Beidou signal error compensation positioning correction method and system in complex urban environment

Abstract

The invention relates to the technical field of navigation and positioning, and discloses a Beidou signal error compensation positioning correction method and system in a complex urban environment, and the method comprises the steps: synchronously collecting Beidou satellite original observation data, inertial navigation data and environmental point cloud data through a multi-source sensor array; performing real-time signal quality evaluation on the Beidou original observation data and extracting a composite error feature vector; inputting the composite error feature vector into a pre-trained adaptive deep neural network model, and outputting a positioning error compensation amount; and in a final positioning calculation link, systematic correction is carried out on an original pseudo-range and carrier phase observation value through an extended Kalman filtering algorithm, and high-precision position coordinates are calculated. The invention aims to solve the problems of insufficient adaptability, multi-source error comprehensive compensation capability and real-time performance in a complex urban environment in the prior art.

Description

Technical Field

[0001] This invention belongs to the field of navigation and positioning technology, specifically relating to a method and system for BeiDou signal error compensation and positioning correction in complex urban environments. Background Technology

[0002] With the increasing application of the BeiDou Navigation Satellite System in areas such as intelligent urban transportation, autonomous driving, and location services, the complex urban environment places higher demands on high-precision positioning. Dense high-rise buildings, narrow streets, and dynamic obstructions in urban environments cause BeiDou signals to frequently encounter problems such as blockage, multipath propagation, and non-line-of-sight propagation, significantly reducing the accuracy and stability of positioning results. Existing positioning technologies typically rely on single error source modeling or local environment adaptation strategies, which are insufficient to comprehensively address the dynamic nature and spatial heterogeneity of multi-source coupled errors in urban environments.

[0003] Among these, BeiDou signal error compensation and positioning correction technology in complex urban environments has become a key direction for improving navigation performance. This type of technology aims to estimate and compensate for comprehensive errors caused by multipath effects, signal attenuation, atmospheric delay, and building reflections in real time by integrating environmental perception, signal feature extraction, and intelligent modeling, thereby restoring or enhancing the availability and reliability of the positioning system.

[0004] In existing technologies, some solutions combine inertial navigation assistance with neural network prediction to correct position when signals are missing. However, these solutions lack sufficient modeling of typical urban error sources such as multipath propagation and occlusion, limiting the model's generalization ability. Other solutions focus on precise correction of spatial segment errors such as orbital clock errors, but neglect local signal distortion caused by building clusters in the near-ground propagation environment of cities, making it difficult to achieve end-to-end comprehensive error suppression. Furthermore, existing methods generally rely on historical data or static environmental assumptions, exhibiting poor real-time adaptability in dynamically changing urban scenarios and failing to meet the dual requirements of low latency and high robustness for high-precision positioning. These shortcomings indicate an urgent need for a BeiDou positioning enhancement method that can deeply integrate multi-source error characteristics, possess strong environmental adaptability, and support real-time dynamic correction to overcome the technical bottlenecks of high-precision navigation in complex urban environments. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for BeiDou signal error compensation and positioning correction in complex urban environments, which can effectively solve the problems in the background art. Existing BeiDou signal error compensation and positioning correction methods still have certain shortcomings in terms of adaptability in complex urban environments, multi-source error comprehensive compensation capability, and real-time performance.

[0006] Therefore, this invention provides a BeiDou signal error compensation positioning correction method and system in complex urban environments, aiming to comprehensively consider the effects of multipath effects, signal blockage and other error sources, optimize the generalization ability and real-time performance of the error compensation model, thereby meeting the demand for high-precision positioning in complex urban environments.

[0007] To achieve the above objectives, the present invention provides the following technical solution: On one hand, a BeiDou signal error compensation and positioning correction system for complex urban environments, comprising the following components: a multi-source heterogeneous data acquisition module, used to receive in real time raw observation data from BeiDou satellites, pose data from an inertial measurement unit, and information on the outlines of surrounding buildings and dynamic obstacles collected by environmental perception sensors; and a signal quality assessment and error feature extraction module, connected to the multi-source heterogeneous data acquisition module, used to perform signal-to-noise ratio, carrier phase continuity, and multipath error envelope analysis on the raw observation data, and extract composite errors characterizing multipath effects, signal obstruction, and atmospheric delay based on the analysis results. The system includes a feature vector and an adaptive error compensation model construction module, connected to the signal quality assessment and error feature extraction module. The module receives the composite error feature vector and uses a deep neural network to construct a nonlinear mapping function that dynamically adjusts its internal weight parameters. This function maps the composite error feature vector to the corresponding positioning error compensation amount. A real-time positioning calculation and correction module, connected to the adaptive error compensation model construction module, receives the positioning error compensation amount and applies it to the original pseudorange or carrier phase observations. Position calculation is performed using weighted least squares or extended Kalman filtering algorithms, and the result is output as a high-precision positioning result after error compensation.

[0008] Preferably, the multi-source heterogeneous data acquisition module specifically includes a BeiDou radio frequency front-end, a microelectromechanical system (MEMS) inertial measurement unit, and a lidar or millimeter-wave radar. The BeiDou radio frequency front-end is responsible for down-conversion and analog-to-digital conversion, outputting a raw observation data stream containing pseudorange, carrier phase, and Doppler frequency shift; the MEMS inertial measurement unit outputs triaxial acceleration and angular velocity data at a sampling frequency of 100Hz; the lidar or millimeter-wave radar scans the surrounding environment at a frequency of 10Hz, generating point cloud data for building contour extraction and dynamic obstacle tracking.

[0009] Furthermore, the signal-to-noise ratio analysis performed by the signal quality assessment and error feature extraction module adopts the sliding window variance calculation method, with the window length set to 2 seconds; the carrier phase continuity analysis identifies cycle slip events by detecting first-order differential jumps in the carrier phase time series that exceed 0.5 cycles; and the multipath error envelope analysis constructs a statistical upper bound model for multipath error based on the power ratio of the reflected signal and the direct signal, as well as the time delay difference.

[0010] Furthermore, the adaptive error compensation model construction module employs a hybrid architecture of a long short-term memory (LSTM) network and an attention mechanism in its deep neural network. The LSTM network is used to capture the long-term dependencies of error features over time, and it has 128 hidden layer units. The attention mechanism is deployed above the output layer of the LSTM network to adaptively weight the contribution of features from different time steps and different error sources to the current positioning error compensation amount. Its attention weight calculation function is as follows: in, This represents the hidden state of the Long Short-Term Memory network at time step t. For context vectors, For the scoring function, an additive model is used; Here, T represents the attention weights, and T represents the sequence length.

[0011] Furthermore, the real-time positioning calculation and correction module employs an extended Kalman filter algorithm whose state vector includes position, velocity, attitude, clock error, and clock drift. This module introduces the positioning error compensation output by the adaptive error compensation model construction module as a systematic bias correction term in the observation equation, specifically manifested as a correction to the original pseudorange observations: in, These are the original pseudorange observations. This is the amount of compensation for the positioning error predicted by the model. These are the corrected pseudorange observations.

[0012] On the other hand, a method for compensating for BeiDou signal errors and correcting positioning in complex urban environments is proposed. The specific steps of this method are as follows: Step S110: Simultaneously collect raw observation data from BeiDou satellites, inertial navigation data, and high-resolution environmental point cloud data using a multi-source sensor array deployed on a mobile vehicle; Step S120: Real-time signal quality assessment is performed on the collected BeiDou raw observation data, the signal-to-noise ratio is calculated, the carrier phase cycle slip is detected and the multipath error envelope is analyzed, and then a composite error feature vector containing multipath, occlusion and atmospheric delay information is extracted. Step S130: Input the composite error feature vector into a pre-trained adaptive deep neural network model. This model is built based on a long short-term memory network and an attention mechanism, and is used for dynamic reasoning and outputting the positioning error compensation amount corresponding to the current environment and signal conditions. In step S140, in the final positioning calculation stage, the positioning error compensation amount obtained in step S130 is integrated into the observation equation. By executing the extended Kalman filter algorithm, the original pseudorange and carrier phase observation values ​​are systematically corrected, thereby calculating the high-precision position coordinates after error compensation.

[0013] Compared with the prior art, the present invention has the following beneficial effects: This invention achieves comprehensive modeling and compensation for multi-source coupling errors such as multipath, occlusion, and atmospheric delay in urban environments by deeply fusing multi-source heterogeneous data and constructing composite error feature vectors, significantly improving the comprehensiveness and accuracy of error estimation.

[0014] This invention employs an adaptive deep learning model based on long short-term memory networks and attention mechanisms, which can effectively capture the dynamic temporal evolution of error characteristics and adaptively focus on key error sources, thereby enhancing the model's generalization ability and robustness in complex and ever-changing urban environments.

[0015] This invention tightly couples the intelligent error compensation model with the classic positioning algorithm, enabling simultaneous error compensation and position calculation. While ensuring high accuracy, it meets the stringent real-time requirements of positioning systems in complex urban application scenarios. The entire solution forms a closed loop from data acquisition, feature extraction, model inference to final calculation, systematically solving the core problems of existing technologies, such as reliance on a single error model, poor environmental adaptability, and insufficient real-time performance. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall technical architecture of the BeiDou signal error compensation and positioning correction method and system proposed in this invention for complex urban environments; Figure 2 This is a schematic diagram of the core principle framework of the adaptive error compensation model construction module in this invention; Figure 3 This is a logical flowchart of the signal quality assessment and error feature extraction module in this invention; Figure 4 This is a schematic diagram of the multi-level interaction relationship and data flow between the multi-source heterogeneous data acquisition module and the real-time positioning calculation and correction module in this invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the specific embodiments according to the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0018] Example 1 At a test site for intelligent connected vehicles in the central urban area, a multi-source heterogeneous data acquisition module carried out comprehensive data acquisition for the BeiDou positioning system in an urban canyon environment.

[0019] See Figure 1 This system includes a multi-source heterogeneous data acquisition module, a signal quality assessment and error feature extraction module, an adaptive error compensation model construction module, and a real-time positioning calculation and correction module. The BeiDou radio frequency front-end continuously receives navigation messages from the BeiDou-3 satellite system at a sampling frequency of 20Hz, outputting a raw data stream containing original pseudorange observations, carrier phase observations, and Doppler frequency shift measurements; the accuracy of the original pseudorange observations reaches 0.5 meters, and the accuracy of the carrier phase observations reaches 2 millimeters. The microelectromechanical system (MEMS) inertial measurement unit synchronously outputs triaxial acceleration and angular velocity data at a sampling frequency of 100Hz, with acceleration measurement ranges covering ±16g and angular velocity measurement ranges covering ±2000 degrees / second, ensuring continuous pose estimation even during brief satellite signal interruptions. The lidar performs a 3D scan of the surrounding environment at a frequency of 10Hz, generating point cloud data containing 128,000 points per frame, achieving a point cloud density of 200 points per square meter. Point cloud segmentation algorithms are used to extract building contour features and dynamic obstacle trajectories.

[0020] See Figure 3 The signal quality assessment and error feature extraction module performs in-depth analysis of the received raw observation data. This module first uses a sliding window variance calculation method to evaluate the signal-to-noise ratio (SNR). A 2-second sliding window is set, and the variance of the signal strength within the window is calculated. When the variance exceeds a preset threshold of 4 dB, signal quality is considered degraded. Carrier phase continuity analysis is achieved by monitoring the first-order difference change in the carrier phase time series. When a carrier phase jump between adjacent sampling points exceeds 0.5 cycles, it is identified as a cycle slip event and its timestamp is recorded. Multipath error envelope analysis is based on the power ratio and delay difference parameters of the direct and reflected signals, constructing a statistical upper bound model. The power ratio threshold is set to -6 dB, and the delay difference threshold is set to 1.5 chip periods. The composite error feature vector extracted through the above analysis contains 12 dimensions, representing the characteristic quantities of error sources such as multipath effect intensity, signal obstruction degree, ionospheric delay influence, and tropospheric delay influence.

[0021] See Figure 2The adaptive error compensation model construction module receives the composite error feature vector and performs nonlinear mapping calculations through a deep neural network. This module adopts a hybrid architecture of a long short-term memory network and an attention mechanism. The long short-term memory network contains 128 hidden layer units. The input layer receives a 12-dimensional composite error feature vector, and the output layer generates a 6-dimensional localization error compensation. The attention mechanism is deployed on the output layer of the long short-term memory network and calculates the attention weights of the features at each time step through an additive scoring function. The specific calculation formula is as follows: in, s represents the hidden state of the Long Short-Term Memory network at time step t, with a dimension of 128; s represents the context vector, with a dimension of 64. The scoring function employs an additive model to calculate the relevance score between the hidden state and the query vector. The weight matrix of the scoring function has a dimension of 192×64. Here, T represents the attention weights, and T is the sequence length. During the training phase, the model uses the mean squared error loss function and employs an adaptive moment estimation algorithm for parameter optimization. The initial learning rate is 0.001, and after 500 training iterations, the validation set loss converges to below 0.00015.

[0022] See Figure 4 The real-time positioning calculation and correction module integrates the positioning error compensation output from the adaptive error compensation model construction module into the extended Kalman filter algorithm. The state vector of this module contains 11 state variables, including three-dimensional position, three-dimensional velocity, quaternion attitude, and receiver clock error and clock drift. A systematic bias correction term is introduced into the observation equation. The specific formula for correcting the original pseudorange observation values ​​is as follows: in, These are the original pseudorange observations. This is the amount of compensation for the positioning error predicted by the model. The corrected pseudorange observations are used, and the dimensions of the positioning error compensation predicted by the model are consistent with the original pseudorange observations, in meters. The process noise covariance matrix of the extended Kalman filter is dynamically adjusted according to the data quality of the inertial measurement unit, while the observation noise covariance matrix is ​​adaptively configured based on the signal quality assessment results. The standard deviation of the observation noise is set to 0.5 meters when the signal quality is good, and increases to 3.0 meters when the signal quality is poor. The position calculation results are output at a frequency of 20Hz. The horizontal positioning accuracy reaches 0.3 meters in open environments and improves to 1.2 meters in urban canyon environments.

[0023] Example 2 In drone logistics delivery scenarios between urban underground parking lots and surface transition areas, multi-source heterogeneous data acquisition modules face more severe signal environment challenges. The BeiDou RF front-end is equipped with a high-sensitivity receiver, maintaining a locked state even when signal strength is below -150 dBW, outputting raw observation data including three frequency points: B1I, B2I, and B3I. The microelectromechanical system (MEMS) inertial measurement unit has been upgraded to tactical-level precision, with accelerometer zero-bias stability reaching 0.01 mg and gyroscope zero-bias stability reaching 0.5 degrees / hour, maintaining high-precision dead reckoning for 30 seconds in an indoor environment where satellite signals are completely interrupted. Millimeter-wave radar replaces lidar as the environmental perception sensor, operating at a 77 GHz frequency, with a maximum detection range of 200 meters and a range resolution of 0.1 meters, capable of penetrating rain and fog interference to accurately identify building edge features.

[0024] The signal quality assessment and error feature extraction module enhances cycle slip detection capabilities based on Example 1. It constructs a geometrically independent combination using combined observations of three-frequency carrier phases, and detects minute cycle slips by monitoring the time-series changes of these combined observations, improving detection sensitivity to 0.05 cycles. Multipath error envelope analysis introduces carrier phase multipath error estimation. Utilizing the correlation between pseudorange multipath error and carrier phase multipath error, a dual-frequency multipath error separation model is constructed, effectively distinguishing the influence of multipath signals from different reflectors. The composite error feature vector is expanded to 16 dimensions, adding parameters such as satellite elevation angle weighting coefficient, signal strength attenuation rate, and multipath signal spatial distribution characteristics.

[0025] The adaptive error compensation model building module introduces gated recurrent unit branches on top of a long short-term memory network, forming a dual-stream network architecture. The main branch processes error features related to BeiDou observation data, while the secondary branch processes auxiliary features related to environmental perception data. The two branches fuse features at the attention layer. Attention weights are calculated using a scaled dot product attention mechanism, with a query vector dimension of 64, a key vector dimension of 64, a value vector dimension of 128, and four attention heads. Model training employs a course-based learning strategy, starting with open environment data and gradually introducing complex environment data, ultimately achieving stable performance on a mixed dataset containing various scenarios such as urban canyons, under viaducts, and tree-lined roads.

[0026] The real-time positioning and correction module employs a compactly combined filtering architecture, fusing the original pseudorange and carrier phase observations with inertial navigation data. The extended Kalman filter's state vector is expanded to 21 dimensions, adding inertial sensor error states and carrier phase ambiguity parameters. During the observation update phase, the noise variance of the corrected pseudorange observations is dynamically adjusted based on signal quality assessment results, and the ambiguity parameters of the carrier phase observations are reset promptly when cycle slips occur. The positioning result output frequency is increased to 100Hz, and the horizontal positioning accuracy remains within 2.5 meters even in signal-blocked environments, while the elevation positioning accuracy reaches 1.8 meters, meeting the location service requirements for precise UAV landing.

[0027] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for BeiDou signal error compensation and positioning correction in complex urban environments, characterized in that, Includes the following steps: Step S110: Simultaneously collect raw observation data from BeiDou satellites, inertial navigation data, and high-resolution environmental point cloud data using a multi-source sensor array deployed on a mobile vehicle; Step S120: Real-time signal quality assessment is performed on the collected BeiDou raw observation data, the signal-to-noise ratio is calculated, the carrier phase cycle slip is detected and the multipath error envelope is analyzed, and then a composite error feature vector containing multipath, occlusion and atmospheric delay information is extracted. Step S130: Input the composite error feature vector into a pre-trained adaptive deep neural network model. This model is built based on a long short-term memory network and an attention mechanism, and is used for dynamic reasoning and outputting the positioning error compensation amount corresponding to the current environment and signal conditions. In step S140, in the final positioning calculation stage, the positioning error compensation amount obtained in step S130 is integrated into the observation equation. By executing the extended Kalman filter algorithm, the original pseudorange and carrier phase observation values ​​are systematically corrected, thereby calculating the high-precision position coordinates after error compensation.

2. The BeiDou signal error compensation and positioning correction method in complex urban environments according to claim 1, characterized in that, In step S120, the signal-to-noise ratio evaluation adopts the sliding window variance calculation method. A sliding window with a duration of 2 seconds is set, and the variance value of the signal strength within the window is calculated. When the variance value exceeds a preset threshold of 4 dB, the signal quality is determined to be degraded. Carrier phase continuity analysis is achieved by monitoring the first-order differential change of the carrier phase time series. When a carrier phase jump between adjacent sampling points is detected to exceed 0.5 cycles, it is identified as a cycle slip event and the occurrence timestamp is recorded. Multipath error envelope analysis is based on the power ratio and delay difference parameters of the direct signal and the reflected signal, and a statistical upper bound model is constructed. The power ratio threshold is set to -6 dB and the delay difference threshold is set to 1.5 chip cycles.

3. The BeiDou signal error compensation and positioning correction method in complex urban environments according to claim 1, characterized in that, In step S140, the extended Kalman filter state vector includes 11 state variables: three-dimensional position, three-dimensional velocity, quaternion attitude, receiver clock error, and clock drift. A systematic bias correction term is introduced into the observation equation, and the correction formula for the original pseudorange observations is specifically expressed as follows: in, These are the original pseudorange observations. This is the amount of compensation for the positioning error predicted by the model. The corrected pseudorange observations are represented by the same units as the original pseudorange observations, in meters. The process noise covariance matrix of the extended Kalman filter is dynamically adjusted according to the data quality of the inertial measurement unit, while the observation noise covariance matrix is ​​adaptively configured according to the signal quality assessment results. When the signal quality is good, the standard deviation of the observation noise is set to 0.5 meters, and when the signal quality is poor, it is increased to 3.0 meters.

4. A BeiDou signal error compensation and positioning correction system for complex urban environments, applied to the BeiDou signal error compensation and positioning correction method for complex urban environments as described in any one of claims 1-3, characterized in that, Includes the following modules: The multi-source heterogeneous data acquisition module is used to receive raw observation data from Beidou satellites, pose data from the inertial measurement unit, and information on the outlines of surrounding buildings and dynamic obstacles collected by environmental perception sensors in real time. The signal quality assessment and error feature extraction module is connected to the multi-source heterogeneous data acquisition module. It is used to perform signal-to-noise ratio, carrier phase continuity and multipath error envelope analysis on the raw observation data, and extract composite error feature vectors that characterize multipath effects, signal blockage and atmospheric delay based on the analysis results. An adaptive error compensation model construction module, connected to the signal quality assessment and error feature extraction module, is used to receive the composite error feature vector and construct a nonlinear mapping function that can dynamically adjust its internal weight parameters using a deep neural network. This function maps the composite error feature vector to the corresponding positioning error compensation amount. The real-time positioning calculation and correction module is connected to the adaptive error compensation model construction module. It receives the positioning error compensation amount and applies it to the original pseudorange or carrier phase observation value. It performs position calculation through weighted least squares or extended Kalman filter algorithm and outputs the high-precision positioning result after error compensation.

5. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The multi-source heterogeneous data acquisition module specifically includes a BeiDou radio frequency front-end, a microelectromechanical system (MEMS) inertial measurement unit, and a lidar or millimeter-wave radar. The BeiDou radio frequency front-end is responsible for down-conversion and analog-to-digital conversion, outputting a raw observation data stream containing pseudorange, carrier phase, and Doppler frequency shift. The MEMS inertial measurement unit outputs triaxial acceleration and angular velocity data at a sampling frequency of 100Hz. The lidar or millimeter-wave radar scans the surrounding environment at a frequency of 10Hz, generating point cloud data for building contour extraction and dynamic obstacle tracking.

6. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The signal-to-noise ratio analysis performed by the signal quality assessment and error feature extraction module adopts the sliding window variance calculation method, with the window length set to 2 seconds; Carrier phase continuity analysis identifies cycle slip events by detecting first-order differential jumps in the carrier phase time series that exceed 0.5 cycles; multipath error envelope analysis constructs a statistical upper bound model for multipath error based on the power ratio of the reflected signal to the direct signal and the time delay difference.

7. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The adaptive error compensation model construction module employs a hybrid architecture of a long short-term memory (LSTM) network and an attention mechanism in its deep neural network. The LSTM network is used to capture the long-term dependencies of error features over time, and it has 128 hidden layer units. The attention mechanism is deployed on top of the LSTM network's output layer to adaptively weight the contributions of features from different time steps and different error sources to the current positioning error compensation. Its attention weight calculation function is as follows: in, This represents the hidden state of the Long Short-Term Memory network at time step t. For context vectors, The scoring function is an additive model. Here, T represents the attention weights, and T represents the sequence length.

8. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The real-time positioning solution and correction module uses the extended Kalman filter algorithm, whose state vector includes position, velocity, attitude, clock error, and clock drift. This module introduces the positioning error compensation amount output by the adaptive error compensation model construction module as a systematic deviation correction term in the observation equation.

9. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The adaptive error compensation model construction module introduces gated recurrent unit branches on the basis of long short-term memory network to form a dual-stream network architecture; the main branch processes error features related to Beidou observation data, the branch processes auxiliary features related to environmental perception data, and the two branches perform feature fusion at the attention layer. Attention weights are calculated using a scaled dot product attention mechanism. The query vector dimension is 64, the key vector dimension is 64, the value vector dimension is 128, and the number of attention heads is set to 4.

10. The BeiDou signal error compensation and positioning correction system for complex urban environments according to claim 4, characterized in that, The real-time positioning calculation and correction module adopts a compact combination filter architecture, which fuses the original pseudorange observations and carrier phase observations with the inertial navigation data respectively; the state vector of the extended Kalman filter is expanded to 21 dimensions, and new inertial sensor error states and carrier phase ambiguity parameters are added; During the observation update phase, the noise variance of the corrected pseudorange observations is dynamically adjusted based on the signal quality assessment results, and the ambiguity parameters of the carrier phase observations are reset in a timely manner when cycle slips occur.

Images