A GNSS / INS integrated navigation method, system and device

Abstract

The present invention relates to a GNSS / INS combined navigation method, system and device, wherein the method comprises: receiving GNSS pseudo-range data acquired by a GNSS component, obtaining GNSS data after error correction; receiving IMU signals acquired by an INS component, obtaining INS data after processing; integrating GNSS data with INS data according to different domains of GNSS data, obtaining GNSS data with at least one degree of integration with INS data; inputting GNSS data with at least one degree of integration with INS data into a pre-constructed Kalman filter model, obtaining at least one set of observation data. Thus, the present invention significantly improves the overall performance of the navigation system by enhancing the robustness of the system, simplifying the complexity of the system and improving the accuracy by using Kalman filtering, so that it can provide stable and reliable navigation information in various complex environments.

Description

Technical Field

[0001] The present invention relates to the field of satellite navigation technology, and in particular to a GNSS / INS integrated navigation method, system and device. Background Art

[0002] GNSS (Global Navigation Satellite System) and INS (Inertial Navigation System) are two navigation methods with their own characteristics, showing a wide range of application value in the field of navigation and positioning. GNSS receives signals from the global navigation satellite system and uses the principle of triangulation to accurately determine the user's position. It has significant advantages such as high precision and global coverage, allowing users to obtain accurate navigation information at any location.

[0003] However, due to the susceptibility to signal shielding, interference and other factors, the positioning accuracy of GNSS is sometimes reduced. In contrast, INS obtains navigation information by integrating the inertial information such as the carrier's own acceleration and angular velocity. It has the characteristics of strong autonomy and high accuracy in a short time, and is particularly suitable for environments where GNSS signals are unavailable or the signal quality is poor. However, long-term integration of INS will lead to error accumulation, thus affecting navigation accuracy.

[0004] In order to give full play to the advantages of GNSS and INS, and to achieve complementary advantages, the two are combined into an effective solution. The GNSS / INS integrated navigation system combines the globality of GNSS and the autonomy of INS, and can provide stable and reliable navigation information in complex environments and dynamic changes. It not only improves the performance of the navigation system, but also provides users with richer navigation services.

[0005] However, most of the existing GNSS / INS integrated navigation methods are based on traditional filtering algorithms or optimization algorithms. Due to the limitations of these methods themselves, they often find it difficult to fully utilize the large amount of information provided by GNSS and INS when dealing with complex environments and dynamically changing situations, resulting in a certain degree of limitation in the improvement of navigation accuracy and stability. Therefore, it is necessary to seek new technical means to further optimize the performance of the GNSS / INS integrated navigation system. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a GNSS / INS integrated navigation method, system and device, which solve the technical problems that the existing GNSS / INS integrated navigation method has limitations when dealing with complex environments and dynamic changes, and it is difficult to fully utilize the information of GNSS and INS, which limits the improvement of navigation accuracy and stability.

[0008] (II) Technical solution

[0009] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0010] In a first aspect, an embodiment of the present invention provides a GNSS / INS integrated navigation method, including:

[0011] Receive GNSS pseudorange data acquired by the GNSS component in the GNSS / INS integrated navigation, and obtain GNSS data after error correction;

[0012] Receive the IMU signal obtained by the INS component in the GNSS / INS integrated navigation, and obtain INS data after processing;

[0013] Integrate the GNSS data with the INS data according to different domains of the GNSS data to obtain GNSS data having at least one degree of integration with the INS data;

[0014] The GNSS data that forms at least one degree of integration with the INS data is input into a pre-built Kalman filter model to obtain at least one set of observation data.

[0015] Optionally, receiving GNSS pseudorange data acquired by a GNSS component in a GNSS / INS integrated navigation, and obtaining GNSS data after error correction includes:

[0016] According to the propagation time of the satellite signal received by the GNSS component, an approximate value of the geometric distance between the GNSS component and the satellite is obtained as GNSS pseudorange data;

[0017] Identify and quantify various error sources including satellite and GNSS component clock errors, ionospheric delays, and tropospheric delays;

[0018] Based on various error sources, error correction is performed on the received GNSS pseudorange data to obtain error-corrected pseudorange measurement values;

[0019] The user's three-dimensional position information and GNSS component clock error are obtained by solving data from at least four satellites;

[0020] Generate GNSS data based on the user's three-dimensional position information, GNSS component clock error and corrected GNSS pseudorange data.

[0021] Optionally, the GNSS pseudorange data is: ;

[0022] In the formula, n Represents the satellite number, r nRepresents the geometric distance between the satellite and the GNSS component, δt u is the clock error of the user receiver, δt n is the satellite clock error, I n is the delay caused by the ionosphere. T n It is the delay caused by the troposphere. ε n The delay is caused by random disturbance factors;

[0023] The error-corrected pseudorange measurement is: ;

[0024] Geometric distances between satellites and GNSS components r n for: ;

[0025] In the formula, and Satellite n and the user's position vector.

[0026] Optionally, integrating the GNSS data with the INS data according to different domains of the GNSS data to obtain the GNSS data having at least one degree of integration with the INS data includes:

[0027] Identify and determine the data domain where the GNSS data is located;

[0028] When GNSS data is in the position domain, a loosely coupled method integrated with INS is adopted to integrate INS data with GNSS data to generate loosely coupled GNSS data with INS integration function;

[0029] When GNSS data is in the measurement domain, a tightly coupled method is adopted for integration with INS to generate tightly coupled GNSS data with INS integration function.

[0030] Optionally, GNSS data that forms at least one degree of integration with INS data is input into a pre-built Kalman filter model to obtain at least one set of observation data including:

[0031] For the generated loosely coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias and accelerometer bias are selected according to the loosely coupled mode characteristics to construct a first state vector;

[0032] Determine a first state transfer matrix with the Earth-centered Earth-fixed coordinate system as a reference reference system;

[0033] Based on the constructed first state vector and the first state transfer matrix, a first Kalman filter model is obtained;

[0034] Inputting the generated loosely coupled GNSS data with INS integration function into the first Kalman filter model to obtain observation data; wherein the input loosely coupled GNSS data includes the difference between the position and velocity data output by the GNSS component and the INS component;

[0035] For the generated tightly coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias, accelerometer bias, clock error and clock drift are selected according to the tightly coupled mode characteristics to construct the second state vector;

[0036] Determine the second state transfer matrix with the ECEF coordinate system as the reference system;

[0037] Based on the constructed second state vector and the second state transfer matrix, a second Kalman filter model is obtained;

[0038] The generated tightly coupled GNSS data with INS integration function is input into the second Kalman filter model to obtain observation data; wherein the input tightly coupled GNSS data includes difference data formed by GNSS pseudorange data and pseudorange rate.

[0039] Optionally, the first state vector is: ;

[0040] In the formula, is the position error, is the speed error, is the attitude error, b g is the gyroscope bias, b a is the accelerometer zero bias;

[0041] The first state transfer matrix is: ;

[0042] In the formula, is the Earth's rotation angular velocity, τ s is the inertial navigation update period, I 3 and O 3 represent the unit matrix and the zero matrix respectively, is the direction cosine matrix from the carrier system to the ECEF system, is the inertial navigation output specific force, is the gravitational acceleration term, where For distance, is the ECEF position, is the satellite position, is the acceleration noise;

[0043] The position and velocity data output by the GNSS component and the INS component are: ;

[0044] In the formula, ( X G , Y G , Z G )and( X I , Y I , Z I ) represent the position output by GNSS and the position output by INS respectively, ( V xG , V yG , V zG )and( V xI , V yI , V zI ) represent the speed output by GNSS and the speed output by INS respectively;

[0045] The second state vector is: ;

[0046] In the formula, For the clock error, For the clock drift;

[0047] The second state transfer matrix is: ;

[0048] GNSS pseudorange data and pseudorange rate are: ;

[0049] In the formula, are INS pseudorange data and INS pseudorange rate, respectively. r i is the launch time coordinate, Receive coordinates for the user, is the coordinate transformation matrix from ECEF coordinate system to I system, is the total pseudorange error, is the coordinate system unit vector, v i is the velocity at the time of launch, The user receives the speed, is the sum of pseudorange rate errors;

[0050] The difference data formed by GNSS pseudorange data and pseudorange rate is: ;

[0051] In the formula, Indicates the difference between GNSS pseudorange data and INS pseudorange data. Indicates the difference between the GNSS pseudorange rate and the INS pseudorange rate.

[0052] Optionally, after inputting GNSS data having at least one level of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data, the method further includes:

[0053] receiving at least one set of final observation data;

[0054] A residual RES determined by a pseudorange residual or an innovative residual obtained in a Kalman filter model is introduced, combined with an azimuth angle AZ and an elevation angle EL of the satellite in the sky, to form auxiliary data in the form of a data tuple of a spherical coordinate system;

[0055] by performing a coordinate transformation on the auxiliary data to transform the spherical coordinate system into a Cartesian coordinate system in an earth-fixed coordinate system centered on the earth;

[0056] The auxiliary data after coordinate transformation is input into a pre-trained convolutional neural network, and the final observation data is corrected based on the output data of the convolutional neural network to obtain the final observation data.

[0057] In a second aspect, an embodiment of the present invention provides a GNSS / INS integrated navigation system, including:

[0058] The GNSS data generation module is used to receive the GNSS pseudo-range data acquired by the GNSS component in the GNSS / INS integrated navigation and obtain the GNSS data after error correction;

[0059] INS data generation module, used to receive the IMU signal acquired by the INS component in the GNSS / INS integrated navigation, and obtain INS data after processing;

[0060] A data coupling module, for integrating the GNSS data with the INS data according to different domains of the GNSS data, to obtain GNSS data having at least one degree of integration with the INS data;

[0061] The data filtering module is used to input the GNSS data that forms at least one degree of integration with the INS data into a pre-built Kalman filter model to obtain at least one set of observation data.

[0062] In a third aspect, an embodiment of the present invention provides a GNSS / INS integrated navigation device, including:

[0063] A GNSS component including a GNSS receiver configured to acquire GNSS pseudorange data from at least one satellite;

[0064] An INS component configured to be coupled to a GNSS receiver for generating IMU data;

[0065] The control component is configured to execute the GNSS / INS integrated navigation method as described above.

[0066] Optionally include:

[0067] The integration filter is configured to receive data from the GNSS receiver and the INS component and perform fusion processing on the received data through integration operation.

[0068] (III) Beneficial effects

[0069] The beneficial effects of the present invention are:

[0070] First of all, it should be clear that the present invention significantly enhances the robustness of the navigation system. In complex and changeable environments, such as when there is signal blocking or interference, the method can still maintain the stable operation of the navigation system. This is because the method not only relies on the globality of GNSS, but also makes full use of the autonomy of INS, so that the navigation system can respond flexibly when faced with challenges, and exhibits strong anti-interference ability and stability. At the same time, the present invention effectively simplifies the complexity of the navigation system by constructing a unified combined navigation model. This model can realize unified processing and analysis of GNSS and INS data, which not only improves the efficiency of data processing, but also reduces the complexity and maintenance cost of the system. This integrated processing method makes the navigation system more concise and efficient, thereby improving the reliability and maintainability of the system.

[0071] Next, the present invention receives and corrects the pseudorange data obtained by the GNSS component to obtain accurate GNSS data. At the same time, it also receives and processes the IMU signal obtained by the INS component to obtain INS data. These two types of data can provide continuous navigation information through complementarity when the GNSS signal is poor or missing. Next, according to the different domains of the GNSS data, it is integrated with the INS data to form GNSS data with multiple levels of integration. This data integration method not only improves the richness and accuracy of the data, but also helps to achieve more robust navigation in complex and changing environments.

[0072] Finally, the present invention inputs the integrated data into a pre-built Kalman filter model to obtain multiple sets of observation data. Kalman filtering is an efficient recursive filter that can estimate the state of a dynamic system from a series of incomplete and noisy measurements. In this way, the accuracy and stability of navigation can be further improved.

[0073] Therefore, the present invention significantly improves the overall performance of the navigation system by enhancing the robustness of the system, simplifying the complexity of the system and improving the accuracy by using Kalman filtering, so that the navigation system can provide stable and reliable navigation information in various complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A flowchart of a GNSS / INS integrated navigation method provided by an embodiment of the present invention;

[0075] Figure 2 A specific flow chart of step S1 of a GNSS / INS integrated navigation method provided in an embodiment of the present invention;

[0076] Figure 3 A schematic diagram of a specific flow chart of step S3 of a GNSS / INS integrated navigation method provided in an embodiment of the present invention;

[0077] Figure 4 A specific flow chart of step S4 of a GNSS / INS integrated navigation method provided in an embodiment of the present invention;

[0078] Figure 5 A schematic diagram of a specific flow chart after step S4 of a GNSS / INS integrated navigation method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0079] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0080] like Figure 1 As shown, a GNSS / INS integrated navigation method proposed in an embodiment of the present invention includes: receiving GNSS pseudorange data acquired by a GNSS component in the GNSS / INS integrated navigation, and obtaining GNSS data after error correction; receiving IMU signals acquired by an INS component in the GNSS / INS integrated navigation, and obtaining INS data after processing; integrating the GNSS data with the INS data according to different domains of the GNSS data, and obtaining GNSS data having at least one degree of integration with the INS data; inputting the GNSS data having at least one degree of integration with the INS data into a pre-constructed Kalman filter model to obtain at least one set of observation data.

[0081] First of all, it should be clear that the present invention significantly enhances the robustness of the navigation system. In complex and changeable environments, such as when there is signal blocking or interference, the method can still maintain the stable operation of the navigation system. This is because the method not only relies on the globality of GNSS, but also makes full use of the autonomy of INS, so that the navigation system can respond flexibly when faced with challenges, and exhibits strong anti-interference ability and stability. At the same time, the present invention effectively simplifies the complexity of the navigation system by constructing a unified combined navigation model. This model can realize unified processing and analysis of GNSS and INS data, which not only improves the efficiency of data processing, but also reduces the complexity and maintenance cost of the system. This integrated processing method makes the navigation system more concise and efficient, thereby improving the reliability and maintainability of the system.

[0082] Next, the present invention receives and corrects the pseudorange data obtained by the GNSS component to obtain accurate GNSS data. At the same time, it also receives and processes the IMU signal obtained by the INS component to obtain INS data. These two types of data can provide continuous navigation information through complementarity when the GNSS signal is poor or missing. Next, according to the different domains of the GNSS data, it is integrated with the INS data to form GNSS data with multiple levels of integration. This data integration method not only improves the richness and accuracy of the data, but also helps to achieve more robust navigation in complex and changing environments.

[0083] Finally, the present invention inputs the integrated data into a pre-built Kalman filter model to obtain multiple sets of observation data. Kalman filtering is an efficient recursive filter that can estimate the state of a dynamic system from a series of incomplete and noisy measurements. In this way, the accuracy and stability of navigation can be further improved.

[0084] Therefore, this GNSS / INS integrated navigation method significantly improves the overall performance of the navigation system by enhancing the robustness of the system, simplifying the complexity of the system, and improving the accuracy by using Kalman filtering, so that it can provide stable and reliable navigation information in various complex environments.

[0085] In order to better understand the above technical solution, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer and more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.

[0086] Specifically, an embodiment of the present invention provides a GNSS / INS integrated navigation method, which includes:

[0087] S1. Receive GNSS pseudorange data acquired by the GNSS component in the GNSS / INS integrated navigation, and obtain GNSS data after error correction.

[0088] Furthermore, if Figure 2 As shown, step S1 includes:

[0089] S11. According to the propagation time of the satellite signal received by the GNSS component, an approximate value of the geometric distance between the GNSS component and the satellite is obtained as GNSS pseudorange data.

[0090] S12. Identify and quantify various error sources including satellite and GNSS component clock errors, ionospheric delay, tropospheric delay, and other unknown noise.

[0091] It is important to understand that GNSS, or the Global Navigation Satellite System, is an integral part of modern science and technology. It is an airborne radio navigation and positioning system that can provide users with all-weather 3D coordinates, speed, and time information anywhere on the Earth's surface or near-Earth space. GNSS receivers are carefully configured to work with antennas coupled to them to ensure accurate reception of GNSS signals from GNSS signal sources. In the application of satellite positioning technology, its positioning methods can be summarized in detail into the following types according to different classification standards: 1. Based on different positioning references, it can be divided into absolute positioning and relative positioning; 2. Based on different motion states, it can be divided into static positioning and dynamic positioning; 3. Based on the number of observation points, it can be divided into single-point positioning and differential positioning; 4. Based on the technical principle, it can be divided into pseudo-range positioning and carrier phase positioning. Among them, the pseudo-range positioning principle, as the basic technology of satellite navigation positioning, plays a vital role.

[0092] The GNSS pseudorange data is: ;

[0093] In the formula, n Represents the satellite number, r n Represents the geometric distance between the satellite and the GNSS component, δt u is the clock error of the user receiver, δt n is the satellite clock error, I n is the delay caused by the ionosphere. T n It is the delay caused by the troposphere. ε n The delay is caused by random disturbance factors.

[0094] S13. Based on various error sources, error correction is performed on the received GNSS pseudorange data to obtain error-corrected pseudorange measurement values.

[0095] Pseudorange refers to the estimated straight-line distance from the receiver to the satellite. It is not the true distance, but an approximation. This approximation is calculated by the propagation time of the signal received by the receiver. However, due to various factors, such as the influence of the atmosphere, the noise inside the receiver, etc., this pseudorange measurement will have a certain error. Therefore, it is necessary to correct this error to improve the accuracy of the measurement. After various error corrections, the pseudorange measurement will be closer to the true distance value, thereby providing more accurate data for positioning and navigation. Therefore, the pseudorange measurement after various error corrections will be closer to the true distance value, thereby providing more accurate data for positioning and navigation. It is expressed as: .

[0096] S14. Obtain the user's three-dimensional position information and GNSS component clock error through data from at least four satellites.

[0097] S15. Generate GNSS data according to the user's three-dimensional position information, the GNSS component clock error, and the corrected GNSS pseudo-range data.

[0098] In one embodiment, the geometric distance between the satellite and the GNSS component r n for: ;

[0099] In the formula, and Satellite n and the user's position vector. The user's position can be solved by the following nonlinear equations: ;

[0100] In the above equation, there are four unknowns to be solved, including the user's three-dimensional position information and the receiver's clock error. In order to accurately solve these unknowns, at least four satellite data must be observed each time. The most commonly used methods for solving such equations include least squares and Kalman filtering. It is worth noting that both methods require linearization of the original equations during the processing process. Once the signal is successfully captured, the receiver will immediately respond and generate the corresponding GNSS data, which will then be accurately transmitted to the INS component.

[0101] S2. Receive the IMU signal obtained by the INS component in the GNSS / INS integrated navigation, and obtain INS data after processing.

[0102] In a specific embodiment, in order to ensure the accuracy and stability of the data, an integral filter technology is introduced into the data processing flow. After precise design and configuration, it is seamlessly coupled with the GNSS (Global Navigation Satellite System) receiver and INS (Inertial Navigation System) components. The integral filter is introduced to perform an integral operation on the input signal, which can filter out noise and interference and extract useful signal components. The relationship between the integral filter input and output is expressed as: ;

[0103] In the formula, x ( t ) represents the input signal, and y ( t ) represents the output signal after filtering. In addition, T Refers to the time constant of the integrator. In the process of GNSS and INS data fusion, the integration filter can balance the advantages and disadvantages of both, making the data fusion result more accurate and stable. In this way, the data from the INS system and the GNSS receiver can be effectively integrated, further improving the performance and reliability of the navigation system, and improving the stability and availability of the data.

[0104] S3. Integrate the GNSS data with the INS data according to different domains of the GNSS data to obtain GNSS data having at least one degree of integration with the INS data.

[0105] Furthermore, if Figure 3 As shown, step S3 includes:

[0106] S31. Identify and determine the data domain where the GNSS data is located.

[0107] Specifically, the received GNSS data is parsed to extract key information, such as satellite signal strength, pseudorange, carrier phase, etc. Then, the domain of the GNSS data is determined based on the parsed data features. For example, if the received data is processed positioning coordinates, it is determined to be in the position domain; if it is original satellite measurement data, it is determined to be in the measurement domain; if it is the original intermediate frequency or radio frequency data of the satellite signal, it is determined to be in the signal domain.

[0108] S32. When the GNSS data is in the position domain, a loosely coupled method of integration with INS is adopted to integrate the INS data with the GNSS data to generate loosely coupled GNSS data with INS integration function.

[0109] When GNSS data is in the position domain, a loosely coupled approach is adopted to integrate with the inertial navigation system (INS). This approach involves effectively combining and integrating INS data with GNSS data to generate loosely coupled GNSS data with INS integration function. This data can not only provide accurate position information, but also further improve the stability and reliability of positioning through the assistance of INS.

[0110] S33. When GNSS data is in the measurement domain, a tightly coupled method integrated with INS is adopted to generate tightly coupled GNSS data with INS integration function.

[0111] When GNSS data is in the measurement domain, a tightly coupled method is adopted to integrate with the inertial navigation system (INS). In this way, GNSS data and INS data are closely combined, and through complex algorithms and processing processes, a tightly coupled GNSS data with INS integration function is generated. This data can provide more accurate and reliable measurement results, especially in complex environments, and can significantly improve the performance of positioning and navigation.

[0112] Furthermore, when the GNSS data is in the signal domain, an ultra-tight coupling method integrated with INS is adopted to generate ultra-tightly coupled GNSS data with INS integration function.

[0113] When GNSS data is in the signal domain, an ultra-tight coupling method is adopted to integrate with the inertial navigation system (INS). In this way, GNSS signals and INS data are highly integrated, and through advanced processing technology, ultra-tightly coupled GNSS data with INS integration function is generated. This data can not only provide high-precision signal information, but also further enhance the stability and anti-interference ability of the signal through the assistance of INS, so as to maintain good positioning performance in various complex environments.

[0114] S4. Inputting the GNSS data that forms at least one level of integration with the INS data into a pre-built Kalman filter model to obtain at least one set of observation data.

[0115] Furthermore, if Figure 4 As shown, step S3 includes:

[0116] S41. For the generated loosely coupled GNSS data with INS integration function, according to the loosely coupled mode characteristics, select variables including position error, velocity error, attitude error, gyroscope bias and accelerometer bias to construct a first state vector.

[0117] S42. Determine a first state transfer matrix with the Earth-centered Earth-fixed coordinate system as a reference reference system.

[0118] S43. Obtain a first Kalman filter model based on the constructed first state vector and first state transfer matrix.

[0119] S44, inputting the generated loosely coupled GNSS data with INS integration function into the first Kalman filter model to obtain observation data; wherein the input loosely coupled GNSS data includes the difference between the position and velocity data output by the GNSS component and the INS component.

[0120] S45. For the generated tightly coupled GNSS data with INS integration function, according to the tightly coupled mode characteristics, select variables including position error, velocity error, attitude error, gyroscope bias, accelerometer bias, clock error and clock drift to construct a second state vector.

[0121] S46, determining a second state transfer matrix with the ECEF coordinate system as a reference system;

[0122] S47. Obtain a second Kalman filter model based on the constructed second state vector and second state transfer matrix.

[0123] S48. Input the generated tightly coupled GNSS data with INS integration function into the second Kalman filter model to obtain observation data; wherein the input tightly coupled GNSS data includes difference data formed by GNSS pseudorange data and pseudorange rate.

[0124] In a specific implementation, different processing methods are adopted for different states of GNSS data during the operation of the integrated navigation system. Specifically, when the GNSS data is in the position domain, the performance advantage of the IMU processor is fully utilized to integrate it with the GNSS data to generate a loosely coupled GNSS with INS integration function. This is aimed at improving the positioning accuracy and stability of the navigation system. According to the unique mode characteristics of the loose combination, after comprehensive analysis, the position error is selected. , speed error , attitude error , Gyro bias b g And the accelerometer bias b a A total of 15 states, including , are used to construct the state vector. The first state vector is: ;

[0125] When considering relevant parameters, the earth-centered earth-fixed (ECEF) coordinate system is selected as the reference system. Φ kIt is determined as the key matrix under this reference system, which is used to describe the evolution relationship of state variables at different times. The first state transfer matrix is: ;

[0126] In the formula, is the Earth's rotation angular velocity, τ s is the inertial navigation update period, I 3 and O 3 represent the unit matrix and the zero matrix respectively, is the direction cosine matrix from the carrier system to the ECEF system, is the inertial navigation output specific force, is the gravitational acceleration term, where For distance, is the ECEF position, is the satellite position, is the acceleration noise.

[0127] The position and velocity data output by the GNSS and INS systems are differentially calculated, and the results are input into the Kalman filter as observation values. The position and velocity data output by the GNSS component and the INS component are: ;

[0128] The observation matrix is ​​expressed as: ;

[0129] In the formula, ( X G , Y G , Z G )and( X I , Y I , Z I ) represent the position output by GNSS and the position output by INS respectively, ( V xG , V yG , V zG )and( V xI , V yI , V zI ) represent the speed output by GNSS and the speed output by INS respectively.

[0130] When GNSS data is in the measurement domain, a tightly coupled method integrated with INS is selected to generate the corresponding GNSS data. This processing method can more effectively utilize the information of GNSS and INS, further improving the reliability and accuracy of the navigation system. In addition, if the GNSS data is in the signal domain, a tightly coupled method is adopted for processing. In the IMU processor, GNSS generation with INS integration is implemented to cope with the complex environment and interference that may exist in the signal domain. This ultra-tight coupling method can significantly improve the performance of the navigation system in signal-limited or unstable environments. In the GNSS / INS tightly combined Kalman filter model, the state equation and the observation equation are the core parts of building the model. The state equation describes the dynamic change of the system state over time, while the observation equation establishes the mathematical relationship between the system state and the observable quantity. According to the unique pattern characteristics of the tight combination, after comprehensive analysis, the state equation containing the position error was selected. , speed error , attitude error , Gyro bias bg , Accelerometer bias ba , clock difference He Zhongpiao A total of 17 states, including , are used to construct the state vector. The second state vector is: ;

[0131] Select the ECEF coordinate system as the reference system, and the second state transfer matrix is: ;

[0132] The predicted pseudorange is the estimated value of the distance from the satellite to the receiver during the signal propagation process, and the pseudorange rate is a parameter that describes the rate at which this estimated distance changes over time. GNSS pseudorange data and pseudorange rate are: ;

[0133] In the formula, They are INS pseudorange data and INS pseudorange rate, r i is the launch time coordinate, Receive coordinates for the user, is the coordinate transformation matrix from ECEF coordinate system to I system, is the total pseudorange error, is the coordinate system unit vector, v i is the velocity at the time of launch, The user receives the speed, is the total pseudorange rate error.

[0134] Will and Input to the Kalman filter. Difference data formed by GNSS pseudorange data and pseudorange rate and , that is, the observed value Zk is: .

[0135] In the formula, Indicates the difference between GNSS pseudorange data and INS pseudorange data. Indicates the difference between the GNSS pseudorange rate and the INS pseudorange rate.

[0136] The observation matrix is ​​expressed as: ;

[0137] In the formula, u i Indicates satellite i The unit observation vector to the user.

[0138] Further, after step S4, as Figure 5 As shown, it also includes:

[0139] S51. Receive at least one set of final observation data.

[0140] S52, introducing the residual RES determined by the pseudorange residual or the innovative residual obtained in the Kalman filter model, combined with the azimuth angle AZ and the elevation angle EL of the satellite in the sky, to form auxiliary data in the form of a data tuple of a spherical coordinate system.

[0141] S53. Performing coordinate transformation on the auxiliary data to transform the spherical coordinate system into a Cartesian coordinate system in an earth-fixed coordinate system centered on the earth.

[0142] S54, inputting the auxiliary data after coordinate transformation into a pre-trained convolutional neural network, and correcting the final observation data based on the output data of the convolutional neural network to obtain the final observation data.

[0143] In a specific embodiment, first, according to the position output by GNSS ( X G , Y G , Z G ), the position of INS output ( X I , Y I , Z I ), GNSS output speed ( V xG , V yG , V zG) and the speed output by INS ( V xI , V yI , V zI ) to obtain a set of training data. This set of data includes information such as position, speed, and attitude, which are the original data obtained by the GNSS and INS systems.

[0144] Next, the geographic location data generated by the GNSS receiver and the auxiliary data related thereto are obtained. The residual RES in the auxiliary data can be expressed as a pseudorange residual, or an innovative residual obtained during the Kalman filtering process performed by the GNSS receiver. The satellite direction information should include the specific azimuth AZ and elevation EL of the satellite in the sky when calculating the geographic location. It is worth emphasizing that the present embodiment particularly adopts auxiliary data in the form of data tuples of a spherical coordinate system to represent the azimuth AZ, elevation EL and residual RES. These auxiliary data are used to train convolutional neural networks to predict positioning errors. In this way, the accuracy and reliability of the navigation system can be improved.

[0145] The convolutional neural network model can be used to correct the output of the GNSS receiver during operation. Specifically, the calculated geographic location is corrected by applying the convolutional neural network to the auxiliary data to obtain the positioning error of the calculated geographic location. The positioning error can be used as a correction term for the calculated geographic location and applied to the calculated geographic location to produce a corrected geographic location. Compared to the initially calculated geographic location, the corrected geographic location may be more accurate, meaning that it has a smaller deviation from the actual geographic location, i.e., it has a smaller positioning error relative to the actual geographic location.

[0146] Convolutional neural network is a special feedforward neural network that includes convolution operation. The structure of convolutional neural network mainly includes input layer, convolution layer, pooling layer, fully connected layer and output layer. When building this network, the convolution layer occupies a crucial position. It is composed of multiple convolution kernels superimposed. Convolution kernel, also known as filter, is essentially a matrix or tensor. The convolution operation is to perform inner product calculation between the convolution kernel and the corresponding input data through a set step size. The core principle of convolutional neural network is to extract features at different positions in the input data according to the learning results through multiple convolution kernels to ensure that the extracted data matches the features of the convolution kernel. The convolution calculation formula is expressed as: ;

[0147] In the formula, x n-1 Indicates n -1 layer output, Indicates n Layer iconvolution kernels. * represents convolution operation. is the corresponding bias.

[0148] Convolution is usually the first step in processing data. This step involves performing a dot product operation on the input data with the convolution kernel to extract features. However, convolution alone is not enough to build a strong predictive model. This is because the output of the convolution layer is still linear, which means that its expressive power is limited and it is difficult to capture the complex structures and patterns in the data. To overcome this limitation, the output of the convolution layer must be transformed nonlinearly. This transformation is achieved through the activation function. The activation function is a nonlinear mapping that can map the input data to a higher dimension and introduce nonlinear factors. In this way, the original linear model is transformed into a nonlinear model, and its expressive power is significantly improved. Through this nonlinear transformation, the neural network can capture more complex data features, thereby improving its ability to interpret the input data and the prediction accuracy.

[0149] In the neural network architecture, the pooling layer is usually used as a subsequent step of the convolution layer. After performing the convolution operation, the convolution layer often generates a large number of network parameters. In order to effectively manage these parameters and reduce the computational complexity, the pooling layer plays a key role. Its core function is to reduce the dimension of the parameter matrix, thereby significantly reducing the number of parameters and the amount of computation. Based on this feature, the pooling layer is often called a downsampling layer. Using average pooling, it is expressed as: ;

[0150] In the formula, N Represents the specific length of the input feature information. i The one currently being processed is marked i Pooling windows, j It refers to the first j a specific value.

[0151] There is a close relationship between the fully connected layer and the terminal pooling layer, and the cooperation between them is seamless. Inside the fully connected layer, each neuron establishes a comprehensive and in-depth interactive relationship with all the elements of the previous layer, thus forming an intricate cross-network structure. This structure enables the fully connected layer to deeply integrate and analyze the feature matrix output by the pooling layer in a unique way. During the integration process, the fully connected layer will reshape the feature matrix into a one-dimensional vector according to actual needs. This one-dimensional vector not only contains all the important information of the original data, but also plays a greater role in subsequent calculations.

[0152] The Softmax classifier is generally regarded as an ideal model for the output layer of a neural network. The model is responsible for receiving the output feature data from the last fully connected layer and mapping it to a probability distribution. Through this process, the Softmax classifier can convert the original feature data into a probabilistic form, so that the network output has a clearer meaning. This mapping method makes the sum of the category probabilities represented by each output node 1, ensuring the normalization of the classification results. Therefore, in the classification task, the application of the Softmax classifier can effectively improve the accuracy and reliability of the model prediction, so that the model can more accurately estimate the probability of each category, providing strong support for the final target prediction. Its mathematical expression is: ;

[0153] In the formula, O j It is the final output of the entire convolutional neural network. θ （K） are the parameters of the model.

[0154] Dropout is a regularization technique that aims to address the problem of redundant parameters and scarce samples in convolutional neural networks. In this situation, the network tends to overfit the training data, that is, it shows excellent performance on the training set, but performs poorly when facing a new database. In order to alleviate this overfitting phenomenon, the Dropout layer is introduced and applied to the fully connected layer. By randomly discarding some neurons to reduce the dependencies between them, the generalization ability of the model is enhanced, thereby effectively preventing the occurrence of overfitting. In convolutional neural networks, the loss function is also called the objective function. Since there may be a slight deviation between the predicted value and the target value of the convolutional neural network, in order to optimize the output result and make it closer to the target value, the loss function is used to calculate this deviation, and the parameters in the network are adjusted inversely based on the deviation to achieve the accuracy of the predicted value. The numerical value of the loss function directly reflects the effect of the network model. When the loss function value reaches a low level, it indicates that the model performs well. Common loss functions include cross entropy loss function and square loss function, and their expressions are as follows: ;

[0155] In the formula, N Indicates the number of types, y i represents the target value, t i Represents the predicted value output. Given that the cross entropy loss function can effectively solve the limitations of the square loss function in terms of learning rate, this paper decides to use the cross entropy loss function as the main calculation tool in the selection of loss function. This decision is intended to ensure that the model can be optimized more effectively during the training process, thereby improving the overall performance.

[0156] The training process of a convolutional neural network can be divided into two stages. In the first stage, the forward propagation stage, the network begins to process the input data. This process starts with the convolutional layer, which is responsible for extracting the important features of the input data through a series of convolution operations. These features are converted into feature vectors, which are the basic units that the network can understand and use. These feature vectors are then further processed by the pooling layer. The pooling layer reduces the complexity of the data by reducing the size of the feature vector while retaining the most significant features. After the processing of the convolutional and pooling layers, the data has been transformed into a more abstract and useful form. Next, these processed feature vectors are sent to the fully connected layer. A fully connected layer is a linear layer in a neural network that connects the feature vector to each output category. Here, the feature vector is transformed by a function called Softmax. The Softmax function converts the feature vector into a probability distribution, which represents the probability that the input data belongs to each category. This step is key to classification because it converts a nonlinear problem into a linear problem, allowing the network to output a class probability distribution for each input data. The second stage is the backpropagation stage, the purpose of this stage is to adjust the parameters of the network based on the difference between the network's output and the actual expected result. If the network's prediction does not match the true label, the error between the output and the expected result is calculated. This error is propagated back through the network, starting from the output layer and propagating layer by layer to the input layer. At each layer, the error for that layer is calculated based on the error and the weights of that layer. This error calculation method provides a feedback signal to each weight of the network, telling it how to adjust the weight to reduce the error. This feedback signal is the gradient of the error function with respect to the weight, which is the core of the calculation in the backpropagation algorithm. By using this gradient information, each parameter of the network is updated to produce more accurate predictions in the next round of training. This process is repeated, and each iteration brings the parameters of the network closer to the optimal state that minimizes the prediction error.

[0157] The present invention further optimizes the structure of the convolutional neural network to improve its efficiency and accuracy in processing navigation data. First, a deeper network layer is introduced, and the number of convolutional layers and pooling layers is increased, so that the network can learn more complex feature representations. This helps to capture subtle changes in navigation data, thereby more accurately predicting positioning errors. Secondly, more advanced activation functions, such as ReLU (Rectified Linear Unit) or SELU (Scaled Exponential Linear Unit), are used to enhance the nonlinear expression ability of the network. These activation functions can more effectively handle nonlinear relationships in navigation data and improve the prediction performance of the model. In addition, batch normalization technology is introduced to alleviate the problem of internal covariate shift during network training. By inserting batch normalization layers in the network, the input of each layer can be kept appropriately distributed, thereby accelerating the training process and improving the stability of the model. Finally, more efficient optimization algorithms, such as Adam (Adaptive Moment Estimation) or RMSprop, are used to optimize the weight parameters of the network. These optimization algorithms can automatically adjust the learning rate and accelerate convergence during training, further improving the training efficiency and performance of the model.

[0158] Through the above optimization measures, the convolutional neural network model of the present invention can more accurately predict the positioning error when processing navigation data, and output the corrected geographic location in real time. This helps to improve the accuracy and stability of the navigation system and provide more reliable navigation services for various application scenarios.

[0159] Additionally, an embodiment of the present invention provides a GNSS / INS integrated navigation system, including:

[0160] The GNSS data generation module is used to receive the GNSS pseudo-range data acquired by the GNSS component in the GNSS / INS integrated navigation, and obtain the GNSS data after error correction.

[0161] The INS data generation module is used to receive the IMU signal obtained by the INS component in the GNSS / INS integrated navigation and obtain INS data after processing.

[0162] The data coupling module is used to integrate the GNSS data with the INS data according to different domains of the GNSS data to obtain the GNSS data having at least one degree of integration with the INS data.

[0163] The data filtering module is used to input the GNSS data that forms at least one degree of integration with the INS data into a pre-built Kalman filter model to obtain at least one set of observation data.

[0164] Furthermore, an embodiment of the present invention provides a GNSS / INS combined navigation device, including: a GNSS component, including a GNSS receiver, the GNSS receiver is configured to obtain GNSS pseudorange data from at least one satellite; an INS component, configured to be coupled to the GNSS receiver to generate IMU data; and a control component, configured to execute the GNSS / INS combined navigation method as described above.

[0165] The functions and working principles of these components are described in detail below.

[0166] First up is the GNSS component, the main part of which is the GNSS receiver. This receiver is carefully configured to capture GNSS pseudorange data from at least one satellite. This is achieved through an antenna connected to the receiver, which is responsible for accurately capturing the signal from the GNSS signal source. Once the signal is successfully received, the receiver responds quickly and generates the corresponding GNSS data. This data contains important information about the device's geographic location and time.

[0167] Next is the INS component, which is tightly coupled with the GNSS receiver and is mainly responsible for generating IMU (Inertial Measurement Unit) data. The IMU sensor is the core of the INS component, which can generate accurate IMU signals. After receiving these signals, the INS component can quickly process and generate corresponding IMU data. These data provide key navigation parameters such as the device's attitude, velocity, and acceleration.

[0168] The control component is the "brain" of the entire system, which is responsible for executing the GNSS / INS combined navigation method. In this process, the control component first receives data from the GNSS receiver and the INS component. Then, it efficiently integrates the two types of data in the navigation processing unit of the INS system. This integration ensures precise matching and fusion between the data, thereby improving the accuracy of navigation.

[0169] Based on the integrated IMU data and GNSS data, the control component will further generate accurate and reliable INS data. These data not only meet the high-precision navigation requirements, but also ensure the real-time positioning. In this way, the GNSS / INS combined navigation device provided by the embodiment of the present invention can provide stable and accurate navigation services in various environments.

[0170] In general, the present invention overcomes the limitations of a single navigation technology by combining GNSS and INS navigation technologies, and achieves the goal of maintaining high-precision navigation in complex environments. This combined navigation device not only improves the accuracy and stability of navigation, but also provides a more reliable navigation solution for various application scenarios.

[0171] In addition, the GNSS / INS integrated navigation device also includes: an integral filter configured to receive data from the GNSS receiver and the INS component, and perform fusion processing on the received data through an integral operation.

[0172] The GNSS / INS integrated navigation equipment is also equipped with a key component - the integration filter. This advanced data fusion device is designed to make full use of the data provided by the global navigation satellite system (GNSS) receiver and the inertial navigation system (INS). The integration filter is specially configured to receive and process the dual data sources from the GNSS receiver and the INS component.

[0173] In the process of data fusion, the integral filter performs comprehensive processing and analysis on the data from these two sources through complex integral operations. This processing method not only ensures the comprehensiveness and diversity of the data, but more importantly, it can effectively eliminate or reduce the errors and uncertainties that may exist between multi-source data. Through precise fusion algorithms, the integral filter can generate high-quality information that has been optimized.

[0174] Therefore, the integral filter plays a vital role in the GNSS / INS integrated navigation device. It can not only improve the accuracy and reliability of the data, but also significantly improve the accuracy of navigation and positioning, so that the entire system can maintain excellent performance in complex and changing environments.

[0175] In summary, the embodiments of the present invention provide a GNSS / INS combined navigation method, system and device. The present invention innovatively adopts a convolutional neural network and integrates it into the GNSS / INS combined navigation technology. On the basis of inheriting the essence of existing navigation technology, a convolutional neural network is introduced to achieve more accurate and stable navigation positioning. This technology not only combines the respective advantages of GNSS (Global Navigation Satellite System) and INS (Inertial Navigation System), but also significantly improves the reliability of the application, the accuracy of the data and the robustness of the signal processing with the help of deep learning. Therefore, it can provide users with uninterrupted, no-dead-angle, all-round GNSS / INS combined navigation services, thereby ensuring the accuracy and stability of navigation information.

[0176] Therefore, the present invention has the following advantages and beneficial effects: (1) Improved navigation accuracy and stability. By introducing a convolutional neural network, deep learning and feature extraction of GNSS and INS data are achieved, the navigation information in the data is fully utilized, and the navigation accuracy and stability are improved. (2) The robustness of the navigation system is enhanced. The method of the present invention can adapt to complex environments and dynamically changing situations, has strong robustness to factors such as signal shielding and interference, and ensures the stable operation of the navigation system. (3) The complexity of the navigation system is simplified. By constructing a unified combined navigation model, unified processing and analysis of GNSS and INS data are achieved, the complexity of the navigation system is simplified, and the reliability and maintainability of the system are improved.

[0177] Since the system / device described in the above embodiments of the present invention is a system / device used to implement the method of the above embodiments of the present invention, a person skilled in the art can understand the specific structure and deformation of the system / device based on the method described in the above embodiments of the present invention, and thus will not be described in detail here. All systems / devices used in the method of the above embodiments of the present invention belong to the scope of protection of the present invention.

[0178] It should be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0179] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions.

[0180] It should be noted that the word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention may be implemented by means of hardware comprising a number of different components and by means of a suitably programmed computer. Among the several means listed, several of these means may be embodied by the same hardware. The use of the words first, second, third, etc., is for convenience of expression only and does not indicate any order. These words may be understood as part of the name of the component.

[0181] In addition, it should be noted that, in the description of this specification, the description of the terms "one embodiment", "some embodiments", "embodiment", "example", "specific example" or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0182] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the technical solution of the present invention and its equivalent technology, the present invention should also include these modifications and variations.

Claims

1. A GNSS / INS integrated navigation method, characterized in that: include: Receive GNSS pseudorange data acquired by the GNSS component in the GNSS / INS integrated navigation, and obtain GNSS data after error correction; Receive the IMU signal obtained by the INS component in the GNSS / INS integrated navigation, and obtain INS data after processing; Integrate the GNSS data with the INS data according to different domains of the GNSS data to obtain GNSS data having at least one degree of integration with the INS data; Inputting GNSS data having at least one level of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data; in, Input GNSS data that form at least one degree of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data including: For the generated loosely coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias and accelerometer bias are selected according to the loosely coupled mode characteristics to construct a first state vector; Determine a first state transfer matrix with the Earth-centered Earth-fixed coordinate system as a reference reference system; Based on the constructed first state vector and the first state transfer matrix, a first Kalman filter model is obtained; Inputting the generated loosely coupled GNSS data with INS integration function into the first Kalman filter model to obtain observation data; wherein the input loosely coupled GNSS data includes the difference between the position and velocity data output by the GNSS component and the INS component; For the generated tightly coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias, accelerometer bias, clock error and clock drift are selected according to the tightly coupled mode characteristics to construct the second state vector; Determine the second state transfer matrix with the ECEF coordinate system as the reference system; Based on the constructed second state vector and the second state transfer matrix, a second Kalman filter model is obtained; Inputting the generated tightly coupled GNSS data with INS integration function into the second Kalman filter model to obtain observation data; wherein the input tightly coupled GNSS data includes difference data formed by GNSS pseudorange data and pseudorange rate; The first state vector is: ； In the formula, is the position error, is the speed error, is the attitude error, b g is the gyroscope bias, b a is the accelerometer zero bias; The first state transfer matrix is: ； In the formula, is the Earth's rotation angular velocity, τ s is the inertial navigation update period, I 3 and O 3 represent the unit matrix and the zero matrix respectively, is the direction cosine matrix from the carrier system to the ECEF system, is the inertial navigation output specific force, is the gravitational acceleration term, where For distance, is the ECEF position, is the satellite position, is the acceleration noise; The position and velocity data output by the GNSS component and the INS component are: ； In the formula, ( X G , Y G , Z G )and( X I , Y I , Z I ) represent the position output by GNSS and the position output by INS respectively, ( V xG , V yG , V zG )and( V xI , V yI , V zI ) represent the speed output by GNSS and the speed output by INS respectively; The second state vector is: ； In the formula, For the clock error, For the clock drift; The second state transfer matrix is: ； GNSS pseudorange data and pseudorange rate are: ； In the formula, are INS pseudorange data and INS pseudorange rate, respectively. r i is the launch time coordinate, Receive coordinates for the user, is the coordinate transformation matrix from ECEF coordinate system to I system, is the total pseudorange error, is the coordinate system unit vector, v i is the velocity at the time of launch, The user receives the speed, is the sum of pseudorange rate errors; The difference data formed by GNSS pseudorange data and pseudorange rate is: ； In the formula, Indicates the difference between GNSS pseudorange data and INS pseudorange data. Indicates the difference between the GNSS pseudorange rate and the INS pseudorange rate.

2. The GNSS / INS integrated navigation method according to claim 1, characterized in that: Receive the GNSS pseudorange data obtained by the GNSS component in the GNSS / INS integrated navigation, and obtain the GNSS data after error correction, including: According to the propagation time of the satellite signal received by the GNSS component, an approximate value of the geometric distance between the GNSS component and the satellite is obtained as GNSS pseudorange data; Identify and quantify various error sources including satellite and GNSS component clock errors, ionospheric delays, and tropospheric delays; Based on various error sources, error correction is performed on the received GNSS pseudorange data to obtain error-corrected pseudorange measurement values; The user's three-dimensional position information and GNSS component clock error are obtained by solving data from at least four satellites; Generate GNSS data based on the user's three-dimensional position information, GNSS component clock error and corrected GNSS pseudorange data.

3. The GNSS / INS integrated navigation method according to claim 2, characterized in that: The GNSS pseudorange data is: ； In the formula, n Represents the satellite number, r n Represents the geometric distance between the satellite and the GNSS component, δt u is the clock error of the user receiver, δt n is the satellite clock error, I n It is the delay caused by the ionosphere. T n It is the delay caused by the troposphere. ε n The delay is caused by random disturbance factors; The error-corrected pseudorange measurement is: ； Geometric distances between satellites and GNSS components r n for: ； In the formula, and Satellite n and the user's position vector.

4. The GNSS / INS integrated navigation method according to claim 1, characterized in that: According to different domains of GNSS data, GNSS data is integrated with INS data to obtain GNSS data that forms at least one degree of integration with INS data, including: Identify and determine the data domain where the GNSS data is located; When GNSS data is in the position domain, a loosely coupled method integrated with INS is adopted to integrate INS data with GNSS data to generate loosely coupled GNSS data with INS integration function; When GNSS data is in the measurement domain, a tightly coupled method is adopted for integration with INS to generate tightly coupled GNSS data with INS integration function.

5. The GNSS / INS integrated navigation method according to any one of claims 1 to 4, characterized in that: After inputting GNSS data that forms at least one level of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data, the method further includes: receiving at least one set of final observation data; A residual RES determined by a pseudorange residual or an innovative residual obtained in a Kalman filter model is introduced, combined with an azimuth angle AZ and an elevation angle EL of the satellite in the sky, to form auxiliary data in the form of a data tuple of a spherical coordinate system; by performing a coordinate transformation on the auxiliary data to transform the spherical coordinate system into a Cartesian coordinate system in an earth-fixed coordinate system centered on the earth; The auxiliary data after coordinate transformation is input into a pre-trained convolutional neural network, and the final observation data is corrected based on the output data of the convolutional neural network to obtain the final observation data.

6. A GNSS / INS integrated navigation system, characterized in that: include: The GNSS data generation module is used to receive the GNSS pseudo-range data acquired by the GNSS component in the GNSS / INS integrated navigation and obtain the GNSS data after error correction; INS data generation module, used to receive the IMU signal acquired by the INS component in the GNSS / INS integrated navigation, and obtain INS data after processing; A data coupling module, for integrating the GNSS data with the INS data according to different domains of the GNSS data, to obtain GNSS data having at least one degree of integration with the INS data; A data filtering module, used for inputting GNSS data having at least one level of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data; in, Input GNSS data that form at least one degree of integration with INS data into a pre-built Kalman filter model to obtain at least one set of observation data including: For the generated loosely coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias and accelerometer bias are selected according to the loosely coupled mode characteristics to construct a first state vector; Determine a first state transfer matrix with the Earth-centered Earth-fixed coordinate system as a reference reference system; Based on the constructed first state vector and the first state transfer matrix, a first Kalman filter model is obtained; Inputting the generated loosely coupled GNSS data with INS integration function into the first Kalman filter model to obtain observation data; wherein the input loosely coupled GNSS data includes the difference between the position and velocity data output by the GNSS component and the INS component; For the generated tightly coupled GNSS data with INS integration function, variables including position error, velocity error, attitude error, gyroscope bias, accelerometer bias, clock error and clock drift are selected according to the tightly coupled mode characteristics to construct the second state vector; Determine the second state transfer matrix with the ECEF coordinate system as the reference system; Based on the constructed second state vector and the second state transfer matrix, a second Kalman filter model is obtained; Inputting the generated tightly coupled GNSS data with INS integration function into the second Kalman filter model to obtain observation data; wherein the input tightly coupled GNSS data includes difference data formed by GNSS pseudorange data and pseudorange rate; The first state vector is: ； In the formula, is the position error, is the speed error, is the attitude error, b g is the gyroscope bias, b a is the accelerometer zero bias; The first state transfer matrix is: ； In the formula, is the Earth's rotation angular velocity, τ s is the inertial navigation update period, I 3 and O 3 represent the unit matrix and the zero matrix respectively, is the direction cosine matrix from the carrier system to the ECEF system, is the inertial navigation output specific force, is the gravitational acceleration term, where For distance, is the ECEF position, is the satellite position, is the acceleration noise; The position and velocity data output by the GNSS component and the INS component are: ； In the formula, ( X G , Y G , Z G )and( X I , Y I , Z I ) represent the position output by GNSS and the position output by INS respectively, ( V xG , V yG , V zG )and( V xI , V yI , V zI ) represent the speed output by GNSS and the speed output by INS respectively; The second state vector is: ； In the formula, For the clock error, For the clock drift; The second state transfer matrix is: ； GNSS pseudorange data and pseudorange rate are: ； In the formula, are INS pseudorange data and INS pseudorange rate, respectively. r i is the launch time coordinate, Receive coordinates for the user, is the coordinate transformation matrix from ECEF coordinate system to I system, is the total pseudorange error, is the coordinate system unit vector, v i is the velocity at the time of launch, The user receives the speed, is the sum of pseudorange rate errors; The difference data formed by GNSS pseudorange data and pseudorange rate is: ； In the formula, Indicates the difference between GNSS pseudorange data and INS pseudorange data. Indicates the difference between the GNSS pseudorange rate and the INS pseudorange rate.

7. A GNSS / INS integrated navigation device, characterized in that: include: A GNSS component including a GNSS receiver configured to acquire GNSS pseudorange data from at least one satellite; an INS component configured to be coupled to a GNSS receiver for generating IMU data; A control component configured to execute the GNSS / INS integrated navigation method as described in any one of claims 1 to 5.

8. The GNSS / INS integrated navigation device according to claim 7, characterized in that: include: The integration filter is configured to receive data from the GNSS receiver and the INS component and perform fusion processing on the received data through integration operation.

Images