidar, the vehicle's raw satellite observation data, inertial data, and environmental point cloud data are collected simultaneously. S2. Based on the approximate location of the vehicle currently obtained by GNSS single-point positioning or INS estimation, query the pre-generated digital twin scene map of the open-pit mine to obtain the environmental feature labels of the current area. The environmental feature labels include multipath effect level, signal obstruction level and road structure complexity. S3. Based on environmental feature labels, the fusion weights and fusion models of GNSS, INS and lidar point cloud matching are dynamically adjusted through an adaptive decision engine to generate the first real-time positioning solution. The adaptive decision engine has a built-in preset environment-policy mapping table, which defines the correspondence between different combinations of environmental feature labels and the corresponding fusion model types and initial weights. S4. Upload the original satellite observation data to the edge computing server in real time through the 5G communication network, receive the network rtk enhancement correction number generated in real time based on the observation data of multiple base stations from the edge computing server, use the received network rtk enhancement correction number to correct the original satellite observation data, and re-perform GNSS / INS tightly coupled solution to obtain the second high-precision positioning solution. S5. Based on the consistency of multi-source data, the matching degree of environmental feature labels, and the confidence of the enhanced correction number, calculate the real-time reliability score of the second high-precision positioning solution and output it.

[0008] Preferably, the method for generating the digital twin scene map of the open-pit mine includes: S2.1 Obtain the benchmark three-dimensional terrain model and high-precision road network vector data of the open-pit mine area, and construct the geometric benchmark layer; S2.2. On the geometric reference layer, historical GNSS observation data are integrated, and a multipath effect probability heatmap representing the intensity of signal reflection interference is generated through statistical analysis, which serves as the environmental semantic layer. S2.3 On the geometric reference layer, mark the fixed signal obstruction area, the typical activity area of ​​dynamic operating equipment, and the area affected by extreme weather to form a scene label layer; S2.4. Spatiotemporally align and fuse the geometric reference layer, environmental semantic layer, and scene label layer to generate a digital twin scene map.

[0009] Preferably, step S3 specifically includes: S3.1 When the environmental feature label indicates an open and stable area, a fusion model with GNSS / INS tight coupling as the main component and lidar point cloud matching as the auxiliary component is adopted, and the highest weight is assigned to the GNSS observation value. S3.2 When the environmental feature label indicates a high incidence of multipath or a signal obstruction area, switch to a fusion model that primarily uses tightly coupled INS estimation and lidar point cloud matching, supplemented by GNSS, and reduce the weight of GNSS observations. S3.3 When the environmental feature label indicates a complex curved area, increase the weight of the lidar point cloud matching and enable the precise pose optimization algorithm based on point cloud features.

[0010] Preferably, in step S4, the method for the edge computing server to generate network RTK enhancement corrections includes: S4.1 Receive raw observation data from multiple GNSS reference stations deployed in the mining area; S4.2 Construct a regional error model on the server side and use carrier phase smoothing pseudorange and ambiguity fixing algorithms for real-time network solution; S4.3. Generate an enhanced correction data stream that includes regional atmospheric delay correction and orbital error correction. The edge computing server combines the multipath effect heat map information in the digital twin scene map to generate multipath suppression auxiliary information, which is then sent out along with the enhanced correction data stream. The vehicle terminal uses the multipath suppression auxiliary information to downweight or remove satellite observations that are severely affected by multipath in the carrier phase smoothing pseudorange processing.

[0011] Preferably, in step S5, the method for calculating the real-time credibility score is as follows: Score = α×C + β×M + γ×T Where C is the consistency coefficient between GNSS / INS / LiDAR positioning results, M is the matching degree between the current point cloud and the digital twin map, and T is the confidence level of the enhanced correction data sent from the cloud. α, β, and γ are weighting coefficients that are dynamically adjusted based on environmental feature labels, and α + β + γ = 1. The dynamic adjustment is achieved by querying a preset weight configuration table, which defines the specific values ​​of α, β, and γ corresponding to different combinations of environmental feature labels. Specifically, in the open and stable region, α ≥ 0.5; in the signal obstruction region, β ≥ 0.6; and in the region where the confidence level of the enhancement correction is higher than 0.9, γ ≥ 0.4.

[0012] The present invention also provides a high-precision positioning system for open-pit mine vehicles to implement the above method, comprising an on-board terminal subsystem and a cloud-based collaborative subsystem: The vehicle-mounted terminal subsystem includes: The data acquisition module is used to simultaneously acquire raw GNSS observation data, IMU inertial data, and lidar point cloud data; The local fusion processing module has a built-in adaptive decision engine, which dynamically adjusts the fusion weights and fusion model of GNSS, INS and lidar point cloud matching based on environmental feature labels to generate the first real-time positioning solution. The 5G communication module is used for data interaction with the cloud; The cloud-based collaborative subsystem includes: The digital twin map service module is used to store and provide query services for digital twin scene maps; Edge computing servers are deployed at the edge of the 5G network in the mining area to receive and process raw satellite observation data, and generate and distribute network RTK enhancement correction data. The central monitoring platform is used to receive the location results and reliability scores reported by each vehicle, and to display and issue warnings in a centralized manner.

[0013] Preferably, the local fusion processing module is implemented using a programmable system-on-a-chip (SoC) and is used to process two GNSS data streams, one IMU data stream, and one lidar point cloud data stream in parallel.

[0014] Preferably, the system further includes an auxiliary positioning module installed on the electric shovel bucket and the mine car box. The auxiliary positioning module includes a UWB ultra-wideband positioning unit, which is used to communicate with the vehicle terminal subsystem during automatic alignment of the shovel and the vehicle to provide centimeter-level relative positioning data to assist the absolute positioning system in completing precise loading.

[0015] Preferably, the central monitoring platform is equipped with a positioning reliability early warning unit. When the real-time reliability score of a vehicle's positioning solution is lower than a preset threshold, the unit automatically sends a deceleration or stopping command to the vehicle and issues an alarm to the remote dispatcher.

[0016] The present invention also provides an unmanned operation system for open-pit mines, including an unmanned mining dump truck, a remotely controlled electric shovel, and the aforementioned high-precision positioning system. The positioning system provides continuous and high-precision position and orientation information for the unmanned mining dump truck and provides positioning assurance for the automatic collaborative loading operation of the electric shovel and the mining truck.

[0017] The technical effects and advantages of this invention are as follows: This invention significantly improves the continuity and robustness of positioning in complex environments. By providing feedforward environmental cognition through digital twin scene maps, the system can proactively adjust multi-source fusion strategies based on environmental feature labels, effectively overcoming the problem of frequent failures of traditional RTK positioning caused by signal obstruction and multipath effects in open-pit mines. This invention achieves a balance between high precision and high reliability. By combining tight-coupled multi-sensor calculation on the vehicle end with enhanced correction via cloud network RTK, it ensures centimeter-level accuracy in positioning results while enhancing the overall robustness and reliability of the system through a cloud-end collaborative architecture. This invention innovatively provides quantifiable positioning safety redundancy. By introducing and calculating a credibility score that integrates multi-dimensional information in real time, it constructs a complete closed loop from perception and positioning to assessment and early warning, providing key system-level state monitoring and active safety barriers for autonomous driving systems. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the overall method of the present invention; Figure 2 This is a detailed flowchart of step S2 of the present invention; Figure 3 This is a detailed flowchart of step S3 of the present invention; Figure 4 This is a detailed flowchart of step S4 of the present invention; Figure 5 This is a system module connection diagram of the present invention.

[0019] The attached diagram is labeled as follows: 100, Vehicle-mounted terminal subsystem; 200, Cloud-based collaborative subsystem; 300, Auxiliary positioning module; 101, Data acquisition module; 102, Local fusion processing module; 103, 5G communication module; 201, Digital twin map service module; 202, Edge computing server; 203, Central monitoring platform; 301, UWB ultra-wideband positioning unit; 2030, Positioning reliability early warning unit. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0021] This invention provides a high-precision vehicle positioning method for open-pit coal mine cars based on RTK post-differential positioning. In its implementation, this method requires the use of the methods described in the appendix. Figure 5 The high-precision positioning system for open-pit mine vehicles shown includes an on-board terminal subsystem 100, a cloud-based collaborative subsystem 200, and an auxiliary positioning module 300. The three components interact with each other through the mine's 5G communication network. The vehicle-mounted terminal subsystem 100 is installed on a mining dump truck and is responsible for multi-source data acquisition, local fusion positioning calculation and communication. The cloud-based collaborative subsystem 200 is deployed at the edge of the mining area network and in the central computer room, and is responsible for high-precision map services, enhanced correction calculation and global monitoring. The auxiliary positioning module 300 is installed at specific locations on the electric shovel bucket and the mine car bucket for precise relative positioning at close range.

[0022] As attached Figure 1-4 As shown, the method of the present invention specifically includes the following steps: S1. Multi-source data synchronous acquisition In this step, the vehicle-mounted terminal subsystem 100 is responsible for synchronously collecting multi-source heterogeneous sensor data. Specifically: GNSS receiver module: It adopts a dual-antenna configuration to receive raw observation data from satellite systems such as BeiDou and GPS in real time, including but not limited to carrier phase observations, pseudorange observations and satellite ephemeris at L1 / L2 frequency points; Inertial Measurement Unit: Employs a tactical or industrial-grade IMU to output vehicle inertial data at high frequencies (e.g., 100Hz), including triaxial angular velocity and triaxial acceleration. LiDAR: Employs automotive-grade mechanical rotating or solid-state LiDAR to scan the surrounding environment in real time and generate environmental point cloud data; To ensure consistent time reference for data fusion, all the aforementioned sensors are connected to a programmable on-chip system, and the system's hardware triggering or software timestamp mechanism ensures strict synchronization between GNSS observation time, IMU sampling time, and lidar frame time.

[0023] S2. Environmental feature query based on digital twin scene map, details are attached. Figure 2 As shown, This step aims to provide predictive environmental awareness for the positioning system, and its implementation relies on a pre-generated digital twin scene map of the open-pit mine. The method for constructing this map includes: S2.1 Constructing the geometric reference layer: Obtaining the reference three-dimensional terrain model generated by airborne lidar measurement of the mining area and the road network vector data obtained by high-precision mapping to form the geometric basis of the map; S2.2 Generation of Environmental Semantic Layer: On the geometric reference layer, historical GNSS observation data (such as CORS data from continuously operating reference stations or a large amount of vehicle-mounted data) are fused, and statistical analysis methods such as kernel density estimation are used (or alternatively, historical data regression analysis or machine learning classification methods can be used) to calculate and generate a probability heat map of multipath effects in the entire mining area, and to quantitatively characterize the intensity level of signal reflection interference at each location. S2.3 Forming a scene label layer: On the geometric reference layer, based on the operating radius of dynamic equipment such as mining trucks and electric shovels, mark the typical activity area of ​​dynamic operating equipment; based on fixed features such as slopes and factory buildings, mark the fixed signal blocking area; and based on historical meteorological data, mark the area affected by extreme weather. S2.4. The geometric reference layer, environmental semantic layer and scene label layer mentioned above are spatiotemporally aligned and fused to form a unified digital twin scene map, which is then deployed in the cloud as a digital twin map service module 201.

[0024] During implementation, the vehicle terminal initiates a query request to the cloud via the 5G network based on the approximate location calculated by GNSS single-point positioning or INS in step S1. The cloud map service returns the environmental feature label corresponding to the area where the location is located. This label is a structured data that includes at least: multipath effect level (e.g., high, medium, low), signal obstruction level (e.g., severe, medium, none), and road structure complexity (e.g., simple straight road, complex curve, loading and unloading platform).

[0025] S3, Adaptive multi-source fusion localization solution, details are attached. Figure 3 As shown, This step is the core processing stage of localization, executed by the adaptive decision engine within the vehicle terminal. The tight coupling involved in this invention refers to an architecture that directly fuses raw observation data or low-level observation constraints from different sensors within a state estimation filter (such as an extended Kalman filter), rather than simply weighting the independent localization results. This engine has a built-in preset environment-policy mapping table. Based on the environmental feature labels obtained in step S2, it queries this mapping table to dynamically select and adjust the fusion strategy. S3.1 When the label indicates an open and stable area: The mapping table output command adopts a fusion model based on tight coupling of GNSS / INS. Specifically, the raw pseudorange of GNSS, carrier phase observations and IMU data are input together into the extended Kalman filter for tight coupling calculation to directly estimate the position, velocity and attitude. In this mode, lidar point cloud matching is used as an auxiliary verification method with a low weight and is mainly used for anti-skid assistance on non-smooth road surfaces. S3.2 When the tag indicates a high-incidence area of ​​multipath or a signal obstruction area: the mapping table output command switches to the mode of INS / LiDAR tight coupling. In this mode, the weight of GNSS observations is significantly reduced (e.g., by increasing their process noise covariance in the filter), or used only for velocity assistance. The core algorithm uses IMU for dead reckoning and simultaneously performs rapid matching between the real-time LiDAR point cloud and the geometric reference layer of the digital twin map in step S2 (e.g., using the iterative nearest point algorithm or its variant). The matching results are used to continuously correct the accumulated error of INS, forming tight coupling to ensure the continuity of positioning when the satellite signal quality is poor. At this point, the system has completed the preliminary positioning solution based on feedforward environmental cognition. S3.3 When the label indicates a complex curved area: The mapping table output instruction increases the weight of the lidar point cloud matching. In addition to basic matching, it enables a precise pose optimization algorithm based on point cloud features, such as extracting stable features like lane lines, retaining walls, and poles, and performing feature-based matching and local optimization to obtain a more accurate lateral position and heading angle in the curve.

[0026] Through the above adaptive fusion, a stable and reliable first real-time localization solution is finally output.

[0027] S4, differential enhancement correction after cloud collaboration, details are attached. Figure 4 As shown, This step aims to leverage cloud computing power to further improve positioning accuracy and reliability, and its implementation involves vehicle-cloud collaboration: S4.1 The vehicle-mounted terminal uploads the raw satellite observation data collected in step S1 to the edge computing server 202 deployed at the edge of the mining area in real time through the 5G communication module 103. S4.2 Edge computing server 202 collects observation data from multiple GNSS reference stations in the mining area and builds a fine regional error model (including ionospheric and tropospheric delay models) on the server side. Then, it uses network RTK algorithms such as carrier phase smoothing pseudorange and non-difference ambiguity fixing to perform real-time calculation and generate a set of high-precision network RTK enhancement corrections. This correction data stream not only includes conventional atmospheric and orbital corrections, but also integrates multipath suppression auxiliary information derived from the digital twin map environment semantic layer (multipath heat map). This information is used to identify satellites or observations that are severely affected by multipath. S4.3 The enhanced correction data is broadcast in real time to the vehicle terminal via the 5G network (end-to-end latency is usually less than 100 milliseconds); S4.4 After receiving the enhanced correction data, the vehicle-mounted terminal applies it to correct the original GNSS observation data, and uses the corrected observation data to re-perform the GNSS / INS tightly coupled solution, thereby obtaining a second high-precision positioning solution with higher accuracy and stronger robustness. Theoretically, it can achieve centimeter-level positioning accuracy, which reflects the global optimization capability of cloud-edge collaboration.

[0028] S5. Location Reliability Assessment and Output This step quantifies the quality of the positioning results, which is crucial for achieving safety redundancy. The specific implementation is as follows: The real-time credibility score is calculated using the following formula: Score = α×C + β×M + γ×T in: C is the consistency coefficient: it reflects the consistency between the output results of the three positioning sources, namely GNSS, INS and lidar. It can be obtained by calculating the Mahalanobis distance between the positions calculated by any two sources and combining the relationship between the three (alternatively, the weighted Euclidean distance or the method of calculating the overlapping area of ​​the covariance ellipse of the solution results can also be used). The higher the consistency, the closer the C value is to 1. M represents the point cloud matching degree: reflecting the degree of matching between the real-time laser point cloud and the geometric reference layer of the digital twin map in step S3. It can be quantified by the fitting residual or the proportion of interior points of the matching algorithm (such as ICP). T represents the confidence level of the cloud correction: generated by the edge computing server 202 based on information such as the residual error of the baseline during the solution process, the fixed solution ratio, and the integrity of the base station network, and distributed along with the correction. α, β, and γ are dynamic weighting coefficients: their values ​​are dynamically determined by querying a preset weight configuration table based on the environmental feature labels queried in step S2. The weight configuration table defines the suggested value ranges of α, β, and γ under different combinations of environmental feature labels. For example, in open and stable areas, α takes the largest value (e.g., 0.5) to emphasize GNSS consistency; in signal-obstructed areas, β takes the largest value (e.g., 0.6) to emphasize lidar matching quality; and in areas with extremely high cloud correction quality (T > 0.9), γ can be appropriately increased (e.g., 0.4). The sum of the three is always 1.

[0029] Finally, the system outputs a second high-precision positioning solution and its real-time reliability score. This score is transmitted to the cloud-based central monitoring platform 203 in real time. The score is not only used as an output, but its dynamic trend can also be fed back to the adaptive decision engine as reference information for fine-tuning the fusion strategy. When the score is lower than the preset safety threshold, the monitoring platform will automatically trigger an early warning mechanism, send a deceleration or stop command to the vehicle, and alert the dispatcher, thus forming a complete adaptive closed loop from perception and positioning to safety assessment.

[0030] The following example illustrates the overall workflow of the system using a driverless mining truck and an electric shovel in a coordinated loading operation: For example: When unmanned mining truck A drives towards the electric shovel operation area, the on-board terminal obtains the approximate location through GNSS single-point positioning, queries the digital twin map, and learns that it is about to enter the electric shovel operation signal blockage area. The adaptive decision engine reduces the GNSS weight in advance according to the mapping table rules and switches to the fusion mode based on INS / LiDAR to generate a stable and continuous first real-time positioning solution. Meanwhile, the vehicle-mounted 5G module uploads the original GNSS observation data and receives the enhanced correction data containing multipath suppression auxiliary information issued by the edge computing server 202. It corrects and re-solves the original observation data to obtain the second positioning solution with centimeter-level accuracy, and calculates the current confidence score as 0.88, marking it as good. When mining truck A enters the preset alignment area near the electric shovel, its on-board terminal communicates with the UWB ultra-wideband auxiliary positioning module 300 installed on the electric shovel bucket and the mining truck box. The UWB module provides centimeter-level precision relative position data (such as the distance and angle between the centers of the bucket and the box). The local fusion processing module 102 merges the precise relative positioning data provided by the UWB with its own high-precision absolute positioning data to generate the final composite positioning signal, guiding the mining truck to achieve precise automatic alignment with the electric shovel bucket. After alignment, the electric shovel control system (which can be remotely controlled) automatically loads the shovel based on the precise location information of the mining card. Throughout the process, the positioning data and reliability score are transmitted back to the central monitoring platform 203 in real time, enabling safe production that is globally visible, controllable, and has early warning capabilities.

[0031] In summary: Through the above implementation methods, this invention constructs a complete high-precision positioning closed loop from environmental perception, intelligent decision-making, cloud collaborative correction to credibility assessment. The steps are not simply connected in series, but rather achieve feedforward prediction through digital twin maps, intelligent decision-making through an adaptive engine, global optimization through cloud collaboration, and closed-loop feedback through credibility scoring. This deep collaboration solves the problems of accuracy, continuity, and reliability of vehicle positioning in the complex environment of open-pit coal mines, providing core technical support for the safe and efficient operation of unmanned mining trucks.

[0032] The above description is merely 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 high-precision vehicle positioning method for open-pit coal mine cars based on RTK post-differential positioning, characterized in that: Includes the following steps: S1. Through the vehicle-mounted GNSS receiving module, inertial measurement unit, and lidar, the vehicle's raw satellite observation data, inertial data, and environmental point cloud data are collected simultaneously. S2. Based on the approximate location of the vehicle currently obtained by GNSS single-point positioning or INS estimation, query the pre-generated digital twin scene map of the open-pit mine to obtain the environmental feature labels of the current area. The environmental feature labels include multipath effect level, signal obstruction level and road structure complexity. S3. Based on environmental feature labels, the fusion weights and fusion models of GNSS, INS and lidar point cloud matching are dynamically adjusted through an adaptive decision engine to generate the first real-time positioning solution. The adaptive decision engine has a built-in preset environment-policy mapping table. S4. Upload the original satellite observation data to the edge computing server (202) in real time through the 5G communication network, receive the network rtk enhancement correction number generated by real-time calculation based on the observation data of multiple base stations issued by the edge computing server (202), use the received network rtk enhancement correction number to correct the original satellite observation data, and re-perform GNSS / INS tightly coupled calculation to obtain the second high-precision positioning solution. S5. Based on the consistency of multi-source data, the matching degree of environmental feature labels, and the confidence of the enhanced correction number, calculate the real-time reliability score of the second high-precision positioning solution and output it.

2. The method according to claim 1, characterized in that: The method for generating the digital twin scene map of the open-pit mine includes: S2.1 Obtain the benchmark three-dimensional terrain model and high-precision road network vector data of the open-pit mine area, and construct the geometric benchmark layer; S2.

2. On the geometric reference layer, historical GNSS observation data are integrated, and a multipath effect probability heatmap representing the intensity of signal reflection interference is generated through statistical analysis, which serves as the environmental semantic layer. S2.3 On the geometric reference layer, mark the fixed signal obstruction area, the typical activity area of ​​dynamic operating equipment, and the area affected by extreme weather to form a scene label layer; S2.

4. Spatiotemporally align and fuse the geometric reference layer, environmental semantic layer, and scene label layer to generate a digital twin scene map.

3. The method according to claim 1, characterized in that: Step S3 specifically includes: S3.1 When the environmental feature label indicates an open and stable area, a fusion model with GNSS / INS tight coupling as the main component and lidar point cloud matching as the auxiliary component is adopted, and the highest weight is assigned to the GNSS observation value. S3.2 When the environmental feature label indicates a high incidence of multipath or a signal obstruction area, switch to a fusion model that primarily uses tightly coupled INS estimation and lidar point cloud matching, supplemented by GNSS, and reduce the weight of GNSS observations. S3.3 When the environmental feature label indicates a complex curved area, increase the weight of the lidar point cloud matching and enable the precise pose optimization algorithm based on point cloud features.

4. The method according to claim 1, characterized in that: In step S4, the method by which the edge computing server (202) generates network RTK enhancement corrections includes: S4.1 Receive raw observation data from multiple GNSS reference stations deployed in the mining area; S4.2 Construct a regional error model on the server side and use carrier phase smoothing pseudorange and ambiguity fixing algorithms for real-time network solution; S4.

3. Generate an enhanced correction data stream that includes regional atmospheric delay correction and orbital error correction. The edge computing server (202) combines the multipath effect heat map information in the digital twin scene map to generate multipath suppression auxiliary information, which is then sent out along with the enhanced correction data stream. The vehicle terminal uses the multipath suppression auxiliary information to downweight or remove satellite observations that are severely affected by multipath in the carrier phase smoothing pseudorange processing.

5. The method according to claim 1, characterized in that: In step S5, the method for calculating the real-time credibility score is as follows: Score = α×C + β×M + γ×T Where C is the consistency coefficient between GNSS / INS / LiDAR positioning results, M is the matching degree between the current point cloud and the digital twin map, and T is the confidence level of the enhanced correction data sent from the cloud. α, β, and γ are weighting coefficients that are dynamically adjusted based on environmental feature labels, and α + β + γ = 1. The dynamic adjustment is achieved by querying a preset weight configuration table, which defines the specific values ​​of α, β, and γ corresponding to different combinations of environmental feature labels. Specifically, in the open and stable region, α ≥ 0.5; in the signal obstruction region, β ≥ 0.6; and in the region where the confidence level of the enhancement correction is higher than 0.9, γ ≥ 0.

4.

6. A high-precision positioning system for open-pit mine vehicles for implementing the method according to any one of claims 1-5, characterized in that: Including the vehicle-mounted terminal subsystem (100) and the cloud-based collaborative subsystem (200): The vehicle-mounted terminal subsystem (100) includes: The data acquisition module (101) is used to simultaneously acquire raw GNSS observation data, IMU inertial data, and lidar point cloud data; The local fusion processing module (102) has a built-in adaptive decision engine, which is used to dynamically adjust the fusion weight and fusion model of GNSS, INS and lidar point cloud matching based on environmental feature labels, and generate the first real-time positioning solution. The 5G communication module (103) is used for data interaction with the cloud; The cloud-based collaborative subsystem (200) includes: The digital twin map service module (201) is used to store and provide query services for digital twin scene maps; An edge computing server (202) is deployed at the edge of the 5G network in the mining area to receive and process raw satellite observation data and generate and distribute network TRK enhancement correction numbers. The central monitoring platform (203) is used to receive the location results and credibility scores reported by each vehicle, and to display and issue warnings in a centralized manner.

7. The system according to claim 6, characterized in that: The local fusion processing module (102) is implemented using a programmable system-on-a-chip and is used to process two GNSS data streams, one IMU data stream, and one lidar point cloud data stream in parallel.

8. The system according to claim 6, characterized in that: The system also includes an auxiliary positioning module (300) installed on the electric shovel bucket and the mine car box. The auxiliary positioning module (300) includes a UWB ultra-wideband positioning unit (301) for communicating with the vehicle terminal subsystem (100) during automatic alignment of the shovel and the vehicle to provide centimeter-level relative positioning data to assist the absolute positioning system in completing accurate loading.

9. The system according to claim 6, characterized in that: The central monitoring platform (203) is equipped with a positioning reliability early warning unit (2030). When the real-time reliability score of the received vehicle positioning solution is lower than a preset threshold, it automatically sends a deceleration or stop command to the vehicle and issues an alarm to the remote dispatcher.

10. An unmanned driving operation system for open-pit mines, characterized in that: The invention includes unmanned mining dump trucks, remotely controlled electric shovels, and a high-precision positioning system as described in any one of claims 6-9. The positioning system provides continuous and high-precision position and orientation information for the unmanned mining dump trucks and provides positioning assurance for the automatic collaborative loading operation of the electric shovel and the mining truck.

Images