Inertial navigation attitude resolving method and device based on sensor fusion

Abstract

The invention provides an inertial navigation attitude resolving method and device based on sensor fusion. The method comprises the following steps: carrying out multi-clock domain synchronous calibration processing on a double-antenna GNSS system, an IMU inertial measurement unit and an antenna servo system in a vehicle-mounted communication-in-motion system; according to the time synchronization reference, carrying out quality evaluation and weight distribution processing on the GNSS satellite signal under the satellite communication in motion antenna pointing constraint; and carrying out multi-sensor fusion resolving processing on the inertial attitude of the vehicle body according to the time synchronization reference and the GNSS observation data weight distribution result. According to the method, millisecond-level time synchronization is realized by establishing a robust extended Kalman filtering model of a multi-dimensional extended state vector, GNSS signal quality is optimized by adopting an adaptive weight distribution strategy of antenna pointing constraint, and attitude fusion precision is enhanced by utilizing high-precision angle feedback of an antenna servo system. The problem that a traditional attitude resolving method in a vehicle-mounted communication-in-motion system is not high in precision is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to an inertial navigation attitude solving method and device based on sensor fusion. BACKGROUND

[0002] With the wide application of vehicle-mounted mobile communication systems in modern transportation and communication fields, the accurate acquisition of vehicle attitude information has become a key technical requirement to ensure mobile communication quality. Traditional inertial navigation attitude solving methods usually adopt GNSS and IMU loose coupling or tight coupling fusion technology to realize the joint estimation of position, speed and attitude through Kalman filtering algorithm. These methods can provide better navigation accuracy in open environments, but still have obvious technical limitations in complex application scenarios.

[0003] The main problem of the prior art is the lack of special design for the special working environment of the vehicle-mounted mobile communication system, especially the neglect of the multiple influences of the mobile communication antenna servo system on the navigation sensors. The mechanical vibration generated by the mobile communication antenna during satellite tracking will affect the measurement accuracy of the IMU, the change of the antenna pointing direction will cause occlusion and multipath interference to the GNSS signal reception, and the clock desynchronization problem between the subsystems will cause time alignment error during data fusion. The combined effect of these factors makes the precision and reliability of the traditional attitude solving method significantly decreased in the application of vehicle-mounted mobile communication. SUMMARY

[0004] The main purpose of the present application is to solve the technical problems of low precision and insufficient reliability of the existing inertial navigation attitude solving method in the application of vehicle-mounted mobile communication system. The first aspect of the present application provides an inertial navigation attitude solving method based on sensor fusion, which comprises: Performing multi-clock domain synchronization calibration processing on the dual-antenna GNSS system, IMU inertial measurement unit and antenna servo system in the vehicle-mounted mobile communication system to obtain a multi-clock domain time synchronization reference; According to the multi-clock domain time synchronization reference, performing quality evaluation and weight distribution processing on the GNSS satellite signal under the constraint of the mobile communication antenna pointing direction to obtain the GNSS observation data weight distribution result optimized by the antenna pointing direction constraint; According to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result, performing multi-sensor fusion solving processing on the vehicle body inertial attitude to obtain the vehicle body three-axis attitude angle solving result.

[0005] Optionally, in the first implementation of the first aspect of the present application, the multi-clock domain synchronization calibration processing on the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system comprises: combining the time synchronization error between the GNSS receiver and the IMU inertial measurement unit with the multi-dimensional navigation state variable of the IMU inertial measurement unit to obtain a multi-dimensional extended state vector containing the time synchronization error; performing linearization processing on the position and velocity measurement values of the dual-antenna GNSS system according to the multi-dimensional extended state vector by second-order Taylor series expansion to obtain linearized measurement equations considering the influence of the time synchronization error; filtering and estimating the multi-dimensional extended state vector by using a robust extended Kalman filter according to the linearized measurement equations to obtain real-time time synchronization error estimation values; performing time offset compensation processing on the observation data of the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system according to the time synchronization error estimation values, and verifying the synchronization effect to obtain the multi-clock domain time synchronization reference.

[0006] Optionally, in the second implementation of the first aspect of the present application, the linearization processing on the position and velocity measurement values of the dual-antenna GNSS system according to the multi-dimensional extended state vector to obtain linearized measurement equations considering the influence of the time synchronization error comprises: performing second-order Taylor series expansion calculation processing on the position measurement values of the dual-antenna GNSS system in the presence of time delay according to the time synchronization error state variable in the multi-dimensional extended state vector to obtain a position expansion expression containing a time delay influence term; performing second-order Taylor series expansion calculation processing on the velocity measurement values of the dual-antenna GNSS system in the presence of time delay according to the position expansion expression to obtain a velocity expansion expression containing a time delay influence term; ignoring the second-order and higher-order terms and retaining the first-order time delay influence term according to the position expansion expression and the velocity expansion expression to obtain linearized position and velocity approximate expressions; constructing a linear relationship matrix between the state variables and the observation values according to the linearized position and velocity approximate expressions to obtain linearized measurement equations considering the influence of the time synchronization error.

[0007] Optionally, in a third implementation form of the first aspect of the present application, the quality evaluation and weight distribution processing of the GNSS satellite signals under the constraint of the antenna pointing of the moving-target indication antenna according to the multi-clock domain time synchronization reference comprises: time calibration processing of the received GNSS satellite signals according to the multi-clock domain time synchronization reference, to obtain time-synchronized GNSS satellite signals; antenna pointing influence analysis processing of the time-synchronized GNSS satellite signals according to the real-time pointing parameters of the moving-target indication antenna, to obtain expected reception quality evaluation results of each satellite signal; quality classification and attenuation compensation processing of the GNSS satellite signals according to the expected reception quality evaluation results, to obtain quality-corrected GNSS satellite signals; observation data extraction and weight distribution processing according to the quality-corrected GNSS satellite signals, to obtain the GNSS observation data weight distribution results optimized under the constraint of the antenna pointing.

[0008] Optionally, in a fourth implementation form of the first aspect of the present application, the quality classification and attenuation compensation processing of the GNSS satellite signals according to the expected reception quality evaluation results, to obtain quality-corrected GNSS satellite signals comprises: quantitative classification processing of the carrier-to-noise ratio and signal strength of each GNSS satellite signal according to the expected reception quality evaluation results, to obtain classified satellite signal quality levels; calculation of the attenuation amount of each satellite signal caused by the antenna shielding and multipath effect according to the satellite signal quality levels and the pointing parameters of the moving-target indication antenna, to obtain signal attenuation compensation coefficients related to the antenna pointing; robust factor weight reduction processing of the GNSS satellite signals with low quality levels and weight enhancement processing of the signals with high quality levels according to the signal attenuation compensation coefficients, to obtain signal weight coefficients adjusted by quality classification; adaptive adjustment processing of the observation noise covariance of each GNSS satellite signal according to the signal weight coefficients, to suppress the influence of low-quality signals on the navigation solution, to obtain quality-corrected GNSS satellite signals.

[0009] Optionally, in a fifth implementation form of the first aspect of the present application, the multi-sensor fusion solution processing of the vehicle body inertial attitude according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution results comprises: According to the multi-clock domain time synchronization reference, angle feedback information extraction processing is performed on an angle encoder and a gyro-stabilized platform of the moving target indication antenna servo system, to obtain external attitude reference data time-aligned with GNSS observation data; According to the external attitude reference data and a three-dimensional space position offset vector between an antenna mounting point of the antenna in the moving target indication antenna servo system and a measurement center of an IMU inertial measurement unit, coordinate system conversion processing is performed on an angle measurement value of the antenna, to obtain equivalent attitude information at a vehicle body gravity center; According to the equivalent attitude information and the GNSS observation data weight distribution result, multi-sensor fusion solving processing is performed on IMU inertial measurement unit data, to obtain a fusion attitude estimation result; According to the fusion attitude estimation result, consistency verification processing is performed on tracking historical trajectory data of the antenna, to obtain a vehicle body three-axis attitude angle solving result.

[0010] Optionally, in the sixth implementation manner of the first aspect of the present application, the multi-sensor fusion solving processing on the IMU inertial measurement unit data according to the equivalent attitude information and the GNSS observation data weight distribution result to obtain the fusion attitude estimation result comprises: According to the equivalent attitude information, data alignment processing is performed on angle feedback data of the moving target indication antenna servo system and three-axis attitude angles of the IMU inertial measurement unit, to obtain multi-source attitude measurement values at the same time; According to the multi-source attitude measurement values and the GNSS observation data weight distribution result, fusion weight coefficients of each sensor attitude data are calculated, to obtain weight parameters of multi-sensor attitude fusion; According to the weight parameters, weight-based attitude matrix fusion calculation processing is performed on the attitude angles of the IMU inertial measurement unit and the attitude feedback of the moving target indication antenna servo system, to obtain a fused attitude rotation matrix; According to the fused attitude rotation matrix, rotation matrix to Euler angle conversion calculation processing is performed, to obtain the fusion attitude estimation result.

[0011] The second aspect of the present application provides a sensor fusion-based inertial navigation attitude solving device, which comprises: A time synchronization module is configured to perform multi-clock domain synchronization calibration processing on a dual-antenna GNSS system, an IMU inertial measurement unit and an antenna servo system in a vehicle-mounted moving target indication system, to obtain a multi-clock domain time synchronization reference; A signal optimization module is configured to perform quality evaluation and weight distribution processing on GNSS satellite signals under moving target indication antenna pointing constraints according to the multi-clock domain time synchronization reference, to obtain GNSS observation data weight distribution results optimized under antenna pointing constraints. A pose fusion module is configured to perform multi-sensor fusion calculation on the vehicle body inertial pose according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result, to obtain a vehicle body three-axis pose angle calculation result.

[0012] The above-mentioned inertial navigation pose calculation method and device based on sensor fusion perform multi-clock domain synchronization calibration processing on the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted moving channel system; perform quality evaluation and weight distribution processing on the GNSS satellite signal under the moving channel antenna pointing constraint according to the time synchronization reference; and perform multi-sensor fusion calculation on the vehicle body inertial pose according to the time synchronization reference and the GNSS observation data weight distribution result. The present application realizes millisecond-level time synchronization by establishing a robust extended Kalman filter model of a multi-dimensional extended state vector, optimizes the GNSS signal quality by using an adaptive weight distribution strategy of antenna pointing constraint, enhances the pose fusion accuracy by using high-precision angle feedback of the antenna servo system, and solves the problem of low accuracy of the traditional pose calculation method in the vehicle-mounted moving channel system.

[0013] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application will be realized and achieved by the structure particularly pointed out in the description, claims and drawings.

[0014] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the following preferred embodiments are specifically described below, and the accompanying drawings are described in detail as follows. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The first embodiment of the inertial navigation pose calculation method based on sensor fusion in the embodiments of the present application is shown in the figure. Figure 2 The first embodiment of the inertial navigation pose calculation method based on sensor fusion in the embodiments of the present application is shown in the figure. DETAILED DESCRIPTION

[0016] In order to make the objects, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions of the present application will be described in detail below with reference to the drawings. Obviously, the described embodiments are part of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0017] The terms "comprising" and "having" and any variations thereof used in the embodiments of the present application are intended to cover a non-exclusive inclusion. For example, a process, method, article, or apparatus that comprises a list of steps or units is not limited to the listed steps or units but can optionally further include other steps or units not listed or can optionally further include steps or units inherent to such process, method, article, or apparatus.

[0018] To facilitate the understanding of the present embodiments, first, a sensor fusion-based inertial navigation attitude calculation method disclosed by the embodiments of the present application is introduced in detail. As shown in the following figure, the method comprises the following steps: Figure 1 S101, the double-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted dynamic communication system are subjected to multi-clock domain synchronization calibration processing to obtain a multi-clock domain time synchronization reference; In an embodiment of the present application, the multi-clock domain synchronization calibration processing of the double-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted dynamic communication system to obtain the multi-clock domain time synchronization reference comprises: taking the time synchronization error between the GNSS receiver and the IMU inertial measurement unit as a state variable, combining it with the multi-dimensional navigation state variable of the IMU inertial measurement unit to obtain a multi-dimensional extended state vector containing the time synchronization error; according to the multi-dimensional extended state vector, performing linearization processing of the position and velocity measurement values of the double-antenna GNSS system by second-order Taylor series expansion to obtain a linearized measurement equation considering the influence of the time synchronization error; according to the linearized measurement equation, filtering and estimating the multi-dimensional extended state vector by using a robust extended Kalman filter to obtain a real-time time synchronization error estimate value; according to the time synchronization error estimate value, performing time offset compensation processing on the observation data of the double-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system, and verifying the synchronization effect to obtain the multi-clock domain time synchronization reference.

[0019] ​Specifically, in on-board on-the-move communication applications, the core of multi-clock domain synchronization calibration lies in establishing an extended state-space model that includes time synchronization error. First, the time synchronization error between the GNSS receiver and the IMU (Inertial Measurement Unit) is used as a 22nd-dimensional state variable and combined with the IMU's existing 21-dimensional navigation state variables. These 21-dimensional navigation state variables specifically include 3D position error, 3D velocity error, 3D attitude error (roll error, pitch error, and yaw error) in the northeast celestial coordinate system, triaxial gyroscope bias, triaxial accelerometer bias, triaxial gyroscope scale factor error, and triaxial accelerometer scale factor error. The introduction of the time synchronization error state variable enables a filter to estimate and compensate for the clock offset between the GNSS and IMU in real time, thereby resolving the problem of multi-clock domain desynchronization. In the extended 22-dimensional state vector, the time synchronization error is modeled as a random constant process, whose dynamic characteristics are described by process noise. This modeling approach effectively captures the slowly varying characteristics of clock drift.

[0020] Based on the multidimensional extended state vector, the dual-antenna GNSS position and velocity measurements require linearization using a second-order Taylor series expansion. Due to time synchronization errors between the timestamps of the GNSS observations and the IMU data, the true GNSS observation at time t corresponds to the physical state at time t minus the time synchronization error. A Taylor series expansion is a mathematical tool that approximates a function near a point using a polynomial. By performing a second-order Taylor series expansion near the standard observation time, the position observation consists of the position itself, the velocity multiplied by the time delay, and half of the acceleration multiplied by the squared time delay. The velocity observation consists of the velocity itself, the acceleration multiplied by the time delay, and half of the rate of change of acceleration multiplied by the squared time delay. Because the magnitude of the time synchronization error is relatively small, the second-order and higher-order terms have little impact on the observations. Therefore, these higher-order terms are ignored in the linearization process, retaining only the first-order time delay term. This simplification ensures computational efficiency while maintaining sufficient modeling accuracy. Through linearization processing, the original nonlinear observation equation about time delay is converted into a linear form about the extended state vector, and a linearized measurement equation considering the influence of time synchronization error is constructed.

[0021] Robust extended Kalman filter is used to estimate the multi-dimensional extended state vector, and the real-time estimation of time synchronization error is realized. Robust extended Kalman filter is an improved algorithm developed on the basis of traditional extended Kalman filter, and its core feature is to introduce a robust mechanism to suppress the influence of abnormal observation values. In the time update stage, the filter predicts the 22-dimensional extended state vector according to the state transition equation, and the time synchronization error state keeps its estimated value at the previous time, because the error is modeled as a slowly changing random constant. In the measurement update stage, the filter calculates the innovation sequence using the linearized measurement equation, and the innovation reflects the difference between the predicted observation value and the actual observation value. The robust mechanism dynamically adjusts the robust factor by analyzing the statistical characteristics of the innovation. When the innovation exceeds the preset threshold, the robust factor automatically reduces the weight of the corresponding observation value, thereby reducing the interference of abnormal data on the estimation of time synchronization error. The recursive estimation process of the filter continues, and the current time synchronization error estimate is output at each time, which quantifies the time offset of GNSS observation data relative to IMU data.

[0022] According to the real-time time synchronization error estimate, the observation data of the double-antenna GNSS, IMU inertial measurement unit and antenna servo sub-devices are time offset compensated. The basic principle of compensation is to add the estimated time synchronization error directly to the time stamp of the corresponding data, so that the data of different sub-devices are accurately aligned in the time dimension. Specifically, the time stamp of IMU data is added to the time synchronization error estimate to align with the time stamp of GNSS data, and the time compensation of antenna servo data is adjusted according to the relative relationship between its clock and GNSS clock. In order to verify the effectiveness of the synchronization effect, a two-factor detection mechanism is used for quantitative evaluation. The synchronization error factor directly evaluates the current synchronization accuracy level by calculating the absolute value of the time synchronization error estimate, and the error consistency factor evaluates the stability and consistency of synchronization by analyzing the standard deviation of the time synchronization error sequence in the sliding time window. When the synchronization error factor is less than the set accuracy threshold and the error consistency factor is less than the set stability threshold, it indicates that the time synchronization has reached the required accuracy and stability standards. At this time, the multi-clock domain time synchronization reference is formally established, which provides a unified time reference framework for each sub-device and ensures the accurate synchronization of all sensor data in the time dimension.

[0023] Further, the linearization processing of the second-order Taylor series expansion of the position and velocity measurement values of the dual-antenna GNSS system according to the multi-dimensional extended state vector to obtain the linearized measurement equation considering the influence of the time synchronization error includes: performing the second-order Taylor series expansion calculation processing of the position measurement value of the dual-antenna GNSS system in the presence of time delay according to the time synchronization error state variable in the multi-dimensional extended state vector to obtain a position expansion expression containing a time delay influence term; performing the second-order Taylor series expansion calculation processing of the velocity measurement value of the dual-antenna GNSS system in the presence of time delay according to the position expansion expression to obtain a velocity expansion expression containing a time delay influence term; according to the position expansion expression and the velocity expansion expression, ignoring the high-order terms of the second order and above, retaining the first-order time delay influence term, and obtaining the linearized position and velocity approximate expression; according to the linearized position and velocity approximate expression, constructing a linear relationship matrix between the state variable and the observation value to obtain the linearized measurement equation considering the influence of the time synchronization error.

[0024] Specifically, in the second-order Taylor series expansion calculation processing of the dual-antenna GNSS position measurement value, the time synchronization error state variable needs to be extracted from the multi-dimensional extended state vector as the basis parameter for expansion. The time synchronization error state variable represents the time deviation between the GNSS observation time and the IMU measurement time, which causes the GNSS observation value at the nominal time t to actually reflect the true physical state at time t minus the time synchronization error. The second-order Taylor series expansion is a mathematical approximation method that simplifies complex function relationships by expressing functions as polynomials near a reference point. In the processing of position measurement values, the position function with time delay is expanded near the standard observation time, and the expansion expression contains three main terms: the zero-order term represents the position value itself at the standard time, the first-order term represents the contribution of the velocity multiplied by the time delay, and the second-order term represents the influence of the acceleration multiplied by the time delay squared divided by two. This expansion method can quantitatively describe the specific influence of time delay on position observation values, where the first-order term reflects the linear relationship between the carrier motion velocity and the time deviation, and the second-order term considers the quadratic influence of carrier acceleration motion on position deviation. Through this mathematical processing, the time delay problem that is difficult to directly process is transformed into a polynomial expression form related to the time synchronization error, thereby obtaining a position expansion expression containing a time delay influence term.

[0025] Based on the mathematical structure of the position expansion expression, the dual-antenna GNSS velocity measurement is also calculated by a similar second-order Taylor series expansion. The expansion process of the velocity measurement follows similar mathematical principles as the position, but the physical meaning is different. The zero-order term of the velocity expansion expression represents the velocity value at the standard time, the first-order term embodies the contribution of the acceleration multiplied by the time delay, and the second-order term reflects the influence of the acceleration rate of change (i.e., jerk) multiplied by the time delay squared and divided by two. The acceleration rate of change describes the trend of the carrier acceleration changing over time, which is mainly caused by the acceleration or deceleration operation of the vehicle in vehicle dynamic applications. The establishment of the velocity expansion expression quantitatively describes the influence of the time delay on the velocity observation value, which converts the complex time-domain coupling problem into an easy-to-handle algebraic relationship. It is worth noting that the coefficients in the velocity expansion expression are closely related to the motion state of the carrier. When the carrier is in uniform linear motion, the influence of the first-order term is relatively small, while in variable speed or turning motion, the contribution of the first-order term becomes significant.

[0026] After obtaining the complete expansion expressions of position and velocity, the second-order and higher-order terms need to be ignored. The theoretical basis for this simplification is the magnitude characteristics of the time synchronization error. In actual vehicle dynamic applications, the typical value of the time synchronization error is in the order of milliseconds, which is a small amount relative to the observation period. When the time synchronization error is small, the numerical values of the second-order, third-order, and higher-order terms become extremely small, and their influence on the observation value can be ignored. At the same time, the acceleration and jerk of the carrier also remain within a reasonable range under normal operating conditions, further reducing the contribution of the higher-order terms. By retaining the first-order time delay influence term and ignoring the higher-order terms, the calculation efficiency of the mathematical model is ensured, and sufficient modeling accuracy is maintained. This processing method converts the original nonlinear expansion expression into a linear form with respect to the time synchronization error. The linearized approximation expression of the position contains the position value itself and the linear term of the velocity multiplied by the time delay, and the linearized approximation expression of the velocity contains the velocity value itself and the linear term of the acceleration multiplied by the time delay. The key advantage of linearization is that the subsequent filtering estimation algorithm can be solved using linear algebra methods, avoiding the complexity and computational burden of nonlinear optimization.

[0027] According to the linearized position and velocity approximation expressions, constructing the linear relationship matrix between the state variables and the observations becomes the core step of the entire linearized measurement equation establishment. The linear relationship matrix is also called the observation matrix or the measurement matrix, and its role is to establish an explicit mathematical mapping relationship between the internal state variables and the external observations. In the construction process, the coefficients of the linearized position and velocity expressions need to be extracted and the matrix arranged according to the order of the state variables. For each state component in the 22-dimensional extended state vector, the partial derivative of its position and velocity observations is calculated, and these partial derivatives constitute the corresponding elements of the linear relationship matrix. The matrix elements corresponding to the time synchronization error state variable are particularly important because they directly reflect the degree of influence of the time delay on the observations. Specifically, the influence coefficient of the time synchronization error on the position observation is equal to the current velocity estimate, and the influence coefficient on the velocity observation is equal to the current acceleration estimate. The influence coefficients of other state variables such as position error, velocity error, and attitude error on the observations are calculated according to the standard formula of inertial navigation. Through this systematic matrix construction process, the linearized measurement equation considering the influence of the time synchronization error is finally obtained, which establishes a linear mathematical relationship between all state variables in the extended state vector and the position and velocity observations, and provides a complete mathematical model for the measurement update step of the robust extended Kalman filter.

[0028] In S102, according to the multi-clock domain time synchronization reference, quality evaluation and weight distribution processing of GNSS satellite signals under the constraint of moving target indication antenna pointing are performed to obtain the weight distribution result of GNSS observation data optimized under the constraint of antenna pointing. In an embodiment of the present application, the quality evaluation and weight distribution processing of GNSS satellite signals under the constraint of moving target indication antenna pointing according to the multi-clock domain time synchronization reference to obtain the weight distribution result of GNSS observation data optimized under the constraint of antenna pointing includes: performing time mark calibration processing on the received GNSS satellite signals according to the multi-clock domain time synchronization reference to obtain time-synchronized GNSS satellite signals; performing antenna pointing influence analysis processing on the time-synchronized GNSS satellite signals according to the real-time pointing parameters of the moving target indication antenna to obtain the expected reception quality evaluation result of each satellite signal; performing quality classification and attenuation compensation processing on the GNSS satellite signals according to the expected reception quality evaluation result to obtain the quality-corrected GNSS satellite signals; and performing observation data extraction and weight distribution processing according to the quality-corrected GNSS satellite signals to obtain the weight distribution result of GNSS observation data optimized under the constraint of antenna pointing.

[0029] Specifically, the implementation process of time mark calibration processing of the received GNSS satellite signals based on the multi-clock domain time synchronization reference involves accurate correction of the original signal timestamps. The received GNSS satellite signals carry time information of the transmission time, but due to the deviation of the internal clock of the receiver from the standard time, these time marks need to be calibrated according to the established time synchronization reference. The calibration process first reads the original timestamp data of the GNSS receiver, and then applies the time offset in the multi-clock domain time synchronization reference to these original timestamps. The specific calibration calculation involves directly adding or subtracting the correction value in the time synchronization reference to the original timestamp, and the specific symbol depends on the direction of the clock deviation. The accuracy of the time mark calibration directly affects the accuracy of the subsequent processing steps, because accurate time information is the basis for signal quality assessment and data fusion. After the calibration process, the time marks of all GNSS satellite signals are unified under the same time reference framework, eliminating the time errors caused by clock asynchronization. This unified time mark enables the satellite signals received at different times to be accurately matched in time with the pointing state of the SATCOM antenna. The GNSS satellite signals after time synchronization not only retain the original signal characteristics, but also have accurate time marks, which are strictly consistent with the multi-clock domain time synchronization reference.

[0030] According to the real-time pointing parameters of the moving satellite antenna, the antenna pointing influence analysis and processing of the time-synchronized GNSS satellite signals need to establish the quantitative relationship between the antenna pointing state and the signal reception quality. The real-time pointing parameters of the moving satellite antenna include two key angle information of azimuth and elevation. These angle parameters describe the pointing direction of the antenna main axis in the three-dimensional space. The core of the antenna pointing influence analysis is to calculate the angle relationship between the antenna main beam direction and the direction of each GNSS satellite, because this angle relationship directly determines the antenna reception gain of different satellite signals. When the GNSS satellite is located in the antenna main beam direction, the signal reception quality is best, and when the satellite is located in the antenna side lobe or back lobe direction, the signal will be obviously attenuated. In the analysis and processing process, first, the azimuth and elevation of the satellite relative to the receiver are calculated according to the orbit parameters of each GNSS satellite and the current time, and then these angle information are compared and calculated with the pointing parameters of the moving satellite antenna. The result of the comparison and calculation is the deviation angle of each satellite relative to the antenna main axis, which is an important indicator to evaluate the signal reception quality. By querying the antenna directivity pattern or using the preset gain function, the deviation angle is converted into the corresponding signal gain value or attenuation value. Different deviation angles correspond to different signal reception quality levels, the smaller the deviation angle, the higher the expected signal reception quality, and vice versa. The result of this analysis and processing is the expected reception quality evaluation result of each satellite signal, which quantitatively describes the specific influence degree of the moving satellite antenna pointing state on the reception performance of each GNSS satellite signal.

[0031] Based on the expected reception quality evaluation results, the GNSS satellite signals are subjected to quality grading and attenuation compensation processing to achieve standardization and optimization of signal quality. Quality grading is to divide each satellite signal into different quality levels according to the expected reception quality evaluation results, and the grading standard is based on comprehensive indicators such as signal strength, carrier-to-noise ratio, and multipath influence degree. The quality level division adopts a threshold judgment method, and the signal quality is divided into multiple levels such as excellent, good, general, and poor, each level corresponding to a specific quality indicator range. The signal of excellent level indicates that its reception quality is almost not affected by the antenna pointing, and the signal strength is high and stable; the signal of good level is slightly affected by the antenna pointing, but can still provide reliable observation data; the signal of general level is obviously affected by the antenna pointing, and the signal quality decreases but can still be used; the signal of poor level is severely affected by the antenna pointing, and the signal quality is low and needs special processing. Attenuation compensation processing is to correct the numerical value of the signal affected by the antenna pointing, and the compensation principle is to correct the signal strength in the reverse direction according to the known antenna gain characteristics. For satellite signals in the antenna side lobe or back lobe direction, the corresponding attenuation amount is calculated according to the deviation angle and the antenna gain pattern, and then this attenuation amount is used as a compensation coefficient applied to the signal carrier-to-noise ratio or signal strength indicator. Compensation processing makes the satellite signals received at different pointing angles have numerical comparability, reducing the influence of antenna pointing changes on signal quality evaluation. After quality grading and attenuation compensation processing, the quality corrected GNSS satellite signals are obtained, and the quality indicators of these signals have considered the influence of antenna pointing and have been corrected accordingly.

[0032] According to the quality-corrected GNSS satellite signals, observation data extraction and weight distribution processing are performed, and finally the observation data weight distribution optimized under the antenna pointing constraint is realized. The observation data extraction process extracts the key information for navigation solution from the quality-corrected satellite signals, including pseudorange observation value, carrier phase observation value, Doppler frequency shift observation value, etc. The extraction process needs to select the appropriate data processing strategy according to the quality level of the signal. For high-quality signals, the standard data extraction algorithm is used, while for low-quality signals, a more robust extraction method is used to reduce the influence of noise. The weight distribution processing is to assign appropriate weight coefficients to each satellite signal according to the correction quality index, and the weight coefficient reflects the importance and reliability of the satellite observation data in the navigation solution. The calculation method of weight distribution is based on the inverse relationship of signal quality index, that is, the satellite with higher signal quality is assigned larger weight, and the satellite with lower signal quality is assigned smaller weight. The weight distribution also considers the geometric distribution of the satellite, and gives additional weight reward to the satellite combination with good geometric position distribution, and correspondingly reduces the weight of the satellite combination with poor geometric dilution of precision. In addition, the weight distribution also integrates a robust processing mechanism, which automatically reduces the weight of a satellite when its signal quality suddenly deteriorates to prevent abnormal data from affecting the overall solution performance. The final GNSS observation data weight distribution result optimized under the antenna pointing constraint is a data structure containing the weight information of all available satellites, which provides optimized observation data input for multi-sensor fusion attitude solution and ensures the accuracy and reliability of the navigation solution under the antenna pointing constraint of the moving antenna.

[0033] Further, the quality grading and attenuation compensation processing of the GNSS satellite signals according to the expected reception quality evaluation result includes: quantitatively grading the carrier-to-noise ratio and signal strength of each GNSS satellite signal according to the expected reception quality evaluation result to obtain the graded satellite signal quality level; calculating the attenuation amount of each satellite signal caused by antenna blocking and multipath effect according to the satellite signal quality level and the antenna pointing parameter of the moving antenna to obtain the antenna pointing related signal attenuation compensation coefficient; performing robust factor weight reduction processing on the GNSS satellite signals with low quality level and weight enhancement processing on the signals with high quality level according to the signal attenuation compensation coefficient to obtain the signal weight coefficient adjusted by quality grading; and performing adaptive adjustment processing on the observation noise covariance of each GNSS satellite signal according to the signal weight coefficient to suppress the influence of low-quality signals on the navigation solution, thereby obtaining the quality-corrected GNSS satellite signals.

[0034] Specifically, based on the expected reception quality evaluation results, the carrier-to-noise ratio and signal strength of each GNSS satellite signal are quantitatively classified and processed, and a detailed classification standard and evaluation system need to be established. The carrier-to-noise ratio refers to the ratio of signal power to noise power, expressed in decibels, which reflects the strength level of the signal relative to the noise. The higher the carrier-to-noise ratio, the better the signal quality. Signal strength directly reflects the received satellite signal power, and is closely related to factors such as satellite transmission power, propagation path loss, and receiving antenna gain. Quantitative classification processing first sets the classification threshold of carrier-to-noise ratio and signal strength, which is determined according to the actual needs and experience data of the vehicle-mounted moving channel application. The carrier-to-noise ratio classification standard divides the signal into four levels: the excellent level corresponds to a carrier-to-noise ratio greater than 45 decibels, indicating that the signal is very strong and stable; the good level corresponds to a carrier-to-noise ratio between 35 and 45 decibels, indicating that the signal strength is good; the general level corresponds to a carrier-to-noise ratio between 25 and 35 decibels, indicating that the signal strength is medium; the poor level corresponds to a carrier-to-noise ratio less than 25 decibels, indicating that the signal is weak and needs special processing. The classification standard of signal strength uses a similar method, dividing the levels according to the numerical range of the received power. In the classification process, the measured carrier-to-noise ratio and signal strength of each satellite are compared with the preset threshold to determine its level in each evaluation dimension, and then the overall quality level of the satellite signal is obtained by integrating these level information. The comprehensive evaluation adopts the weighted average method, and the carrier-to-noise ratio and signal strength are respectively given different weight coefficients, and the weight distribution is determined according to their importance to the signal quality. Through this quantitative classification processing, each GNSS satellite signal is assigned a clear quality level identifier.

[0035] According to the satellite signal quality level and the moving satellite antenna pointing parameter, the attenuation amount of each satellite signal caused by the antenna shielding and multipath effect is calculated, and the comprehensive influence of various physical factors needs to be considered in this calculation process. The antenna shielding refers to the blocking of the GNSS satellite signal propagation path by the physical structure of the moving satellite antenna, including the signal blocking caused by the antenna cover, support structure, servo mechanism and other components. The calculation of the shielding effect is based on the three-dimensional geometric model of the antenna and the spatial position relationship of each satellite relative to the antenna. When the satellite is located in the antenna structure shielding area, the signal will be attenuated to different degrees. The calculation of the shielding attenuation considers the material properties, thickness, shape and other parameters of the shielding body. Different materials have different attenuation characteristics for electromagnetic waves. Metal structures will produce strong reflection and absorption, while dielectric materials will mainly produce refraction and a small amount of absorption. The multipath effect refers to the interference phenomenon caused by the GNSS signal reaching the receiving antenna after being reflected by surrounding objects, which leads to signal strength fluctuation and measurement error. In the vehicle-mounted moving satellite environment, the multipath reflection mainly comes from the reflection surfaces such as the vehicle body structure, surrounding buildings and ground. The strength and phase of the reflected signal are related to the material properties, geometric shape and relative position of the reflection surface. The calculation of the multipath effect attenuation needs to analyze the geometric relationship and electromagnetic propagation characteristics of each reflection path, considering the influence of reflection coefficient, path length difference, incident angle and other factors. In the calculation process, the antenna pointing parameter is input as a time-varying parameter, because the change of the antenna pointing will change the position relationship of the satellite relative to the antenna structure, thereby affecting the strength of the shielding and multipath effect. By comprehensively considering the shielding and multipath effects, the total attenuation of each satellite signal is calculated, which is expressed in decibels and reflects the specific influence of the antenna pointing state on the signal reception quality of the satellite. The final signal attenuation compensation coefficient related to the antenna pointing is a dynamic parameter related to time and antenna pointing angle, which quantitatively describes the numerical size of the signal strength compensation required.

[0036] Based on the signal attenuation compensation coefficient, GNSS satellite signals of different quality levels are treated with differentiated robustness factor weighting. This approach employs different weight adjustment strategies based on signal quality. The robustness factor is a mathematical tool used to suppress the impact of outliers. By dynamically adjusting the weights of observations, it reduces the negative impact of low-quality data on the overall solution. For low-quality GNSS satellite signals, a weight reduction strategy is employed, with the degree of reduction determined by the signal quality level and the attenuation compensation coefficient. The weight reduction calculation is based on an inverse proportional relationship: the worse the signal quality and the more severe the attenuation, the greater the weight reduction. In implementation, the attenuation compensation coefficient is converted into a weight adjustment factor, which is then applied to the original weight to obtain the adjusted weight value. For high-quality signals, a weight enhancement strategy is employed. The enhancement aims to fully utilize the information content of high-quality signals to improve the accuracy and reliability of the overall solution. The weight enhancement calculation considers the signal quality advantage and the quality difference relative to other signals. The greater the quality advantage, the greater the weight enhancement. The enhancement process also considers the geometric distribution of the signals, giving additional weight bonuses to high-quality signals with favorable geometric locations. During the weight adjustment process, it is necessary to ensure that the sum of all signal weights meets normalization requirements to avoid systematic deviations from the overall solution. The adjusted weights also require bounds to ensure that no signal's weight is too large or too small, preventing a single signal from excessively influencing the solution or completely ignoring some useful signals. This differentiated weighting process results in quality-graded signal weight coefficients, which reflect the relative importance and credibility of each satellite signal under the current antenna pointing state.

[0037] According to the signal weight coefficient, the observation noise covariance of each GNSS satellite signal is adaptively adjusted to effectively suppress the influence of low-quality signals. The observation noise covariance is a statistical parameter that describes the uncertainty of the observation data. The larger the covariance value, the higher the uncertainty of the observation data, and the smaller the weight in the filtering calculation. The core idea of adaptive adjustment is to dynamically modify the observation noise covariance of the signal according to its actual quality, so that the noise covariance can truly reflect the quality state of the current signal. The adjustment process first calculates the ratio of the weight coefficient of each satellite signal to the standard weight, which reflects the deviation of the signal quality from the standard level. For low-quality signals with small weight coefficients, the observation noise covariance is increased, and the increase is inversely proportional to the degree of weight reduction. The smaller the weight, the higher the covariance increase. This adjustment allows low-quality signals to automatically obtain a smaller influence weight in the filtering calculation, effectively suppressing their negative impact on the navigation solution. For high-quality signals with large weight coefficients, the observation noise covariance is appropriately reduced, and the reduction is proportional to the degree of weight enhancement, so that high-quality signals can play a greater role in the solution. The covariance adjustment also considers the time-varying characteristics of the signal. When the signal quality changes, the covariance parameter is also dynamically adjusted to maintain the consistency of the covariance and the actual signal quality. The adjusted observation noise covariance is integrated into the measurement noise covariance matrix, which plays a key role in the measurement update step of the Kalman filter and directly affects the trust degree of the filter for different observation data. Through this adaptive adjustment mechanism, the quality-corrected GNSS satellite signals are obtained, which not only contain the original observation information but also carry the quality evaluation results and corresponding weight information optimized by the antenna pointing constraint, providing high-quality input data for multi-sensor fusion attitude solution.

[0038] S103、According to the multi-clock domain time synchronization reference and the GNSS observation data weight allocation result, a multi-sensor fusion solution is performed on the vehicle body inertial attitude to obtain a vehicle body three-axis attitude angle solution.

[0039] In an embodiment of the present application, the multi-sensor fusion calculation processing of the vehicle body inertial attitude according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result includes: according to the multi-clock domain time synchronization reference, angle feedback information extraction processing of the angle encoder of the moving channel antenna servo system and the gyro stabilized platform is performed to obtain external attitude reference data time-aligned with the GNSS observation data; according to the external attitude reference data and the three-dimensional space position offset vector between the antenna installation point of the antenna in the moving channel antenna servo system and the IMU inertial measurement unit measurement center, coordinate system conversion processing of the angle measurement value of the antenna is performed to obtain equivalent attitude information at the center of gravity of the vehicle body; according to the equivalent attitude information and the GNSS observation data weight distribution result, multi-sensor fusion calculation processing of the IMU inertial measurement unit data is performed to obtain a fusion attitude estimation result; and according to the fusion attitude estimation result, consistency verification processing of the tracking historical trajectory data of the antenna is performed to obtain a vehicle body three-axis attitude angle calculation result.

[0040] Specifically, based on the multi-clock domain time synchronization reference, angle feedback information extraction processing of the angle encoder of the moving channel antenna servo device and the gyro stabilized platform is performed, which needs to ensure that the extracted angle information is strictly aligned with the GNSS observation data in the time dimension. The angle encoder is a high-precision angle measurement device that can output the azimuth angle and elevation angle information of the antenna relative to the carrier in real time, with a measurement accuracy of angular degree level, which is much higher than the angle measurement accuracy of ordinary IMU. The gyro stabilized platform is an inertial reference device in the antenna servo device for isolating the influence of carrier motion, which maintains the inertial space orientation of the platform through a high-precision gyroscope to provide a stable reference coordinate system for the antenna. The information extraction processing first reads the real-time azimuth angle and elevation angle data of the antenna from the angle encoder, which are transmitted to the processing unit in the form of digital signals through a serial communication interface. At the same time, the platform's attitude information relative to the inertial space is obtained from the gyro stabilized platform, including the three-axis attitude angle and angular velocity data of the platform. Since these devices have independent clock sources, their time stamps are offset from the GNSS observation data, so they need to be corrected according to the multi-clock domain time synchronization reference. The specific method of time correction is to apply the correction parameters in the time synchronization reference to the original time stamps to ensure that the time markers of all angle feedback information are in the same time reference framework as the time markers of GNSS observation data. The corrected angle feedback information not only maintains the original high-precision characteristics, but also has the basic conditions for time synchronization fusion with other sensor data. Through this extraction and correction processing, external attitude reference data time-aligned with the GNSS observation data is obtained, which contains accurate angle information of the antenna relative to the carrier and attitude information of the platform relative to the inertial space.

[0041] According to the external attitude reference data and the three-dimensional spatial position offset vector between the antenna mounting point and the IMU inertial measurement unit measurement center, the angle measurement value of the antenna is processed by coordinate system conversion. This conversion process needs to consider the spatial geometric relationship and rigid body kinematics principle. The three-dimensional spatial position offset vector describes the spatial position relationship of the antenna mounting point relative to the IMU measurement center. The vector contains three components of east, north and sky, which is obtained by precise measurement. The existence of the offset vector makes the attitude angle at the antenna different from the attitude angle at the vehicle body center of gravity. Especially when the vehicle body has angular motion, the spatial position offset will produce an additional linear motion component at the antenna. The purpose of the coordinate system conversion process is to convert the angle information measured at the antenna to the equivalent angle information at the vehicle body center of gravity, and eliminate the influence of spatial position offset on attitude measurement. The conversion calculation first establishes the transformation relationship between the antenna mounting point coordinate system and the vehicle body center of gravity coordinate system, which is determined by the position offset vector and the attitude angle. In the conversion process, the influence of the six degrees of freedom motion of the vehicle body on the angle measurement at the antenna needs to be considered, including the coupling effect of three translation motions and three rotation motions. The translation motion mainly affects the linear displacement of the antenna, while the rotation motion directly affects the angular attitude of the antenna. The influence calculation of the rotation motion is based on the rigid body kinematics theory. When the vehicle body rotates around its center of gravity, the antenna mounting point will produce corresponding angular displacement and linear displacement, which need to be separated from the angle measurement value of the antenna. The separation calculation considers the length and direction of the offset vector, as well as the angle and angular velocity of the vehicle body rotation, and obtains the correction amount of the offset influence through vector operation. The corrected angle value reflects the true attitude information at the vehicle body center of gravity, and eliminates the measurement deviation caused by the installation position offset. Through this coordinate system conversion process, the equivalent attitude information at the vehicle body center of gravity is obtained, which is consistent with the measurement reference of the IMU inertial measurement unit, and provides a unified attitude reference frame for multi-sensor fusion calculation.

[0042] Based on the equivalent attitude information and the weight distribution results of GNSS observation data, the IMU inertial measurement unit data is processed by multi-sensor fusion calculation. The fusion process uses a weighted fusion algorithm to comprehensively utilize multi-source attitude information. The core idea of multi-sensor fusion calculation is to assign appropriate weights to each sensor data according to its quality and reliability, and then obtain the optimal fusion result through mathematical operation. The fusion calculation first preprocesses and formats the various data sources involved in the fusion, including the three-axis angular velocity and three-axis acceleration data of the IMU, the equivalent attitude information after coordinate system conversion, and the GNSS observation data with weight information. The IMU data is integrated to obtain attitude angle information, and the integral process takes into account the influence of gyroscope bias and scale factor error. The fourth-order Runge-Kutta method is used for numerical integration to ensure calculation accuracy. The equivalent attitude information is directly involved in the fusion calculation as a high-precision external reference, and its weight is determined according to the accuracy characteristics and current working state of the antenna servo device. The weight distribution results of GNSS observation data are used to adjust the contribution of GNSS attitude information in the fusion, and the greater the weight, the greater the influence on the fusion result. The fusion algorithm uses an extended Kalman filter framework to establish a state space model containing attitude angle, angular velocity bias, acceleration bias, and other state variables. In the time update step of the filter, the angular velocity data of the IMU is used to predict the attitude state at the next time, and in the measurement update step, the predicted results are corrected using the equivalent attitude information and GNSS attitude information. The weight distribution in the correction process directly affects the contribution of different observation data to the final result, and the observation data with high weight has a greater impact on the filter state estimation. The filter also integrates an adaptive mechanism that can dynamically adjust the fusion weight according to the real-time quality of each sensor data. When the data quality of a certain sensor decreases, its weight in the fusion is automatically reduced. Through this multi-sensor fusion calculation process, the fusion attitude estimation result is obtained, which integrates the information of IMU, antenna servo device, GNSS, and other sensors, and has higher precision and reliability than single sensor.

[0043] According to the fusion attitude estimation result, consistency verification processing is performed on the tracking historical trajectory data of the antenna. This verification method uses the continuity and consistency of the antenna pointing to verify the correctness of the attitude solution result. The tracking historical trajectory data of the antenna records the pointing trajectory of the antenna in a period of time, including the azimuth and elevation angle change sequence of the target satellite. The basic principle of consistency verification is to compare and analyze the fusion attitude estimation result with the antenna tracking trajectory to verify whether there is logical consistency between the two. The verification processing first calculates the theoretical angle sequence that the antenna should point to according to the fusion attitude estimation result and the known target satellite orbit parameters, and then compares this theoretical angle sequence with the actual antenna tracking trajectory. The comparison analysis considers the influence of the vehicle body attitude change on the antenna pointing. When the vehicle body rolls, pitches or yaws, the antenna needs to adjust its pointing angle accordingly to maintain tracking of the target satellite. If the fusion attitude estimation result is correct, the theoretical antenna pointing angle calculated based on the result should be highly consistent with the actual antenna tracking trajectory. The consistency evaluation uses statistical analysis method to calculate the deviation sequence between the theoretical value and the actual value, and analyzes the statistical characteristics of the deviation. Under normal circumstances, the deviation should present a random distribution with zero mean. If the deviation has a systematic trend or exceeds the preset tolerance range, it indicates that the fusion attitude estimation result has a problem. The verification processing also considers the influence of the antenna servo performance and environmental interference factors, and identifies and eliminates the trajectory deviation caused by servo delay or external interference. When the consistency verification passes, the fusion attitude estimation result is confirmed as valid vehicle body attitude information, including the roll angle, pitch angle and yaw angle three components. These attitude angle information has been double-protected by multi-sensor fusion and consistency verification, and has the characteristics of high precision and high reliability. The final three-axis attitude angle solution result of the vehicle body not only meets the precision requirements of the moving transponder application, but also has real-time and robustness, providing a reliable attitude reference for the stable operation of the vehicle-mounted moving transponder device.

[0044] Further, the multi-sensor fusion calculation processing of the IMU inertial measurement unit data according to the equivalent attitude information and the GNSS observation data weight distribution result includes: according to the equivalent attitude information, the angle feedback data of the moving channel antenna servo system and the three-axis attitude angle of the IMU inertial measurement unit are subjected to data alignment processing to obtain multi-source attitude measurement values at the same time; according to the multi-source attitude measurement values and the GNSS observation data weight distribution result, the fusion weight coefficients of each sensor attitude data are calculated to obtain the weight parameters of multi-sensor attitude fusion; according to the weight parameters, the attitude angle of the IMU inertial measurement unit and the attitude feedback of the moving channel antenna servo system are subjected to weight-based attitude matrix fusion calculation processing to obtain the fused attitude rotation matrix; and according to the fused attitude rotation matrix, rotation matrix to Euler angle conversion calculation processing is performed to obtain the fusion attitude estimation result.

[0045] Specifically, based on the equivalent attitude information, the angle feedback data of the moving satellite antenna servo device is aligned with the three-axis attitude angle of the IMU inertial measurement unit. This alignment process needs to solve the differences in sampling frequency, data format and time reference of different sensors. The angle feedback data of the moving satellite antenna servo device contains two components of azimuth angle and pitch angle, with a sampling frequency of 50 Hz, a data format of 16-bit integer angle value, and an accuracy of 0.01 degrees. The three-axis attitude angle output by the IMU inertial measurement unit includes roll angle, pitch angle and yaw angle, with a sampling frequency of 200 Hz and a data format of 32-bit floating point number, which is obtained by integrating the three-axis angular velocity. Due to the difference in sampling frequency, the high-frequency IMU data needs to be down-sampled or the low-frequency antenna data needs to be interpolated to realize frequency alignment. The down-sampling processing adopts the method of low-pass filtering and decimation, first low-pass filters the 200 Hz IMU attitude angle data to prevent aliasing, and then decimates the data sequence to 50 Hz according to the ratio of 4:1. Data format alignment needs to convert the integer angle value of the antenna servo device to floating point format, and the original numerical accuracy is kept unchanged during the conversion process. Time reference alignment uses the multi-clock domain time synchronization reference established in the early stage to unify the timestamps of the two kinds of data to the same time reference framework. The alignment process also needs to handle the difference in coordinate system. The antenna servo device outputs the attitude angle of the antenna relative to the carrier, while the IMU outputs the attitude angle of the carrier relative to the navigation coordinate system, and the reference bases of the two are different. Coordinate system conversion is realized through rotation matrix operation, which converts the angle value in the antenna coordinate system to the same navigation coordinate system as the IMU. The conversion calculation considers the coupling relationship between the antenna installation angle and the carrier attitude, ensuring that the converted angle value has the same physical meaning. After frequency alignment, format conversion, time synchronization and coordinate system conversion, the multi-source attitude measurement values at the same time are obtained, which are completely consistent in time, frequency, format and coordinate system.

[0046] According to the multi-source attitude measurement values and the GNSS observation data weight distribution results, the fusion weight coefficients of the attitude data of each sensor are calculated, and the calculation process needs to comprehensively consider the inherent accuracy characteristics, the current working state and the external environmental influence of the sensor. The weight calculation of the IMU (Inertial Measurement Unit) is based on the noise characteristics of the gyroscope and the accelerometer of the IMU. The random walk coefficient and the zero bias stability of the gyroscope directly affect the accuracy of the attitude angle integration. The longer the integration time is, the larger the cumulative error is, and the smaller the corresponding weight should be. The noise level and the zero bias error of the accelerometer affect the measurement accuracy of the horizontal attitude angle. In the vehicle motion state, the accelerometer is also disturbed by the linear acceleration of the carrier, which reduces the reliability of its attitude measurement. The weight calculation of the servo device of the moving antenna is based on the accuracy specification and the current tracking performance of the angle encoder of the servo device. The angle encoder has high instantaneous accuracy, but its measurement value reflects the angle of the antenna relative to the carrier, and the absolute attitude information needs to be obtained through the carrier attitude calculation. When the antenna is in the fast tracking state, the dynamic error of the servo device will increase, and the corresponding weight needs to be reduced. The weight distribution results of the GNSS observation data provide another important weight information source. The GNSS double-antenna configuration can directly measure the heading angle of the carrier, and the size of the weight reflects the current quality and geometric distribution of the GNSS signal. The fusion weight calculation adopts a comprehensive evaluation method to combine the theoretical accuracy, the real-time performance index and the external quality evaluation results of each sensor. The theoretical accuracy is determined based on the technical specifications of the sensor, the real-time performance index is obtained by analyzing the statistical characteristics of the sensor output data, and the external quality evaluation results come from the signal quality monitoring and the consistency test. The weight calculation also introduces the time-varying characteristics. When the performance of a certain sensor changes, the weight of the sensor is also dynamically adjusted accordingly. The dynamic adjustment mechanism monitors the innovation sequence of the data of each sensor. When the innovation exceeds the expected range, the weight of the sensor is automatically reduced. The final weight parameters of the multi-sensor attitude fusion are a parameter set containing the weight information of all the sensors participating in the fusion, and the parameter set reflects the relative importance and the reliability of the attitude data of each sensor at the current time.

[0047] Based on the weight parameters, the attitude angles of the IMU (inertial measurement unit) and the attitude feedback of the antenna servo device are fused into a weight-based attitude matrix. This fusion method converts the attitude angles into a rotation matrix for mathematical operations. The rotation matrix is a mathematical tool for describing the rotation transformation in three-dimensional space, which can avoid the singularity problem in Euler angle representation and provide a more stable mathematical framework for attitude fusion. The conversion of attitude angles to rotation matrix uses a standard rotation sequence, in which the roll angle, pitch angle, and yaw angle are applied in a specific rotation order to generate the corresponding rotation matrix. During the conversion process, the angle values are input in radians, and the elements of the rotation matrix are obtained through trigonometric operations. The rotation matrix obtained by converting the IMU attitude angles reflects the attitude relationship of the carrier relative to the navigation coordinate system, while the rotation matrix obtained by converting the antenna servo attitude feedback reflects the attitude relationship of the antenna relative to the carrier. The matrix fusion calculation uses a weighted average method, but since the rotation matrix is on the special orthogonal group, it cannot be directly linearly weighted, and a fusion algorithm suitable for the rotation matrix is needed. The fusion algorithm first converts each rotation matrix to a quaternion representation. The quaternion is another mathematical tool for describing rotation, which has a compact representation and good numerical stability. The quaternion fusion uses a spherical linear interpolation method, which performs weighted interpolation operations in the quaternion space according to the weight parameters. The interpolation calculation considers the unit length constraint and double coverage characteristics of the quaternion to ensure that the fusion result is still a valid unit quaternion. The fused quaternion is converted back to a rotation matrix to obtain the fused attitude rotation matrix. This rotation matrix integrates the attitude information of multiple sensors, and its accuracy and reliability are better than the measurement results of any single sensor.

[0048] According to the fused attitude rotation matrix, a rotation matrix to Euler angle conversion calculation process is performed, and this conversion process needs to process the numerical stability of the rotation matrix and the singularity problem of the Euler angle. The conversion of the rotation matrix to the Euler angle is a standard transformation of the attitude representation, and three Euler angles are obtained by extracting specific elements in the rotation matrix and performing inverse trigonometric function operations. The conversion calculation adopts the ZYX rotation sequence, that is, first rotating around the Z axis by the yaw angle, then rotating around the Y axis by the pitch angle, and finally rotating around the X axis by the roll angle. The calculation of the yaw angle is based on the inverse tangent operation of the first row first column and the second row first column elements of the rotation matrix, the pitch angle is obtained by the inverse sine operation of the third row first column element, and the roll angle is determined by the inverse tangent operation of the third row second column and the third row third column elements. In the conversion process, the singularity of the pitch angle close to plus or minus 90 degrees needs to be specially handled, at this time some elements of the rotation matrix are close to zero, which will cause the instability of numerical calculation. The singularity processing adopts the conditional judgment and alternative calculation method, when the singularity state is detected, a more stable calculation path or the angle value of the previous moment is used. The conversion calculation also needs to process the periodicity of the angle value to ensure that the output Euler angle is within a reasonable numerical range. The range of the roll angle and the yaw angle is limited between-180 degrees and 180 degrees, and the range of the pitch angle is limited between-90 degrees and 90 degrees. The angle range constraint is realized through modulus operation and conditional judgment, when the angle value exceeds the range, the periodic adjustment is automatically performed. The converted Euler angle also needs to be smoothed to eliminate small amplitude jumps caused by numerical calculation errors. The smoothing processing adopts a first-order low-pass filter, and the cutoff frequency of the filter is set according to the application requirements, which can track the real attitude change and suppress high-frequency noise. After the rotation matrix conversion, singularity processing, range constraint and smoothing filtering, the final fused attitude estimation result is obtained, which contains the roll angle, pitch angle and yaw angle of the vehicle body, has the characteristics of high precision, high stability and strong robustness, and meets the strict requirements of the vehicle-mounted dynamic channel application for attitude information.

[0049] In the embodiment, the double-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted dynamic channel system are subjected to multi-clock domain synchronous calibration processing; quality evaluation and weight distribution processing of the GNSS satellite signal under the constraint of the dynamic channel antenna pointing are performed according to the time synchronization reference; and multi-sensor fusion solving processing of the vehicle body inertia attitude is performed according to the time synchronization reference and the GNSS observation data weight distribution result. The present application realizes millisecond-level time synchronization by establishing a robust extended Kalman filter model of a multi-dimensional extended state vector, adopts an adaptive weight distribution strategy of antenna pointing constraint to optimize the GNSS signal quality, and utilizes high-precision angle feedback of the antenna servo system to enhance the attitude fusion precision, thereby solving the problem of low precision of the traditional attitude solving method in the vehicle-mounted dynamic channel system.

[0050] The above describes the sensor fusion-based inertial navigation attitude solving method in the embodiment of the application, and the following describes a sensor fusion-based inertial navigation attitude solving device in the embodiment of the application, please refer to Figure 2 One embodiment of the sensor fusion-based inertial navigation attitude solving device in the embodiment of the application includes: The time synchronization module 201 is configured to perform multi-clock domain synchronization calibration processing on the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted dynamic communication system, to obtain a multi-clock domain time synchronization reference; The signal optimization module 202 is configured to perform quality evaluation and weight distribution processing on GNSS satellite signals under the constraint of dynamic communication antenna pointing according to the multi-clock domain time synchronization reference, to obtain a GNSS observation data weight distribution result optimized under the constraint of antenna pointing; The attitude fusion module 203 is configured to perform multi-sensor fusion solving processing on the vehicle body inertial attitude according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result, to obtain a vehicle body three-axis attitude angle solving result.

[0051] In the embodiment of the application, the sensor fusion-based inertial navigation attitude solving device runs the above-described sensor fusion-based inertial navigation attitude solving method. The sensor fusion-based inertial navigation attitude solving device performs multi-clock domain synchronization calibration processing on the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system in the vehicle-mounted dynamic communication system; performs quality evaluation and weight distribution processing on GNSS satellite signals under the constraint of dynamic communication antenna pointing according to the time synchronization reference; and performs multi-sensor fusion solving processing on the vehicle body inertial attitude according to the time synchronization reference and the GNSS observation data weight distribution result. The application realizes millisecond-level time synchronization by establishing a robust extended Kalman filter model of a multi-dimensional extended state vector, optimizes GNSS signal quality by using an adaptive weight distribution strategy of antenna pointing constraint, and enhances attitude fusion accuracy by using high-precision angle feedback of the antenna servo system, thereby solving the problem of low accuracy of the traditional attitude solving method in the vehicle-mounted dynamic communication system.

[0052] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system or device, unit can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0053] The integrated unit, if implemented in the form of a software function unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the entire or part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0054] The above-described embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the same; even though the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that modifications can still be made to the technical solutions recorded in the foregoing embodiments, or equivalent replacements can be made to some of the technical features; and these modifications or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A sensor fusion-based inertial navigation attitude solution method, characterized in that: The inertial navigation attitude solution method based on sensor fusion includes: Perform multi-clock domain synchronization calibration on the dual-antenna GNSS system, IMU inertial measurement unit, and antenna servo system in the vehicle-mounted mobile communication system to obtain a multi-clock domain time synchronization benchmark; Performing quality assessment and weight distribution processing on the GNSS satellite signals under the antenna pointing constraint of the communication in motion according to the multi-clock domain time synchronization benchmark, and obtaining a GNSS observation data weight distribution result optimized by the antenna pointing constraint; According to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result, the vehicle body inertial attitude is subjected to multi-sensor fusion solution processing to obtain the vehicle body three-axis attitude angle solution result.

2. The inertial navigation attitude solution method based on sensor fusion according to claim 1 is characterized in that: The multi-clock domain synchronization calibration process is performed on the dual-antenna GNSS system, the IMU inertial measurement unit, and the antenna servo system in the vehicle-mounted mobile communication system to obtain a multi-clock domain time synchronization benchmark, including: The time synchronization error between the GNSS receiver and the IMU inertial measurement unit is used as a state variable and combined with the multi-dimensional navigation state variable of the IMU inertial measurement unit to obtain a multi-dimensional extended state vector including the time synchronization error. According to the multidimensional extended state vector, the position and velocity measurement values ​​of the dual-antenna GNSS system are linearized by a second-order Taylor series expansion to obtain a linearized measurement equation that takes into account the influence of time synchronization error; According to the linearized measurement equation, a robust extended Kalman filter is used to filter and estimate the multidimensional extended state vector to obtain a real-time time synchronization error estimate; According to the time synchronization error estimate, time offset compensation processing is performed on the observation data of the dual-antenna GNSS system, the IMU inertial measurement unit and the antenna servo system, and the synchronization effect is verified to obtain a multi-clock domain time synchronization benchmark.

3. The inertial navigation attitude solution method based on sensor fusion according to claim 2 is characterized in that: The linearized measurement equation considering the influence of time synchronization error is obtained by performing a second-order Taylor series expansion linearization process on the position and velocity measurement values ​​of the dual-antenna GNSS system according to the multidimensional extended state vector. The linearized measurement equation includes: performing a second-order Taylor series expansion calculation on the position measurement value of the dual-antenna GNSS system in the presence of time delay based on the time synchronization error state variable in the multidimensional extended state vector to obtain a position expansion expression including a time delay influence term; According to the position expansion expression, a second-order Taylor series expansion calculation is performed on the velocity measurement value of the dual-antenna GNSS system in the presence of time delay to obtain a velocity expansion expression including a time delay influence term; According to the position expansion expression and the velocity expansion expression, the second-order and higher-order terms are ignored, and the first-order time delay influence term is retained to obtain the linearized position and velocity approximate expressions; According to the linearized position and velocity approximate expressions, a linear relationship matrix between state variables and observation values ​​is constructed to obtain a linearized measurement equation that takes into account the influence of time synchronization error.

4. The inertial navigation attitude solution method based on sensor fusion according to claim 1 is characterized in that: The step of performing quality assessment and weight distribution processing on the GNSS satellite signal under the antenna pointing constraint of the communication in motion according to the multi-clock domain time synchronization reference to obtain the GNSS observation data weight distribution result optimized by the antenna pointing constraint includes: Performing time stamp calibration processing on the received GNSS satellite signal according to the multi-clock domain time synchronization reference to obtain a time-synchronized GNSS satellite signal; Perform antenna pointing impact analysis on the time-synchronized GNSS satellite signals based on the real-time pointing parameters of the mobile communication antenna to obtain an expected reception quality assessment result for each satellite signal; performing quality grading and attenuation compensation processing on the GNSS satellite signal according to the expected reception quality assessment result to obtain a quality-corrected GNSS satellite signal; Observation data extraction and weight distribution processing are performed based on the quality-corrected GNSS satellite signals to obtain a GNSS observation data weight distribution result optimized through antenna pointing constraints.

5. The inertial navigation attitude solution method based on sensor fusion according to claim 4 is characterized in that: The performing quality grading and attenuation compensation processing on the GNSS satellite signal according to the expected reception quality assessment result to obtain a quality-corrected GNSS satellite signal includes: performing quantitative grading processing on the carrier-to-noise ratio and signal strength of each GNSS satellite signal based on the expected reception quality assessment result to obtain a graded satellite signal quality level; Calculating the attenuation of each satellite signal due to antenna shielding and multipath effects based on the satellite signal quality level and the CSM antenna pointing parameters, and obtaining a signal attenuation compensation coefficient related to the antenna pointing; According to the signal attenuation compensation coefficient, a robust factor weight reduction process is performed on low-quality GNSS satellite signals, and a weight enhancement process is performed on high-quality signals to obtain a signal weight coefficient adjusted by quality grading; According to the signal weight coefficient, the observation noise covariance of each GNSS satellite signal is adaptively adjusted to suppress the influence of low-quality signals on navigation solution, and obtain a GNSS satellite signal with corrected quality.

6. The inertial navigation attitude solution method based on sensor fusion according to claim 1 is characterized in that: The performing of multi-sensor fusion solution processing on the vehicle body inertial attitude according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result to obtain the vehicle body three-axis attitude angle solution result includes: Extracting and processing angle feedback information from the angle encoder and gyro stabilization platform of the CMS antenna servo system based on the multi-clock domain time synchronization benchmark to obtain external attitude reference data that is time-aligned with the GNSS observation data; Based on the external attitude reference data and the three-dimensional spatial position offset vector between the antenna installation point of the antenna in the mobile communication antenna servo system and the IMU inertial measurement unit measurement center, the antenna angle measurement value is subjected to coordinate system conversion processing to obtain equivalent attitude information at the center of gravity of the vehicle body; Performing multi-sensor fusion processing on the IMU inertial measurement unit data according to the equivalent attitude information and the weight distribution result of the GNSS observation data to obtain a fusion attitude estimation result; According to the fusion attitude estimation result, consistency verification processing is performed on the antenna's tracking history trajectory data to obtain the vehicle's three-axis attitude angle solution result.

7. The inertial navigation attitude solution method based on sensor fusion according to claim 6 is characterized in that: The performing multi-sensor fusion solution processing on the IMU inertial measurement unit data according to the equivalent attitude information and the GNSS observation data weight distribution result to obtain the fusion attitude estimation result includes: According to the equivalent attitude information, the angle feedback data of the moving antenna servo system is aligned with the three-axis attitude angle of the IMU inertial measurement unit to obtain multi-source attitude measurement values ​​at the same time; Calculating the fusion weight coefficient of each sensor attitude data according to the multi-source attitude measurement value and the GNSS observation data weight distribution result to obtain the weight parameter of multi-sensor attitude fusion; According to the weight parameters, a weighted attitude matrix fusion calculation process is performed on the attitude angle of the IMU inertial measurement unit and the attitude feedback of the moving antenna servo system to obtain a fused attitude rotation matrix; According to the fused posture rotation matrix, a conversion calculation process is performed from the rotation matrix to Euler angles to obtain a fused posture estimation result.

8. An inertial navigation attitude solving device based on sensor fusion, characterized in that: The inertial navigation attitude solving device based on sensor fusion includes: The time synchronization module is used to perform multi-clock domain synchronization calibration on the dual-antenna GNSS system, IMU inertial measurement unit, and antenna servo system in the vehicle-mounted mobile communication system to obtain a multi-clock domain time synchronization benchmark; a signal optimization module for performing quality assessment and weight distribution processing on the GNSS satellite signals under the antenna pointing constraint of the communication in motion according to the multi-clock domain time synchronization reference, and obtaining a GNSS observation data weight distribution result optimized by the antenna pointing constraint; The attitude fusion module is used to perform multi-sensor fusion solution processing on the vehicle body inertial attitude according to the multi-clock domain time synchronization reference and the GNSS observation data weight distribution result to obtain the vehicle body three-axis attitude angle solution result.

Images