A GNSS pseudorange smoothing method, system and device considering multipath effect

Abstract

The application discloses a GNSS pseudorange smoothing method, system and equipment considering multipath effect. The method acquires multi-constellation multi-frequency GNSS original observation data in real time and carries out pretreatment; based on multi-feature fusion technology, code load combination, signal-to-noise ratio change and satellite geometric features are comprehensively utilized to construct a multi-dimensional feature vector, and the multi-dimensional feature vector is input into a pre-trained multipath pollution evaluation model to output a continuous multipath pollution intensity index MPI; according to the MPI and the receiver motion state, a smoothing filter time constant is dynamically and adaptively adjusted, and each observation value is allocated a weight; a robust estimation algorithm combined with high-precision carrier phase constraint is used to perform robust processing on the pseudorange observation value, and high-precision smoothed pseudorange is output. Through intelligent evaluation and multi-source information adaptive fusion processing of machine learning, the application effectively suppresses the influence of multipath effect and gross error, and significantly improves the precision and reliability of GNSS pseudorange positioning in a complex environment.

Description

Technical Field

[0001] This invention relates to the field of satellite navigation and positioning technology, and in particular to a GNSS pseudorange smoothing method, system, and device for suppressing multipath errors. Background Technology

[0002] In complex terrains such as urban areas and mountainous regions, Global Navigation Satellite Systems (GNSS) signals are easily affected by reflective objects such as buildings, ground surfaces, and trees, resulting in multipath effects. This has become a key technical bottleneck restricting the improvement of GNSS positioning accuracy. Multipath effects occur when a receiver receives both the direct signal transmitted by the satellite and reflected signals from surrounding objects. The interference and superposition of these two signals cause additional time or phase delays, leading to observations that deviate from the true value, resulting in so-called multipath errors.

[0003] Multipath signals can be divided into line-of-sight (LOS) and non-line-of-sight (NLOS) multipath errors. LOS multipath error is caused by the receiver receiving reflected / diffracted signals, and its impact on pseudorange observations is generally on the order of meters. It can be suppressed by using choke antennas or improving receiver signal tracking algorithms. NLOS multipath effect is caused by satellite signals on non-direct paths being received by the receiver through reflection / diffraction. Depending on the observation environment, its error can reach tens or even hundreds of meters.

[0004] Global Navigation Satellite Systems (GNSS) provide users with all-weather positioning services. However, in complex environments such as cities and canyons, satellite signals are easily reflected and blocked by obstacles such as buildings, resulting in multipath effects and non-direct signal (NLOS) reception problems. This leads to pseudorange observations containing a large number of errors, severely reducing positioning accuracy and reliability.

[0005] Traditional pseudorange smoothing methods often employ carrier phase smoothing pseudorange techniques, but their fixed smoothing windows cannot adapt to changing environments. In cases of severe multipath effects, a fixed long smoothing window can introduce historical contamination data, leading to error accumulation. Furthermore, traditional methods lack effective mechanisms to suppress significant observational gross errors. While some recent studies have attempted to incorporate single features such as signal-to-noise ratio (SNR) or elevation angle for weighting, these methods have failed to fully utilize multi-dimensional observational information, resulting in insufficiently accurate quantitative assessments of multipath effects and limited processing effectiveness.

[0006] Therefore, there is an urgent need for a pseudorange smoothing method that can intelligently assess the degree of multipath contamination and adaptively process it to significantly improve positioning performance in complex environments. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the prior art and provide a GNSS pseudorange smoothing method, system and device that takes into account the multipath effect. It can intelligently quantify multipath error and dynamically adjust the processing strategy to finally output a high-precision smoothed pseudorange result.

[0008] In a first aspect, embodiments of this application provide a GNSS pseudorange smoothing method that takes into account multipath effects, the method comprising:

[0009] S1. Acquire raw GNSS observation data in real time, including at least pseudorange observations, carrier phase observations, signal-to-noise ratio (SNR), and satellite elevation information;

[0010] S2. Real-time multipath contamination detection and quantitative assessment of the observation data based on multi-feature fusion technology, wherein the multi-feature fusion technology comprehensively utilizes the code-on-board combined CMC features, the signal-to-noise ratio change features reflecting signal strength, and the geometric correlation features based on satellite elevation and azimuth angles, and uses machine learning algorithms to construct a multipath contamination assessment model.

[0011] S3. Based on the quantification results of the multipath contamination level and the receiver motion state, dynamically and adaptively adjust the time constant of the smoothing filter and assign weights to each observation, wherein when the multipath contamination is severe, the smoothing time constant is automatically reduced and the low-quality observations are deweighted.

[0012] S4. A robust estimation algorithm is adopted, and combined with the high-precision constraint conditions formed by the carrier phase observations, the pseudorange observations affected by multipath are subjected to robust processing to suppress the influence of gross errors and output the final smoothed high-precision pseudorange result.

[0013] S5. Integrate the various functional modules into a complete pseudorange smoothing system to optimize algorithm performance and operating efficiency.

[0014] Optionally, in one implementation of the first aspect of the present invention, the real-time acquisition of raw GNSS observation data in step S1 specifically includes:

[0015] Receive multi-frequency observation data from a multi-constellation GNSS system, which includes at least two of the GPS, GLONASS, BDS, and Galileo systems;

[0016] The observation data also includes carrier-to-noise ratio (C / N0), Doppler shift observations, and satellite azimuth information;

[0017] The raw observation data is preprocessed, including gross error removal, cycle slip detection and repair, and observation integrity check.

[0018] Optionally, in one implementation of the first aspect of the present invention, the step S2 of performing real-time multi-path contamination detection and quantitative assessment of the observation data based on multi-feature fusion technology specifically includes:

[0019] Construct a multidimensional feature vector, which consists of at least the following features: the code carrier combination CMC sequence and its time difference calculated based on pseudorange and carrier phase observations, the attenuation and fluctuation statistics of signal-to-noise ratio (SNR) or carrier-to-noise ratio (C / N0), and the reflector geometric risk factor determined by the satellite elevation angle and azimuth angle.

[0020] The multidimensional feature vector is input into the pre-trained multipath pollution assessment model, which takes the multidimensional feature vector as input and outputs a continuous multipath pollution intensity index (MPI).

[0021] The machine learning classification algorithm is one of Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM), or Random Forest. The model is trained using a multi-path effect dataset containing various typical environment annotations.

[0022] Optionally, in one implementation of the first aspect of the present invention, the multipath pollution assessment model is implemented using an improved recurrent neural network structure based on a spatiotemporal attention mechanism, and the core calculation process of the multipath pollution intensity index (MPI) during its forward propagation includes the following steps:

[0023] Step 1: Spatiotemporal Feature Enhancement: Enhance the input multidimensional feature vector The transformation is performed, fusing feature sequences with a historical window length of L, and then weighted and fused using a spatiotemporal attention weight matrix A to generate an enhanced feature representation. :

[0024] ,

[0025] Where ⊙ represents the Hadamarda product, Indicates time;

[0026] The spatiotemporal attention weight matrix Calculated using the following formula:

[0027] ,

[0028] Wherein, Q, K, and V represent Query, Key, and Value, respectively, and are obtained from the historical feature sequence through different learnable linear transformation layers. This is the scaling factor;

[0029] Step 2: Calculation of Multipath Pollution Intensity Index (MPI):

[0030] The enhanced spatiotemporal features The input is fed into a fully connected output layer to compute the final result. value:

[0031] ,

[0032] in, Indicates the operation of the gated loop unit; and The learnable weight matrix and bias vector for the output layer; This is the Sigmoid activation function.

[0033] Optionally, in one implementation of the first aspect of the present invention, the step S3 of dynamically and adaptively adjusting the time constant of the smoothing filter and assigning weights to each observation specifically includes the following sub-steps:

[0034] Based on the multipath pollution intensity index at the current moment The smoothing time constant is calculated using the following nonlinear mapping function. :

[0035] ,

[0036] in, The preset maximum time constant reference value, The attenuation coefficient is greater than 0;

[0037] S3.2, Adaptive Motion State:

[0038] The Doppler frequency shift rate of carrier phase observations was calculated. To determine the motion state of the receiver:

[0039] ,

[0040] in, This indicates the calculation of standard deviation. This represents the change in Doppler frequency shift. Set the sliding time window length; set the motion state threshold. Make a judgment;

[0041] In a static state, the time constant Multiply by the static gain factor ;

[0042] In a dynamic state, the time constant Multiply by dynamic gain factor ;

[0043] S3.3, Overall Weight Allocation of Observations:

[0044] Assign a comprehensive mass weight to the observations of each satellite. :

[0045] ,

[0046] Among them, the signal-to-noise ratio weight , The highest signal-to-noise ratio among all current satellites; multipath contamination weighting. , For the first Multipath pollution intensity index of a satellite.

[0047] Optionally, in one implementation of the first aspect of the present invention, step S4, which involves employing a robust estimation algorithm and combining it with carrier phase observations for robust processing, specifically includes the following sub-steps:

[0048] Define the cost function for:

[0049] ,

[0050] Among them, residual , For the residual Huber loss function;

[0051] S4.2 Introducing carrier phase variation as a constraint: utilizing the high-precision variation of carrier phase observations between adjacent epochs. Establish the change in pseudorange Given the constraints, construct the following constrained least squares problem;

[0052] S4.3, Performing Reasonableness Verification and Correction by Integrating Historical Smoothing Results: Using time series analysis methods, calculate the smoothed pseudo-mist value at the current time. Relative to its historical window smoothed value sequence Standardized residuals and with threshold Compare and replace outliers.

[0053] Optionally, in one implementation of the first aspect of the present invention, step S5 specifically includes:

[0054] The multipath contamination assessment model, the dynamic adaptive smoothing filter, and the robust differential processing module are integrated into a complete pseudorange smoothing system. Through software optimization and hardware acceleration technologies, the computational resource allocation and scheduling of the system's algorithm flow are optimized to improve the system's real-time processing capability and operating efficiency.

[0055] In a second aspect, embodiments of this application provide a GNSS pseudorange smoothing system that takes into account multipath effects, applied to the GNSS pseudorange smoothing method that takes into account multipath effects as described in the second aspect, the system comprising:

[0056] The data acquisition module is used to acquire raw GNSS observation data in real time. The data includes at least pseudorange observations, carrier phase observations, signal-to-noise ratio (SNR), and satellite elevation information.

[0057] The multipath assessment module is used to perform real-time multipath contamination detection and quantitative assessment of the observation data based on multi-feature fusion technology. The multi-feature fusion technology comprehensively utilizes the on-board combined CMC features, the signal-to-noise ratio change features reflecting signal strength, and the geometric correlation features based on satellite elevation and azimuth angles, and uses machine learning algorithms to construct a multipath contamination assessment model.

[0058] An adaptive filtering control module is used to dynamically and adaptively adjust the time constant of the smoothing filter and assign weights to each observation value based on the quantization result of the multipath contamination degree and the receiver motion state. When the multipath contamination is severe, the smoothing time constant is automatically reduced and the low-quality observation values ​​are deweighted.

[0059] The robust error handling module is used to perform robust estimation algorithm and combine high-precision constraints composed of carrier phase observations to perform error handling on pseudorange observations affected by multipath, suppress gross errors, and output the final smoothed high-precision pseudorange result.

[0060] The system integration and optimization module is used to integrate the various functional modules into a complete pseudorange smoothing system, thereby optimizing algorithm performance and operating efficiency.

[0061] Thirdly, embodiments of this application provide an electronic device, including:

[0062] processor;

[0063] Memory used to store processor-executable instructions;

[0064] The processor is configured to implement the GNSS pseudorange smoothing method that takes into account multipath effects as described in the first aspect when executing the instructions.

[0065] Fourthly, embodiments of this application provide a computer-readable storage medium storing a program that instructs a device to perform the GNSS pseudorange smoothing method taking into account multipath effects as described in the first aspect.

[0066] This invention discloses a GNSS pseudorange smoothing method, system, and device that takes into account multipath effects. The method first acquires and preprocesses raw GNSS observation data from multiple constellations and frequencies in real time. Then, based on multi-feature fusion technology, it constructs a multi-dimensional feature vector by comprehensively utilizing code combination, signal-to-noise ratio variation, and satellite geometric features, and inputs it into a pre-trained multipath contamination assessment model, outputting a continuous multipath contamination intensity index (MPI). Next, it dynamically and adaptively adjusts the smoothing filter time constant and assigns weights to each observation value based on the MPI and receiver motion state. Finally, a robust estimation algorithm incorporating high-precision carrier phase constraints is used to robustly process the pseudorange observations, outputting a high-precision smoothed pseudorange. This invention effectively suppresses multipath effects and gross errors through machine learning intelligent assessment and adaptive fusion processing of multi-source information, significantly improving the accuracy and reliability of GNSS pseudorange positioning in complex environments.

[0067] Beneficial effects:

[0068] 1. Significantly improve positioning accuracy: By intelligently assessing the degree of multipath contamination and dynamically adjusting the filtering strategy, combined with robust estimation algorithms and carrier phase constraints, multipath effects and gross errors in observations are effectively suppressed, directly improving the quality of pseudorange observations, thereby significantly improving the absolute and relative positioning accuracy of GNSS in complex environments.

[0069] 2. Enhance system reliability and robustness: The adaptive weight allocation and robustness processing mechanism can automatically reduce the impact of unreliable observations, and the rationality check of historical smoothing values ​​further ensures the continuity of output results, enabling the system to maintain stable and reliable performance even in harsh observation environments such as strong multipath and strong occlusion.

[0070] 3. Achieve intelligent processing: Utilize machine learning models (such as GBDT and Attention-GRU networks) to automatically learn the complex mapping relationship between multi-path features and pollution levels, overcoming the limitations of traditional models that rely on human experience thresholds and fixed parameters, and achieving more refined and adaptive quantitative evaluation and processing of multi-path effects. Attached Figure Description

[0071] Figure 1 This is a schematic diagram of a GNSS pseudorange smoothing method that takes into account multipath effects, provided as an embodiment of this application.

[0072] Figure 2 This is a structural diagram of a multipath pollution assessment model provided in an embodiment of this application.

[0073] Figure 3 A flowchart of a robust estimation algorithm provided in an embodiment of this application.

[0074] Figure 4A GNSS pseudorange smoothing system architecture diagram considering multipath effects is provided in one embodiment of this application.

[0075] Figure 5 A schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0076] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.

[0077] It should be noted that, in the embodiments of this application, "at least one" refers to one or more, and "more than one" refers to two or more. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application.

[0078] It should be noted that in the embodiments of this application, the terms "first," "second," etc., are used only for descriptive purposes and should not be construed as indicating or implying relative importance, nor as indicating or implying order. Features specified as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0079] Based on the embodiments described in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0080] Example 1

[0081] Figure 1 This is a schematic flowchart of a GNSS pseudorange smoothing method considering multipath effects, provided as an embodiment of this application. Figure 1 As shown, a GNSS pseudorange smoothing method that takes into account multipath effects includes:

[0082] S1. Acquire raw GNSS observation data in real time. This data includes at least pseudorange observations, carrier phase observations, signal-to-noise ratio (SNR), and satellite elevation information. This data can be acquired in real time via a GNSS receiver or API interface and processed and analyzed using relevant algorithms.

[0083] Specifically, in this embodiment, the real-time acquisition of raw GNSS observation data in step S1 includes receiving multi-frequency observation data from a multi-constellation GNSS system, wherein the multi-constellation GNSS system includes at least two of the GPS, GLONASS, BDS and Galileo systems.

[0084] The observation data also includes carrier-to-noise ratio (SNR), Doppler shift observations, and satellite azimuth information. SNR is an important indicator of signal quality, Doppler shift observations are used to assess the Doppler effect of the signal, and satellite azimuth information is used to assess the geometric conditions of the satellite signal. This information is used in GNSS data preprocessing to evaluate data quality and positioning accuracy.

[0085] The raw observation data undergoes preprocessing, including gross error removal, cycle slip detection and repair, and observation integrity checks, to obtain standard data. Gross error removal eliminates outliers, cycle slip detection and repair handles cycle slip issues in the observations, and observation integrity checks ensure data integrity. Specifically, pseudorange and carrier phase observations with a dual-frequency signal-to-noise ratio less than 28 dBHz, and observations corresponding to epochs where pseudorange observations differ by more than 50 m, are removed. Cycle slip detection and repair using multi-frequency combined observations (such as multi-frequency combined observations) can detect and repair cycle slips in real time. This provides a reliable data foundation for subsequent high-precision processing based on machine learning.

[0086] The integrity of observations is ensured through data integrity checks, such as those using methods like TEQC, BNC, and gNut-Anubis. Multi-GNSS data preprocessing and analysis can be performed using GNSS data preprocessing tools (such as Anubis, GAMIT / GLOBK, etc.). These tools support the processing of multi-frequency observations and cycle slip detection and repair.

[0087] GNSS data quality assessment includes data integrity, cycle slip detection, multipath error, ionospheric delay effects, carrier-to-noise ratio, pseudorange and carrier phase noise, and satellite elevation and azimuth angles. These assessments help ensure data reliability and positioning accuracy.

[0088] S2. Real-time multipath contamination detection and quantitative assessment of the observation data based on multi-feature fusion technology, wherein the multi-feature fusion technology comprehensively utilizes the on-board combined CMC features, the signal-to-noise ratio change features reflecting signal strength, and the geometric correlation features based on satellite elevation and azimuth angles, and uses machine learning algorithms to construct a multipath contamination assessment model.

[0089] Multi-feature fusion technology is a method that combines multiple features (such as code combination, signal-to-noise ratio, satellite elevation angle, etc.) to improve model performance and interpretability. In machine learning, feature fusion is often used to enhance model performance and interpretability. Multipath effects are one of the main sources of error in GNSS positioning, with impacts including signal delay and increased noise. Detection and quantitative assessment of multipath contamination are key issues in GNSS data processing. Methods for detecting and mitigating multipath effects include analysis based on features such as multi-frequency signals, signal-to-noise ratio, and satellite elevation angle. Artificial intelligence (AI) is used to enhance GNSS PPP algorithms to mitigate multipath effects, and feature engineering (such as signal-to-noise ratio, elevation angle, code residuals, etc.) is used for the detection and quantification of multipath effects. Multi-feature fusion technology combines multiple features (such as code combination, signal-to-noise ratio, satellite elevation angle, etc.) and constructs a multipath contamination assessment model through machine learning algorithms to achieve the detection and quantitative assessment of multipath contamination. Furthermore, the combination of multi-feature fusion technology and machine learning algorithms can improve the accuracy of multipath contamination detection and quantitative assessment.

[0090] Multi-dimensional feature fusion sub-step: Based on the preprocessed standard data, a fusion feature vector for quantitatively evaluating multipath effects is constructed; this feature vector fuses the following three types of heterogeneous features: a) Code domain features: the code carrier combination CMC sequence and its time difference calculated from pseudorange and carrier phase observations; b) Signal strength features: the attenuation of signal-to-noise ratio (SNR) or carrier-to-noise ratio (C / N0) and its fluctuation statistics; c) Geometric domain features: the reflector geometric risk factor determined by the satellite elevation angle and azimuth angle; Intelligent evaluation sub-step: The fusion feature vector is input into a pre-trained multipath pollution evaluation model, which maps and outputs a continuous multipath pollution intensity index (MPI); wherein, the MPI value range is [0,1], and the larger the value, the more severe the multipath pollution.

[0091] Model selection and training sub-steps: The multi-path pollution assessment model is implemented using a supervised machine learning algorithm, specifically selected from Gradient Boosting Decision Tree (GBDT), Support Vector Machine (SVM), or Random Forest. The model is trained using a multi-path effect dataset collected and labeled under various typical observation environments to learn the complex nonlinear mapping relationship between the fused feature vector and the degree of multi-path pollution.

[0092] Specifically, in this embodiment, the real-time multipath contamination level detection and quantitative assessment of the observation data based on multi-feature fusion technology in S2 includes: constructing a multi-dimensional feature vector, which consists of at least the following features: a code carrier combination (CMC) sequence calculated based on pseudorange and carrier phase observations and its time difference, the attenuation and fluctuation statistics of signal-to-noise ratio (SNR) or carrier-to-noise ratio (C / N0), and a reflector geometric risk factor determined by the satellite elevation angle and azimuth angle; inputting the multi-dimensional feature vector into a pre-trained multipath contamination assessment model, which uses a machine learning classification algorithm, takes the multi-dimensional feature vector as input, and outputs a continuous multipath contamination intensity index (MPI); the machine learning classification algorithm is one of gradient boosting decision tree (GBDT), support vector machine (SVM), or random forest, and the model is trained using a multipath effect dataset containing various typical environmental annotations for supervised learning.

[0093] Figure 2 This is a structural diagram of a multipath pollution assessment model provided in an embodiment of this application. Figure 2 As shown, the multipath pollution assessment model is implemented using an improved recurrent neural network structure based on a spatiotemporal attention mechanism. The core calculation process of the multipath pollution intensity index (MPI) during its forward propagation includes the following steps:

[0094] Step 1: Spatiotemporal Feature Enhancement: Enhance the input multidimensional feature vector The transformation considers not only the features at the current time step but also incorporates feature sequences from historical window length L, resulting in a total of L+1 time step feature sequences. The enhanced feature representation is generated by weighted fusion using the spatiotemporal attention weight matrix A. :

[0095] ,

[0096] The attention weight matrix is ​​calculated using the Hadamard Product. The feature representation is fused with the feature sequence to generate an enhanced feature representation. Where ⊙ represents the Hadamard product, and the attention weight matrix is... Calculated using the following formula:

[0097] ,

[0098] Here, Q, K, and V are the three core vectors in attention calculation, representing Query, Key, and Value, respectively, which are obtained from the historical feature sequence through different learnable linear transformation layers. This is the scaling factor.

[0099] Through the above steps, not only are the features at the current moment considered, but also the features within the historical window are fused. These features are then weighted and fused using an attention mechanism, thereby generating an enhanced feature representation. This process effectively combines temporal and spatial information, improving the model's expressive power and performance.

[0100] Step 2: Calculation of Multipath Pollution Intensity Index (MPI): This involves calculating the enhanced spatiotemporal characteristics. The input is fed into a gated recurrent unit (GRU) for nonlinear transformation and state update, and the final value is calculated. value:

[0101] ,

[0102] in, This represents the nonlinear transformation and state update performed by the gated cyclic unit on the spatiotemporal feature sequence; and The learnable weight matrix and bias vector for the output layer; The MPI is the sigmoid activation function, which maps the output value to the [0,1] interval. The closer the MPI value is to 1, the more severe the multipath pollution.

[0103] A model based on a gated recurrent unit (GRU) is proposed to predict the severity of multipath contamination (MPI). The model performs nonlinear transformations and state updates on the spatiotemporal feature sequences using the GRU, and then passes the results through an output layer containing a learnable weight matrix. and bias vector The prediction is performed, and finally the Sigmoid activation function is used to map the output to the [0,1] interval to represent the severity of multipath pollution.

[0104] S3. Based on the quantification results of the multipath contamination level and the receiver motion state, dynamically and adaptively adjust the time constant of the smoothing filter and assign weights to each observation, wherein when the multipath contamination is severe, the smoothing time constant is automatically reduced and the low-quality observations are deweighted.

[0105] Specifically, in this embodiment, step S3, which involves dynamically and adaptively adjusting the time constant of the smoothing filter and assigning weights to each observation, includes the following sub-steps:

[0106] (a) Calculation of dynamic time constant: based on the multipath pollution intensity index The smoothing time constant at the current moment is dynamically calculated using the following nonlinear function. :

[0107] ,

[0108] in, The preset maximum time constant reference value, The attenuation coefficient is greater than 0. This is the multipath pollution intensity index; the function establishes a negative correlation mapping relationship between the time constant and the multipath intensity, when... Increasing the time constant indicates severe multipath contamination. Automatic reduction to decrease reliance on historical contamination data.

[0109] (b) Motion state adaptation: by calculating the Doppler frequency shift rate of carrier phase observations. To determine the motion state of the receiver:

[0110] ,

[0111] in, This indicates the calculation of the standard deviation. This represents the change in Doppler frequency shift. Set the time window length; set a motion state threshold. ,when The receiver is determined to be in a dynamic state if it is in a dynamic state, otherwise it is in a static state.

[0112] Under static conditions, the calculated time constant Multiply by a static gain coefficient greater than 1 .

[0113] Under dynamic conditions, τ is multiplied by a dynamic gain coefficient less than 1. This is to achieve the final time constant adjustment for motion state adaptation, ultimately obtaining a smooth time constant for motion state adaptation. .

[0114] (c) Overall weighting of observations: A comprehensive quality weight is assigned to each satellite observation. This weight is determined by the signal-to-noise ratio weight. and multi-path pollution weight Joint decision:

[0115] ,

[0116] Among them, the signal-to-noise ratio weight , Among all current satellites Maximum value (maximum signal-to-noise ratio); multipath contamination weight , For the first The multipath pollution intensity index corresponding to each satellite; finally, in the smoothing filter, the weights of the observations are... It is proportional to the value of low-quality observations, thus enabling automatic downweighting.

[0117] By dynamically adjusting the time constant and observation weights of the smoothing filter, and combining information on multipath contamination intensity, motion state, and signal-to-noise ratio, adaptive processing of observation data is achieved to improve positioning accuracy and robustness.

[0118] S4. A robust estimation algorithm is adopted, and combined with the high-precision constraint conditions formed by the carrier phase observations, the pseudorange observations affected by multipath are subjected to robust error processing to suppress the influence of gross errors, and the final smoothed high-precision pseudorange result is output.

[0119] Specifically, in this embodiment, step S4 involves using a robust estimation algorithm and combining it with carrier phase observations for robust processing. Figure 3 A flowchart of a robust estimation algorithm provided in one embodiment of this application is shown. Figure 3 As shown, the specific steps include the following:

[0120] (a) Construct an M-estimation model based on the Huber loss function for robust smoothing: Define the cost function J(Δx) as:

[0121] ,

[0122] Among them, residual , For the residual The Huber loss function has the following specific form:

[0123] ,

[0124] in, As an adjustment constant, Let be the smoothed pseudo-distance value to be determined. These are values ​​calculated using the pseudorange observation model. To design the matrix, The state correction vector is to be estimated. The loss function uses a quadratic term to process normal observations with small residuals to preserve their information, and automatically reduces the weight of abnormal observations with large residuals to linear processing, thereby suppressing the gross errors caused by multipath.

[0125] (b) Introducing carrier phase variation as a constraint: utilizing the high-precision variation of carrier phase observations between adjacent epochs. Establish the change in pseudorange Given the constraints, construct the following constrained least squares problem:

[0126] ,subject to ,

[0127] in, The carrier wavelength; when the change in pseudorange at the current epoch is detected. The change derived from the carrier phase When there is a significant discrepancy, this constraint condition is applied to the change in the smoothed pseudo-millimeter. Make corrections and prioritize trusting carrier phase information.

[0128] (c) Integrating historical smoothing results for rationality verification and correction: Using time series analysis methods, calculate the smoothed pseudo-mist value at the current time. Relative to its historical window smoothed value sequence Standardized residuals :

[0129] ,

[0130] in, , These are the mean and standard deviation of the smoothed values ​​within the historical window, respectively; a threshold is set. ,when At that time, determine the current smoothing value. Outliers are identified and replaced with predicted values ​​or medians based on historical values ​​to ensure the continuity and reliability of the output.

[0131] S5. Integrate the various functional modules into a complete pseudorange smoothing system to optimize algorithm performance and operating efficiency.

[0132] Specifically, in this embodiment, step S5 includes:

[0133] The multipath contamination assessment model, the dynamic adaptive smoothing filter, and the robust differential processing module are integrated into a complete pseudorange smoothing system. Through software optimization and hardware acceleration technologies, the computational resource allocation and scheduling of the system's algorithm flow are optimized to improve the system's real-time processing capability and operating efficiency.

[0134] Traditional multipath detection methods often rely on single observation features, resulting in limited detection accuracy. This project innovatively proposes a multi-feature fusion detection technique, constructing a comprehensive detection model that includes the following features: **Code-borne Combination Features:** Utilizing a combination of pseudorange and carrier phase observations, the influence of receiver clock bias and satellite clock bias is eliminated, highlighting multipath error characteristics. **Signal Quality Features:** Analyzing the trends in signal-to-noise ratio (SNR) and carrier-to-noise ratio (C / N0) to identify signal quality degradation caused by multipath. **Geometric Correlation Features:** Considering the impact of geometric factors such as satellite elevation and azimuth on multipath effects, a multipath risk assessment model based on satellite geometric distribution is established. The multi-feature fusion model is trained using machine learning algorithms to achieve accurate quantification of multipath contamination levels, improving detection accuracy by more than 40% compared to traditional methods.

[0135] Example 2

[0136] like Figure 4 As shown, this application provides a GNSS pseudorange smoothing system architecture that takes into account multipath effects, applied to the GNSS pseudorange smoothing system that takes into account multipath effects as described in Embodiment 1, including a data acquisition module 11, a multipath evaluation module 12, an adaptive filtering control module 13, a robust differential processing module 14, and a system integration and optimization module 15.

[0137] The data acquisition module 11 is used to acquire raw GNSS observation data in real time. The data includes at least pseudorange observations, carrier phase observations, signal-to-noise ratio (SNR), and satellite elevation information.

[0138] The multipath assessment module 12 is used to perform real-time multipath contamination detection and quantitative assessment of the observation data based on multi-feature fusion technology. The multi-feature fusion technology comprehensively utilizes the on-board combined CMC features, the signal-to-noise ratio change features reflecting signal strength, and the geometric correlation features based on the satellite elevation angle and azimuth angle, and uses machine learning algorithms to construct a multipath contamination assessment model.

[0139] The adaptive filtering control module 13 is used to dynamically and adaptively adjust the time constant of the smoothing filter and assign weights to each observation value according to the quantization result of the multipath contamination degree and the receiver motion state, wherein the smoothing time constant is automatically reduced and the low-quality observation value is deweighted when the multipath contamination is severe.

[0140] The robust error processing module 14 is used to perform robust estimation algorithm and combine high-precision constraints composed of carrier phase observations to perform error processing on pseudorange observations affected by multipath, suppress gross errors, and output the final smoothed high-precision pseudorange result.

[0141] The system integration and optimization module 15 is used to integrate the various functional modules into a complete pseudorange smoothing system, and optimize algorithm performance and operating efficiency.

[0142] Figure 5 This is an electronic device provided in one embodiment of this application. For example... Figure 5 As shown, the electronic device includes at least the following components: processor 101 and memory 100, communication interface 103, and bus 102.

[0143] In this embodiment of the application, memory 100 is used to store executable instructions of processor 101, which, when configured to execute instructions, implements the method as described in the first aspect.

[0144] In embodiments of this application, a computer-readable storage medium includes instructions that instruct a device to perform the method as described in the first aspect. For example, the instructions instruct the device to perform... Figure 1 The method is shown in the process steps.

[0145] In one embodiment of this application, the program operating in the electronic device may be a program that controls a central processing unit (CPU) or similar device to achieve the functions of the above-described embodiments of the present invention (a program that enables the computer to function). Information processed by these devices is then temporarily stored in random access memory (RAM) during processing, and subsequently stored in various ROMs such as read-only memory (FlashROM) and hard disk drives (HDDs), and read, corrected, and written by the CPU as needed.

[0146] It should be noted that a portion of the electronic device described above can also be implemented using a computer. In this case, the program for implementing the control function can be recorded on a computer-readable recording medium, and the program recorded on the recording medium can be read into the computer and executed.

[0147] It should be noted that the term "computer" as used here refers to a computer built into an electronic device, employing hardware including an operating system and peripheral devices. Furthermore, "computer-readable recording media" refers to removable media such as floppy disks, magneto-optical disks, ROMs, and CD-ROMs, as well as storage devices such as hard drives built into a computer.

[0148] Furthermore, a "computer-readable recording medium" can include: a medium that dynamically stores a program for a short period of time, such as a communication line used when transmitting a program via a network such as the Internet or a communication line such as a telephone line; or a medium that stores a program for a fixed period of time, such as volatile memory inside a computer that serves as a server or client in this case. In addition, the aforementioned program can be a program used to implement the above-mentioned functions, or it can be a program that can implement the above-mentioned functions by combining with programs already recorded in the computer.

[0149] Furthermore, the electronic device in the above embodiments can also be implemented as an assembly (device group) composed of multiple devices. Each device constituting the device group can possess some or all of the functions or functional blocks of the electronic device in the above embodiments. As a device group, it is sufficient to have all the functions or functional blocks of the electronic device.

[0150] Those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any appropriate changes and variations made to the above embodiments within the essential spirit and scope of this application fall within the scope of protection claimed in this application.

Claims

1. A GNSS pseudorange smoothing method considering multipath effects, characterized in that, The method comprises: S1, real-time acquisition of GNSS raw observation data, the data at least including pseudorange observation value, carrier phase observation value, signal-to-noise ratio SNR and satellite elevation angle information; S2, real-time multipath pollution degree detection and quantitative evaluation of the observation data based on multi-feature fusion technology, wherein the multi-feature fusion technology comprehensively utilizes code carrier combination CMC feature, signal-to-noise ratio change feature reflecting signal strength and geometric correlation feature based on satellite elevation angle and azimuth angle, and adopts machine learning algorithm to construct multipath pollution evaluation model; S3, dynamically and adaptively adjusting time constant of smoothing filter and assigning weight to each observation value according to multipath pollution degree quantitative result and receiver motion state, wherein smoothing time constant is automatically reduced and low-quality observation value is down-weighted when multipath pollution is serious; S4, adopting robust estimation algorithm and combining high-precision constraint condition constituted by carrier phase observation value to carry out robust processing on pseudorange observation value affected by multipath, suppresses gross error influence and outputs final smoothed high-precision pseudorange result; S5, integrating each functional module into complete pseudorange smoothing system, optimizing algorithm performance and running efficiency; The step S2 comprises: constructing a multi-dimensional feature vector, the vector at least consisting of the following features: code carrier combination CMC sequence and its time difference based on pseudorange and carrier phase observation value calculation, attenuation amount and fluctuation statistics of signal-to-noise ratio SNR or carrier-to-noise ratio C / N0, and reflection surface geometric risk factor determined by satellite elevation angle and azimuth angle; inputting the multi-dimensional feature vector into the pre-trained multipath pollution evaluation model, the model taking multi-dimensional feature vector as input and outputting a continuous multipath pollution intensity index MPI; The machine learning classification algorithm is one of gradient boosting decision tree GBDT, support vector machine SVM or random forest RandomForest, and the model is obtained by supervised learning training using a multipath effect data set containing multiple typical environment annotations; The multipath pollution evaluation model adopts an improved recurrent neural network structure based on space-time attention mechanism, and the core calculation process of the multipath pollution intensity index MPI in the forward propagation process comprises the following steps: Step one, spatio-temporal feature enhancement: transform the input multi-dimensional feature vector transformed, the feature sequence with a history window length of L is fused, weighted fusion is performed through a spatio-temporal attention weight matrix A, and an enhanced feature representation is generated : ， wherein, ⊙ denotes Hadamard product, denotes time instant; the spatio-temporal attention weight matrix is calculated by the following equation: ， where Q, K, V represent Query, Key and Value respectively, which are obtained by different learnable linear transformation layers from the history feature sequence, is a scaling factor; Step two, multipath pollution intensity index MPI calculation: The enhanced spatio-temporal features are input to a fully connected output layer, which computes the final values: ， wherein, denotes the operation of a gated recurrent unit; and are the learnable weight matrix and bias vector for the output layer; is a sigmoid activation function. The step S3 comprises the following sub-steps: S3.1, dynamic time constant calculation: Multi-path pollution intensity index based on current time The smoothing time constant is calculated by the following non-linear mapping function : ， wherein is a preset maximum time constant reference value, is a decay coefficient greater than 0; S3.2, motion state adaptation: Determining receiver motion state by calculating doppler shift rate of change of carrier phase observations : ， wherein, represents a standard deviation calculation, represents a change in Doppler shift, is a sliding time window length; set a motion state threshold make a judgment; In the static state, the time constant is multiplied by the static gain factor ; In the dynamic state, the time constant is multiplied by the dynamic gain factor ; S3.3, observation value comprehensive weight distribution: assigning a combined quality weight to the observations for each satellite : ， Wherein, the signal-to-noise ratio weight is , , The maximum signal-to-noise ratio in the current all satellite signal-to-noise ratios The multipath pollution weight , , The multipath pollution intensity index of the first Satellite.

2. The GNSS pseudorange smoothing method considering multipath effect according to claim 1, wherein, The step S1 comprises: receiving multi-frequency point observation data from multi-constellation GNSS system, the multi-constellation GNSS system including at least two of GPS, GLONASS, BDS and Galileo system; The observation data further includes carrier-to-noise ratio C / N0, Doppler frequency shift observation value and satellite azimuth angle information; The original observation data are preprocessed, and the preprocessing includes gross error elimination, cycle slip detection and repair, and observation value integrity check.

3. The GNSS pseudorange smoothing method considering multipath effect according to claim 1, characterized in that, The step S4 of adopting a robust estimation algorithm and combining carrier phase observation values for robust processing specifically includes the following sub-steps: S4.1, constructing an M-estimation model based on a Huber loss function for robust smoothing: Defining a cost function is: ， wherein the residual , is a Huber loss function on the residual ; is the smoothed pseudo-range value to be solved, is the pseudo-range observation model calculated value, is the design matrix, is the state correction vector to be estimated; S4.2, Introduce the carrier phase change amount as a constraint condition: use the high-precision change amount of the carrier phase observation value between adjacent epochs , Establish a constraint relationship with the change amount of the pseudo-range , Construct the following constraint least squares problem; S4.3, fuse history smoothing results to perform rationality test and correction: using time series analysis method, calculate the current time smoothing pseudo-range value relative to the history window within the smoothing value sequence The standardized residual , and compared with threshold , replace the abnormal value.

4. The GNSS pseudorange smoothing method considering multipath effect according to claim 3, characterized in that, The step S5 specifically includes: The multi-path pollution evaluation model, the dynamic adaptive smoothing filter, and the robust robustness processing module are integrated into a complete pseudorange smoothing system; through software optimization and hardware acceleration technology, the algorithm flow of the system is allocated and scheduled to optimize the real-time processing capacity and running efficiency of the system.

5. A GNSS pseudorange smoothing system considering multipath effects, applied to the GNSS pseudorange smoothing method considering multipath effects according to any one of claims 1 to 4, characterized in that, The system comprises: A data acquisition module for acquiring GNSS original observation data in real time, the data at least including pseudorange observation values, carrier phase observation values, signal-to-noise ratio SNR, and satellite elevation angle information; A multi-path evaluation module for detecting and quantitatively evaluating the degree of multi-path pollution of the observation data in real time based on a multi-feature fusion technology, wherein the multi-feature fusion technology comprehensively utilizes code carrier combination CMC features, signal-to-noise ratio change features reflecting signal strength, and geometric correlation features based on satellite elevation angle and azimuth angle, and a machine learning algorithm is used to construct a multi-path pollution evaluation model; An adaptive filtering control module for dynamically and adaptively adjusting the time constant of the smoothing filter and assigning weights to each observation value according to the multi-path pollution degree quantitative result and the receiver motion state, wherein the smoothing time constant is automatically reduced and the low-quality observation value is de-weighted when the multi-path pollution is serious; A robust robustness processing module for adopting a robust estimation algorithm and combining high-precision constraints of carrier phase observation values to perform robust processing on the pseudorange observation values affected by multi-path, suppress the influence of gross errors, and output the final smoothed high-precision pseudorange result; A system integration and optimization module for integrating the functional modules into a complete pseudorange smoothing system to optimize the algorithm performance and running efficiency.

6. An electronic device, comprising: A processor; A memory for storing processor-executable instructions; wherein the processor is configured to implement the GNSS pseudorange smoothing method considering multi-path effect as claimed in any one of claims 1 to 4 when executing the instructions.

7. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a program instructing the device to execute the GNSS pseudorange smoothing method considering multi-path effect as claimed in any one of claims 1 to 4.

Images