An inertial navigation positioning calibration method and system based on multi-source error compensation

Abstract

This invention belongs to the field of high-precision positioning technology. It discloses an inertial navigation positioning calibration method and system based on multi-source error compensation. The method includes acquiring acceleration and angular velocity data output by an inertial measurement unit (IMU), obtaining position, velocity, and attitude information through integration, and constructing navigation state variables under a unified time series. Errors in the navigation state variables are characterized and decomposed into zero-bias error, scaling factor error, dynamic coupling error, and environmental disturbance error, forming a multi-source error superposition relationship and generating an error sequence that evolves over time. Error evolution prediction modeling is performed based on the error sequence, dividing the error sequence into steady-state intervals and transition intervals. An evolution prediction function constrained by motion is constructed within the steady-state interval, and an error-driven enhancement prediction function is constructed within the transition interval. Continuity constraints are applied to adjacent intervals to obtain continuous error prediction results, thereby improving the accuracy and stability of inertial navigation positioning.

Description

Technical Field

[0001] This invention relates to the field of high-precision positioning technology, and more specifically, to an inertial navigation positioning calibration method and system based on multi-source error compensation. Background Technology

[0002] Existing inertial navigation positioning calibration methods and systems mainly suffer from the following problems: Inertial navigation systems (INS) calculate the position, velocity, and attitude of a vehicle by integrating the acceleration and angular velocity output by the inertial measurement unit (IMU). They are widely used in unmanned systems, intelligent equipment, and aerospace. However, a typical problem with these systems is that errors accumulate over time, and without effective calibration, the positioning results will gradually deviate from the true values.

[0003] Current inertial navigation and positioning calibration methods still have several shortcomings in error handling. First, many methods typically employ a single error model or simply superimpose different errors, making it difficult to simultaneously characterize the differences and interactions between different types of errors, resulting in an incomplete error representation. Second, most methods primarily handle errors based on first-order changes or simple filtering, lacking the ability to characterize the changing trends of errors, i.e., insufficient reflection of the acceleration of error changes, thus making it difficult to identify abrupt or non-stationary changes in errors.

[0004] Furthermore, in error analysis, existing methods typically fail to meticulously segment the error sequence or rely solely on a single threshold for differentiation, making it difficult to effectively distinguish between stationary and abrupt error phases, resulting in a lack of targeted subsequent processing. Simultaneously, current technologies lack effective modeling methods for the interrelationships between multiple error sources, failing to reflect the structural characteristics of the errors from a holistic perspective, leading to incomplete error analysis results. Moreover, some methods rely solely on a single time dimension or a single error quantity for judgment, making them susceptible to noise interference and misjudgments, thus affecting the overall processing effectiveness.

[0005] In error prediction, existing technologies typically employ a uniform prediction model to process the entire error sequence, failing to differentiate the varying characteristics of errors at different stages, resulting in poor model adaptability. When error changes are relatively gradual, these methods often fail to consider the carrier's motion state, making it difficult to reflect the coupling relationship between error and motion. Conversely, when errors change rapidly or abruptly, traditional low-order models struggle to characterize their nonlinear enhancement features, easily leading to prediction lag or inaccuracies. Furthermore, these methods usually cannot dynamically adjust the model's scope based on the intensity of error changes, resulting in insufficient refinement in characterizing local abrupt changes.

[0006] In view of this, the present invention proposes an inertial navigation positioning calibration method based on multi-source error compensation to solve the above problems. Summary of the Invention

[0007] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: an inertial navigation positioning calibration method based on multi-source error compensation, comprising: S1. Obtain the acceleration and angular velocity data output by the inertial measurement unit, and obtain the position, velocity and attitude information through integration to construct the navigation state variables under a unified time series. S2. The errors in the navigation state variables are characterized and decomposed into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forming a multi-source error superposition relationship and generating an error sequence that evolves over time. S3. Based on the error sequence, perform error evolution prediction modeling and divide the error sequence into steady-state intervals and transition intervals; construct a motion-constrained evolution prediction function in the steady-state interval, construct an error-driven enhancement prediction function in the transition interval, and apply continuity constraints to adjacent intervals to obtain continuous error prediction results. S4. Generate the corresponding compensation amount based on the error prediction result, and compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning result. S5. The compensated navigation and positioning results are used as the navigation state input for the next moment, driving the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning.

[0008] Preferably, the method for acquiring the acceleration and angular velocity data output by the inertial measurement unit includes: The triaxial accelerometer and triaxial gyroscope in the inertial measurement unit are sampled synchronously to obtain the triaxial acceleration measurement value and triaxial angular velocity measurement value of the navigation vehicle, and a unified timestamp is assigned to the triaxial acceleration measurement value and triaxial angular velocity measurement value; The triaxial acceleration and triaxial angular velocity measurements are processed by a sliding time window to obtain smooth triaxial acceleration and triaxial angular velocity sequences. The triaxial acceleration and triaxial angular velocity sequences are then organized according to a unified time index to form a continuous inertial measurement data sequence.

[0009] Preferably, the method for constructing navigation state quantities under a unified time series includes: The attitude of the navigation vehicle is recursively updated based on the three-axis angular velocity sequence. The attitude increment is obtained by integrating the three-axis angular velocity at each sampling time, and the attitude at the current time is updated by combining the initial attitude of the navigation vehicle, resulting in a continuous attitude sequence. Based on the attitude sequence, a coordinate transformation relationship is constructed, and the three-axis acceleration sequence is transformed from the carrier coordinate system to the navigation coordinate system. Gravity component compensation is performed on the transformed three-axis acceleration to obtain the equivalent acceleration sequence in the navigation coordinate system. The equivalent acceleration sequence is integrated once in time, and the velocity at each sampling time is recursively calculated using the velocity at the previous time as the initial condition to obtain the velocity sequence corresponding to the time index. The velocity sequence is integralized twice over time, and the position sequence corresponding to the time index is obtained by recursion calculation at each sampling time with the position of the previous time as the initial condition. The attitude sequence, velocity sequence and position sequence are aligned and fused according to a unified time index to construct the navigation state variables under a unified time sequence.

[0010] Preferably, the method for forming the multi-source error superposition relationship includes: Based on the navigation state quantities under a unified time series, a state deviation sequence corresponding to each sampling time is constructed. The state deviation sequence is calculated from the difference of the navigation state quantities at adjacent sampling times. The navigation state quantities are statistically calculated within a sliding time window of a fixed length to obtain the local mean sequence corresponding to each time window. The local mean sequence is used as a slow-varying trend component, and the slow-varying trend component is defined as zero bias error. Based on the input-output relationship of navigation state variables in the integral evolution process, for any sampling moment, the equivalent acceleration in the navigation coordinate system is taken as the driving quantity, and the velocity obtained by the corresponding integration is taken as the output quantity. A proportional mapping relationship between the output quantity and the driving quantity is constructed. By calculating the deviation between the velocity increment at the current moment and the corresponding acceleration integration result, the deviation is determined as the proportional factor error. Using triaxial angular velocity, triaxial acceleration, and attitude parameters as basic variables, a multivariate expression relationship containing the cross product term between triaxial angular velocity and triaxial acceleration is constructed during the transformation of triaxial acceleration from the carrier coordinate system to the navigation coordinate system. By calculating the difference between the coordinate transformation results at the current time and the previous time, the acceleration additional term caused by the change in angular velocity is extracted, and this additional term is determined as the dynamic coupling error. After detrending the state deviation sequence, the remaining residual sequence is extracted, and the part of the residual sequence that does not meet the expression form of zero bias error, scale factor error and dynamic coupling error is defined as environmental disturbance error. The zero bias error, scale factor error, dynamic coupling error and environmental disturbance error are superimposed according to a unified time index to form a multi-source error superposition relationship of navigation state quantity.

[0011] Preferably, the method for generating the error sequence that evolves over time includes: Based on a unified time index, the zero bias error, scaling factor error, dynamic coupling error, and environmental disturbance error are aligned at each time point to construct the time series corresponding to each type of error component. The error components at the same time point are superimposed to obtain the total error vector at the corresponding time point. The total error vector at each time point is organized and arranged in chronological order to form an error sequence that evolves over time.

[0012] Preferably, the method for dividing the error sequence into a steady-state interval and a transition interval includes: After obtaining the error sequence that evolves over time, the error values ​​corresponding to each time point are extracted according to a unified time index to construct an evolution analysis sequence of error over time. Second-order difference calculation is performed on the error sequence, and gradient change information between multi-source errors is extracted based on the spatial distribution relationship of various error components at the same time point to characterize the degree of coupling change between different error components. The second-order difference result is combined with the gradient change information to construct an error structure criterion function, and the value of the error structure criterion function is obtained. Based on the comparison result between the error structure criterion function value and the preset interval division threshold, the error state at each time is determined. When the error structure criterion function value is less than the preset interval division threshold, the corresponding time is divided into a steady state interval. When the preset interval division threshold is greater than or equal to the preset interval division threshold, the corresponding time is divided into a transition interval.

[0013] Preferably, the method for obtaining continuous error prediction results includes: After dividing the error sequence into steady-state and transition intervals, the intervals are organized in chronological order, and corresponding error prediction functions are constructed for the differences in error evolution characteristics in different intervals. In the steady-state interval, an evolution prediction function containing time polynomial terms is constructed, and motion constraint terms related to navigation state changes are introduced. By coupling error changes with changes in carrier motion state, the prediction of the error change process can be achieved. Within the transition interval, an enhanced prediction function driven by the error magnitude and its variation characteristics is constructed. The transition process is characterized piecewise by setting basis functions, and the range of action of the basis functions is adaptively adjusted according to the error variation intensity at each time step. After constructing prediction functions for different intervals, a continuity constraint is applied to adjacent intervals to ensure that the function values ​​at the interval boundary points remain consistent and that the first-order change trend of the function at that point remains continuous. This ensures a smooth transition in the numerical values ​​and change trends of the prediction results for different intervals, ultimately yielding error prediction results that change continuously throughout the entire time series.

[0014] Preferably, the method for outputting the compensated navigation and positioning results includes: After obtaining the error prediction results at each time point, the error prediction values ​​at the corresponding time point are extracted and quantized according to the error type to obtain a set of error components corresponding to the navigation state variables. Based on the set of error components, the error components are mapped to the position, velocity and attitude parameters in the navigation state variables to construct an error vector that corresponds one-to-one with each state parameter. The compensation amount is generated based on the error vector. The direction of the compensation amount is opposite to that of the corresponding error component, and its magnitude is determined by the error prediction value, so as to cancel the error. For position parameters and velocity parameters, the compensation amount is superimposed on the original position parameter and velocity parameter state variables in the corresponding coordinate system for correction, so as to obtain the compensated position parameters and velocity parameters. For attitude parameters, quaternion-based compensation is used. By constructing a compensation quaternion corresponding to the attitude error and combining it with the current attitude quaternion, the compensated attitude parameters are obtained. After completing the compensation and correction of position, velocity, and attitude, the various state parameters are integrated according to a unified time index to form the compensated navigation state variables, and the corresponding navigation and positioning results are output.

[0015] Preferably, the method for achieving continuous dynamic calibration of inertial navigation positioning includes: After the compensation and correction of the navigation state quantity is completed at the current sampling time, the compensated navigation state quantity is obtained and used as the initial state input for the navigation solution at the next sampling time. At the next sampling time, the updated navigation state quantity is obtained by integrating the acceleration and angular velocity data output by the inertial measurement unit and the initial state. The updated navigation state variables are repeatedly subjected to error characterization and decomposition processing to extract zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, and generate corresponding error sequences. Based on the error sequences, error evolution characteristic analysis and interval division are further performed to determine the steady state interval and transition interval of the error state. Based on the interval determination results, the corresponding evolution prediction function or enhancement prediction function is called to perform error prediction, obtain the error prediction result at the current time, and generate a new compensation amount based on the error prediction result; the new compensation amount is applied to the navigation state quantity at the current time to complete the compensation correction; by repeating the process at each time, the compensated navigation state quantity is continuously used as the navigation state quantity input at the next time, thereby forming a closed-loop iteration of error representation, prediction modeling and compensation correction, realizing the continuous dynamic calibration of inertial navigation and positioning results in the time dimension.

[0016] An inertial navigation positioning calibration system based on multi-source error compensation includes: The inertial navigation construction module is used to acquire the acceleration and angular velocity data output by the inertial measurement unit, and obtain position, velocity and attitude information through integration calculations to construct navigation state variables under a unified time series. The multi-source error characterization module is used to characterize and decompose the errors in navigation state variables. It divides the errors into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forms a multi-source error superposition relationship, and generates an error sequence that evolves over time. The segmented prediction modeling module is used to perform error evolution prediction modeling based on the error sequence, and divide the error sequence into steady-state intervals and transition intervals; within the steady-state interval, an evolution prediction function constrained by motion is constructed, and within the transition interval, an error-driven enhancement prediction function is constructed, and continuity constraints are applied to adjacent intervals to obtain continuous error prediction results. The error compensation and correction module is used to generate corresponding compensation amounts based on the error prediction results, and to compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning results. The calibration feedback update module is used to take the compensated navigation and positioning results as the input for the next moment, drive the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning.

[0017] Compared with the prior art, the present invention has the following beneficial effects: This invention maps multi-source errors to discrete fields in the error component space and extracts their gradient change information, enabling different types of errors to be expressed within a unified framework, thereby improving the completeness and consistency of error modeling. By introducing the second-order difference component of the error sequence, the changes in the error rate are effectively reflected, thereby identifying accelerated or abrupt changes in the error and improving the ability to perceive non-stationary errors. By simultaneously introducing the second-order difference component and error gradient information, a multi-dimensional judgment basis is formed, reducing the uncertainty caused by a single feature and improving the stability and anti-interference ability of interval division. By structurally dividing the error sequence, a basis is provided for adopting different error prediction and compensation strategies in different intervals, thereby improving the overall error compensation effect. By more accurately identifying the error evolution state and performing targeted processing, error accumulation is effectively suppressed, and the accuracy and long-term stability of navigation results in complex environments are improved.

[0018] By dividing the error sequence into steady-state and transition intervals and constructing differentiated prediction functions in different intervals, the model can adaptively process different error stages, improving overall prediction accuracy. In the steady-state interval, a low-order model based on time polynomials is introduced, and motion constraint terms for navigation state changes are superimposed to achieve coupled modeling between error and carrier motion state, thereby improving the fitting ability for slowly changing errors. In the transition interval, by introducing basis function superposition and exponential enhancement mechanisms, the error mutation process is locally modeled, and the basis function scale is adaptively adjusted according to the intensity of error change, thereby effectively capturing the rapid change characteristics of error. By dynamically adjusting the center position and scale parameters of the basis functions, the model can focus on error mutation regions on the time axis, improving its responsiveness to local anomalies. Continuous and consistent error prediction results provide stable input for subsequent error compensation, effectively avoiding oscillations caused by prediction mutations during compensation, thereby enhancing the overall stability of the inertial navigation system. Targeted modeling strategies are employed at different error evolution stages, enabling the system to maintain high prediction accuracy and robustness in dynamically changing environments, thus improving the accuracy and long-term stability of inertial navigation positioning results. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the inertial navigation positioning calibration method based on multi-source error compensation according to the present invention. Figure 2 This is a schematic diagram of an inertial navigation positioning calibration system based on multi-source error compensation according to the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0021] Please see Figure 1 As shown, this embodiment provides an inertial navigation positioning calibration method based on multi-source error compensation, specifically including the following steps: S1. Obtain the acceleration and angular velocity data output by the inertial measurement unit, and obtain the position, velocity and attitude information through integration to construct the navigation state variables under a unified time series. S2. The errors in the navigation state variables are characterized and decomposed into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forming a multi-source error superposition relationship and generating an error sequence that evolves over time. S3. Based on the error sequence, perform error evolution prediction modeling and divide the error sequence into steady-state intervals and transition intervals; construct a motion-constrained evolution prediction function in the steady-state interval, construct an error-driven enhancement prediction function in the transition interval, and apply continuity constraints to adjacent intervals to obtain continuous error prediction results. S4. Generate the corresponding compensation amount based on the error prediction result, and compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning result. S5. The compensated navigation and positioning results are used as the navigation state input for the next moment, driving the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning.

[0022] Methods for obtaining acceleration and angular velocity data output by an inertial measurement unit include: The triaxial accelerometer and triaxial gyroscope in the inertial measurement unit are sampled synchronously to obtain the triaxial acceleration measurement value and triaxial angular velocity measurement value of the navigation vehicle, and a unified timestamp is assigned to the triaxial acceleration measurement value and triaxial angular velocity measurement value; The triaxial acceleration and triaxial angular velocity measurements are processed by a sliding time window to obtain smooth triaxial acceleration and triaxial angular velocity sequences. The triaxial acceleration and triaxial angular velocity sequences are then organized according to a unified time index to form a continuous inertial measurement data sequence.

[0023] Methods for constructing navigation state variables under a unified time series include: The attitude of the navigation vehicle is recursively updated based on the three-axis angular velocity sequence. The attitude increment is obtained by integrating the three-axis angular velocity at each sampling time, and the attitude at the current time is updated by combining the initial attitude of the navigation vehicle, resulting in a continuous attitude sequence. It should be noted that in this embodiment, the attitude of the navigation vehicle is recursively updated using the three-axis angular velocity sequence output by the inertial measurement unit. Specifically, the three-axis angular velocity measurement value is acquired at the k-th time moment, and combined with the time interval between adjacent time moments, the angular motion within this time interval is approximated as uniform rotation. By integrating the three-axis angular velocity over time, the attitude increment within this sampling period is obtained. The attitude increment is used to characterize the minute rotational changes of the navigation vehicle around each coordinate axis within the current sampling period. After obtaining the attitude increments corresponding to each time moment, the attitude of the navigation vehicle at the initial time moment is used as the starting point for recursion, and the attitude at subsequent time moments is updated step by step.

[0024] Specifically, the attitude at the current moment is combined with the attitude increment at the corresponding moment to obtain the attitude parameters for the next moment. In a preferred implementation, quaternions are used to represent the attitude. Let the current attitude be a quaternion. An incremental quaternion is constructed based on the attitude increment vector obtained from the integration of the three-axis angular velocities. The rotation axis of the incremental quaternion is determined by the direction of the attitude increment vector, and the rotation angle is determined by the magnitude of the attitude increment vector. Then, the current attitude and the attitude increment are combined through quaternion multiplication to obtain the attitude for the next moment. That is, the updated attitude parameters are obtained by multiplying the current attitude quaternion with the incremental quaternion. The combination operation reflects the rotational superposition relationship of the attitude in three-dimensional space. By repeating the above recursive update process at each moment, the attitude is continuously propagated in the time dimension.

[0025] Based on the attitude sequence, a coordinate transformation relationship is constructed, and the three-axis acceleration sequence is transformed from the carrier coordinate system to the navigation coordinate system. Gravity component compensation is performed on the transformed three-axis acceleration to obtain the equivalent acceleration sequence in the navigation coordinate system. The equivalent acceleration sequence is integrated once in time, and the velocity at each sampling time is recursively calculated using the velocity at the previous time as the initial condition to obtain the velocity sequence corresponding to the time index. It should be noted that, in this embodiment, after obtaining the continuous attitude sequence, a coordinate transformation relationship between the carrier coordinate system and the navigation coordinate system is constructed based on the attitude sequence to realize the conversion of acceleration measurement values ​​between different coordinate systems. Specifically, at each sampling moment, the corresponding attitude parameters are obtained to characterize the spatial orientation relationship between the carrier coordinate system and the navigation coordinate system, and a coordinate transformation matrix from the carrier coordinate system to the navigation coordinate system is generated based on the attitude parameters. When the attitude parameters are represented by quaternions, a corresponding direction cosine matrix is ​​constructed according to the combination relationship between the components of the quaternion, so that the direction cosine matrix satisfies the orthogonality constraint, thereby realizing rotational mapping in three-dimensional space. When attitude parameters are represented using Euler angles, the rotation angles of each axis are converted into corresponding basic rotation matrices according to a preset rotation order, and then combined through matrix multiplication to obtain the overall coordinate transformation matrix. When attitude parameters are represented using direction cosine matrices, these attitude parameters are directly used as the coordinate transformation matrix, thereby obtaining the corresponding coordinate transformation relationship at each sampling time. This is used to map physical quantities in the carrier coordinate system to the navigation coordinate system. This coordinate transformation matrix is ​​used to characterize the spatial orientation relationship of the carrier attitude relative to the navigation coordinate system and is updated synchronously with the time series, thus forming a coordinate transformation sequence that corresponds one-to-one with the attitude sequence.

[0026] After obtaining the coordinate transformation relationships at each sampling time, the triaxial acceleration sequence is mapped from the vehicle coordinate system to the navigation coordinate system. Specifically, for the triaxial acceleration measurement value at any sampling time, a linear transformation operation is performed using the coordinate transformation matrix of the corresponding time, and the value is projected into the navigation coordinate system to obtain the acceleration components in the navigation coordinate system. By performing this transformation process time-by-time on the entire time series, an acceleration sequence in the navigation coordinate system consistent with the time index can be obtained, thereby achieving a unified spatial representation of acceleration data.

[0027] Furthermore, since the acceleration measured by the inertial measurement unit includes a gravitational component, in order to obtain acceleration information that reflects the true motion state of the carrier, gravity component compensation processing is performed on the acceleration in the navigation coordinate system after coordinate transformation. Specifically, based on the preset direction and magnitude of gravitational acceleration in the navigation coordinate system, the corresponding gravity component is subtracted from the acceleration component at each sampling time to eliminate the influence of gravity on the measurement results. Through the above coordinate transformation and gravity compensation processing, an equivalent acceleration sequence that only reflects the acceleration changes caused by the motion of the navigation carrier is obtained.

[0028] For example, at a certain sampling moment, the acceleration in the carrier coordinate system is (2,0,0). When the attitude is rotated 90° around the Z-axis, the acceleration in the navigation coordinate system is (0,2,0) after coordinate transformation. After compensating for the gravity component of this acceleration, the equivalent acceleration is (0,2,-9.8). Based on the velocity at the previous moment (1,0,0), the velocity at the current moment is obtained by time integration as (1,0.2,-0.98).

[0029] The velocity sequence is integralized twice over time, and the position sequence corresponding to the time index is obtained by recursion calculation at each sampling time with the position of the previous time as the initial condition. The attitude sequence, velocity sequence and position sequence are aligned and fused according to a unified time index to construct the navigation state variables under a unified time sequence.

[0030] Methods for establishing multi-source error superposition relationships include: Based on the navigation state quantities under a unified time series, a state deviation sequence corresponding to each sampling time is constructed. The state deviation sequence is calculated from the difference of the navigation state quantities at adjacent sampling times. The navigation state quantities are statistically calculated within a sliding time window of a fixed length to obtain the local mean sequence corresponding to each time window. The local mean sequence is used as a slow-varying trend component, and the slow-varying trend component is defined as zero bias error. Based on the input-output relationship of navigation state variables in the integral evolution process, for any sampling moment, the equivalent acceleration in the navigation coordinate system is taken as the driving quantity, and the velocity obtained by the corresponding integration is taken as the output quantity. A proportional mapping relationship between the output quantity and the driving quantity is constructed. By calculating the deviation between the velocity increment at the current moment and the corresponding acceleration integration result, the deviation is determined as the proportional factor error. For example, at a certain moment: the lower edge of the navigation coordinate system Equivalent acceleration components in the axial direction: The time interval between two adjacent moments (sampling period): The previous moment at the lower edge of the navigation coordinate system Velocity in the axial direction: ;in, Indicates the first The previous moment; The theoretical velocity increment is obtained by integrating the acceleration: The theoretical velocity obtained based on the integral of acceleration is: ;in, express At that moment; Assuming the current moment is at the lower edge of the navigation coordinate system Actual velocity in the axial direction: Actual speed increment: Then, the scaling factor (the ratio between actual output and theoretical output) ; Scale factor error: ;in, It represents the ideal ratio; under ideal error-free conditions, the scale factor equals 1; by calculating the difference between the scale factor and the unit ratio, the scale factor error is obtained, which is used to characterize the degree of deviation of the actual output from the ideal output.

[0031] Using triaxial angular velocity, triaxial acceleration, and attitude parameters as basic variables, a multivariate expression relationship containing the cross product term between triaxial angular velocity and triaxial acceleration is constructed during the transformation of triaxial acceleration from the carrier coordinate system to the navigation coordinate system. By calculating the difference between the coordinate transformation results at the current time and the previous time, the acceleration additional term caused by the change in angular velocity is extracted, and this additional term is determined as the dynamic coupling error. For example, at the same moment: angular velocity (rotation about the Z-axis): Triaxial acceleration (forward) in the carrier coordinate system: The time interval between two adjacent sampling moments (sampling period): ; Acceleration in the navigation coordinate system calculated based on the attitude at the previous moment: ; Calculate the cross term of angular velocity and acceleration: Multiplication time: Acceleration in the navigation coordinate system calculated based on the current attitude: Dynamic coupling error: ; When the carrier rotates around one axis and accelerates along another axis, the coordinate system changes over time, causing the acceleration that was originally in a single direction to generate an additional component in the navigation coordinate system. This additional component can be obtained by the difference between the acceleration transformation results corresponding to the current attitude and the attitude at the previous moment, and this difference is used as the dynamic coupling error.

[0032] After detrending the state deviation sequence, the remaining residual sequence is extracted, and the part of the residual sequence that does not meet the expression form of zero bias error, scale factor error and dynamic coupling error is defined as environmental disturbance error. The zero bias error, scale factor error, dynamic coupling error and environmental disturbance error are superimposed according to a unified time index to form a multi-source error superposition relationship of navigation state quantity.

[0033] Methods for generating error sequences that evolve over time include: Based on a unified time index, the zero bias error, scaling factor error, dynamic coupling error, and environmental disturbance error are aligned at each time point to construct the time series corresponding to each type of error component. The error components at the same time point are superimposed to obtain the total error vector at the corresponding time point. The total error vector at each time point is organized and arranged in chronological order to form an error sequence that evolves over time.

[0034] Methods for dividing the error sequence into steady-state and transition intervals include: After obtaining the error sequence that evolves over time, the error values ​​corresponding to each time point are extracted according to a unified time index to construct an evolution analysis sequence of error over time. Second-order difference calculation is performed on the error sequence, and gradient change information between multi-source errors is extracted based on the spatial distribution relationship of various error components at the same time point to characterize the degree of coupling change between different error components. It should be noted that the multi-source error vector is considered as a discrete function defined in the error component index space, where the error component category is the discrete independent variable and the error value is the function value. This maps the distribution relationship of multi-source errors at the same time point into a one-dimensional or multi-dimensional discrete field. Based on this discrete field structure, the numerical changes between adjacent error components are differentially calculated to obtain the local changes between each error component, which are used to characterize the gradient of change between different error sources.

[0035] Furthermore, for any error component, a set of relative changes among the error components is constructed by calculating the difference between it and other error components. This set of changes is then normalized to eliminate the influence of differences in the dimensions and numerical ranges of different error components on the gradient calculation results. After normalization, the difference results among the error components are combined to form a gradient vector describing the overall trend of multi-source error changes.

[0036] In a preferred implementation, the difference results between each error component are weighted and combined according to a preset weight to highlight the error component that has a greater impact on the system, thereby obtaining a comprehensive gradient index that reflects the intensity of the coupling change of multi-source errors; or, by calculating the norm of the gradient vector, it is converted into a scalar form to quantify the overall change intensity of multi-source errors at the current moment; through the above process, the gradient change information between multi-source errors is extracted based on the spatial distribution relationship of various error components at the same sampling moment.

[0037] Second-order difference calculation: ;in, It represents the second-order difference component of the error sequence in the time dimension, and is used to characterize the acceleration characteristics of error change, that is, the degree of change of the error rate. This represents the error value two sampling periods after the current moment; This represents the error value one sampling period after the current moment; This represents the time interval between adjacent sampling moments, i.e., the sampling period; Indicates time The error value at any given time originates from the error sequence that has been constructed over time. The second-order difference result is combined with the gradient change information to construct an error structure criterion function, and the value of the error structure criterion function is obtained. Based on the comparison result between the error structure criterion function value and the preset interval division threshold, the error state at each time is determined. When the error structure criterion function value is less than the preset interval division threshold, the corresponding time is divided into a steady state interval. When the preset interval division threshold is greater than or equal to the preset interval division threshold, the corresponding time is divided into a transition interval.

[0038] Error structure criterion function: ;in, Indicates time The error structure criterion function value at time t is used to comprehensively characterize the intensity of error change and the characteristics of error structure change; It represents the absolute value of the second-order difference component, used to reflect the drastic degree of error change; This represents the weighting coefficient, used to adjust the relative influence between time-varying characteristics and multi-source error structure variation characteristics; Indicates time The multi-source errors at any given time include zero bias error, scaling factor error, dynamic coupling error, and environmental disturbance error; It represents the gradient information of multi-source errors at the current time, and is used to characterize the differences in change and spatial coupling relationship between each error component; The norm of the error gradient is used to quantify the overall change intensity of multi-source errors; the weighting coefficients and preset interval division thresholds can be preset according to the sensor accuracy and application scenario or obtained through experimental calibration.

[0039] Methods for obtaining continuous error prediction results include: After dividing the error sequence into steady-state and transition intervals, the intervals are organized in chronological order, and corresponding error prediction functions are constructed for the differences in error evolution characteristics in different intervals. In the steady-state interval, an evolution prediction function containing time polynomial terms is constructed, and motion constraint terms related to navigation state changes are introduced. By coupling error changes with changes in carrier motion state, the prediction of the error change process can be achieved. Evolutionary prediction function: ;in, Indicates time Predicted error value within the steady-state interval at any given time; Represents the polynomial coefficients, used to characterize the baseline level of the error in the steady-state region and its trend over time; The initial bias term represents the error; The coefficient representing the first-order trend of error; The coefficient representing the second-order trend of error; This represents the filtered navigation state variables (including at least one of position, velocity, or attitude). It represents the rate of change of navigation state quantities over time, used to reflect changes in the motion state of the vehicle; This represents the motion constraint weighting coefficient, which is used to adjust the degree of influence of changes in motion state on the error prediction results. Within the transition interval, an enhanced prediction function driven by the error magnitude and its variation characteristics is constructed. The transition process is characterized piecewise by setting basis functions, and the range of action of the basis functions is adaptively adjusted according to the error variation intensity at each time step. Enhanced prediction function: ;in, Indicates time Predicted error values ​​within the time transition interval; This represents the number of basis functions used to characterize the transition process; Indicates the base function index; Indicates the first The weight coefficients of each basis function are used to adjust the contribution of that basis function to the overall prediction result; Indicates the first A basis function is used to describe the variation characteristics of the error over a local time range; Indicates the first The center time corresponding to each basis function is used to locate the position of the basis function on the time axis; Indicates the first The scaling parameter of each basis function is used to control the range of influence of that basis function on the time axis; This represents an exponential function, used to construct locally enhanced responses; Represents a small positive constant to prevent the denominator from being zero; After constructing prediction functions for different intervals, a continuity constraint is applied to adjacent intervals to ensure that the function values ​​at the interval boundary points remain consistent and that the first-order change trend of the function at that point remains continuous. This ensures a smooth transition in the numerical values ​​and change trends of the prediction results for different intervals, ultimately yielding error prediction results that change continuously throughout the entire time series.

[0040] Continuity constraints: ;in, This represents the function value of the evolution prediction function at the interval boundary point; This represents the function value of the enhanced prediction function at the interval boundary point; This represents the first derivative of the evolution prediction function at the interval boundary point; This represents the first derivative of the enhanced prediction function at the interval boundary point; This indicates the moment when the point approaches the boundary of the interval from one side of the steady-state interval. This indicates the moment when the point approaches the boundary of the transition interval from one side. Methods for outputting compensated navigation and positioning results include: After obtaining the error prediction results at each time point, the error prediction values ​​at the corresponding time point are extracted and quantized according to the error type to obtain a set of error components corresponding to the navigation state variables. Based on the set of error components, the error components are mapped to the position, velocity and attitude parameters in the navigation state variables to construct an error vector that corresponds one-to-one with each state parameter. The compensation amount is generated based on the error vector. The direction of the compensation amount is opposite to that of the corresponding error component, and its magnitude is determined by the error prediction value, so as to cancel the error. For position parameters and velocity parameters, the compensation amount is superimposed on the original position parameter and velocity parameter state variables in the corresponding coordinate system for correction, so as to obtain the compensated position parameters and velocity parameters. For attitude parameters, quaternion-based compensation is used. By constructing a compensation quaternion corresponding to the attitude error and combining it with the current attitude quaternion, the compensated attitude parameters are obtained. After completing the compensation and correction of position, velocity, and attitude, the various state parameters are integrated according to a unified time index to form the compensated navigation state variables, and the corresponding navigation and positioning results are output.

[0041] Methods for achieving continuous dynamic calibration of inertial navigation positioning include: After the compensation and correction of the navigation state quantity is completed at the current sampling time, the compensated navigation state quantity is obtained and used as the initial state input for the navigation solution at the next sampling time. At the next sampling time, the updated navigation state quantity is obtained by integrating the acceleration and angular velocity data output by the inertial measurement unit and the initial state. The updated navigation state variables are repeatedly subjected to error characterization and decomposition processing to extract zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, and generate corresponding error sequences. Based on the error sequences, error evolution characteristic analysis and interval division are further performed to determine the steady state interval and transition interval of the error state. Based on the interval determination results, the corresponding evolution prediction function or enhancement prediction function is called to perform error prediction, obtain the error prediction result at the current time, and generate a new compensation amount based on the error prediction result; the new compensation amount is applied to the navigation state quantity at the current time to complete the compensation correction; by repeating the process at each time, the compensated navigation state quantity is continuously used as the navigation state quantity input at the next time, thereby forming a closed-loop iteration of error representation, prediction modeling and compensation correction, realizing the continuous dynamic calibration of inertial navigation and positioning results in the time dimension.

[0042] In this embodiment, by mapping multi-source errors to discrete fields in the error component space and extracting their gradient change information, different types of errors are expressed under a unified framework, thereby improving the completeness and consistency of error modeling. Furthermore, by introducing the second-order difference component of the error sequence, the changes in the error rate are effectively reflected, enabling the identification of accelerated or abrupt changes in errors and improving the ability to perceive non-stationary errors. By simultaneously introducing second-order difference components and error gradient information, a multi-dimensional judgment basis is formed, reducing the uncertainty caused by a single feature and improving the stability and anti-interference ability of interval division. By structurally dividing the error sequence, a basis is provided for adopting different error prediction and compensation strategies in different intervals, thereby improving the overall error compensation effect. By more accurately identifying the error evolution state and performing targeted processing, error accumulation is effectively suppressed, and the accuracy and long-term stability of navigation results in complex environments are improved.

[0043] By dividing the error sequence into steady-state and transition intervals and constructing differentiated prediction functions in different intervals, the model can adaptively process different error stages, improving overall prediction accuracy. In the steady-state interval, a low-order model based on time polynomials is introduced, and motion constraint terms for navigation state changes are superimposed to achieve coupled modeling between error and carrier motion state, thereby improving the fitting ability for slowly changing errors. In the transition interval, by introducing basis function superposition and exponential enhancement mechanisms, the error mutation process is locally modeled, and the basis function scale is adaptively adjusted according to the intensity of error change, thereby effectively capturing the rapid change characteristics of error. By dynamically adjusting the center position and scale parameters of the basis functions, the model can focus on error mutation regions on the time axis, improving its responsiveness to local anomalies. Continuous and consistent error prediction results provide stable input for subsequent error compensation, effectively avoiding oscillations caused by prediction mutations during compensation, thereby enhancing the overall stability of the inertial navigation system. Targeted modeling strategies are employed at different error evolution stages, enabling the system to maintain high prediction accuracy and robustness in dynamically changing environments, thus improving the accuracy and long-term stability of inertial navigation positioning results. Example

[0044] Please see Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. An inertial navigation positioning calibration system based on multi-source error compensation is provided, including: The inertial navigation construction module is used to acquire the acceleration and angular velocity data output by the inertial measurement unit, and obtain position, velocity and attitude information through integration calculations to construct navigation state variables under a unified time series. The multi-source error characterization module is used to characterize and decompose the errors in navigation state variables. It divides the errors into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forms a multi-source error superposition relationship, and generates an error sequence that evolves over time. The segmented prediction modeling module is used to perform error evolution prediction modeling based on the error sequence, and divide the error sequence into steady-state intervals and transition intervals; within the steady-state interval, an evolution prediction function constrained by motion is constructed, and within the transition interval, an error-driven enhancement prediction function is constructed, and continuity constraints are applied to adjacent intervals to obtain continuous error prediction results. The error compensation and correction module is used to generate corresponding compensation amounts based on the error prediction results, and to compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning results. The calibration feedback update module is used to take the compensated navigation and positioning results as the input for the next moment, drive the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning. Example

[0045] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the inertial navigation positioning calibration method based on multi-source error compensation described above.

[0046] Since the electronic device described in this embodiment is the one used in implementing the inertial navigation positioning calibration method and system based on multi-source error compensation described in this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the inertial navigation positioning calibration method and system based on multi-source error compensation described in this application. Therefore, how the electronic device implements the method in this application will not be described in detail here. Any electronic device used by those skilled in the art in implementing the inertial navigation positioning calibration method and system based on multi-source error compensation described in this application falls within the scope of protection of this application.

[0047] It should be noted that all formulas in this manual are calculated by removing dimensions and taking their numerical values. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0048] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An inertial navigation positioning calibration method based on multi-source error compensation, characterized in that, include: S1. Obtain the acceleration and angular velocity data output by the inertial measurement unit, and obtain the position, velocity and attitude information through integration to construct the navigation state variables under a unified time series. S2. The errors in the navigation state variables are characterized and decomposed into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forming a multi-source error superposition relationship and generating an error sequence that evolves over time. S3. Based on the error sequence, perform error evolution prediction modeling and divide the error sequence into steady-state intervals and transition intervals; construct a motion-constrained evolution prediction function in the steady-state interval, construct an error-driven enhancement prediction function in the transition interval, and apply continuity constraints to adjacent intervals to obtain continuous error prediction results. S4. Generate the corresponding compensation amount based on the error prediction result, and compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning result. S5. The compensated navigation and positioning results are used as the navigation state input for the next moment, driving the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning.

2. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 1, characterized in that, The method for acquiring the acceleration and angular velocity data output by the inertial measurement unit includes: The triaxial accelerometer and triaxial gyroscope in the inertial measurement unit are sampled synchronously to obtain the triaxial acceleration measurement value and triaxial angular velocity measurement value of the navigation vehicle, and a unified timestamp is assigned to the triaxial acceleration measurement value and triaxial angular velocity measurement value; The triaxial acceleration and triaxial angular velocity measurements are processed by a sliding time window to obtain smooth triaxial acceleration and triaxial angular velocity sequences. The triaxial acceleration and triaxial angular velocity sequences are then organized according to a unified time index to form a continuous inertial measurement data sequence.

3. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 2, characterized in that, The method for constructing navigation state quantities under a unified time series includes: The attitude of the navigation vehicle is recursively updated based on the three-axis angular velocity sequence. The attitude increment is obtained by integrating the three-axis angular velocity at each sampling time, and the attitude at the current time is updated by combining the initial attitude of the navigation vehicle, resulting in a continuous attitude sequence. Based on the attitude sequence, a coordinate transformation relationship is constructed, and the three-axis acceleration sequence is transformed from the carrier coordinate system to the navigation coordinate system. Gravity component compensation is performed on the transformed three-axis acceleration to obtain the equivalent acceleration sequence in the navigation coordinate system. The equivalent acceleration sequence is integrated once in time, and the velocity at each sampling time is recursively calculated using the velocity at the previous time as the initial condition to obtain the velocity sequence corresponding to the time index. The velocity sequence is integralized twice over time, and the position sequence corresponding to the time index is obtained by recursion calculation at each sampling time with the position of the previous time as the initial condition. The attitude sequence, velocity sequence and position sequence are aligned and fused according to a unified time index to construct the navigation state variables under a unified time sequence.

4. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 3, characterized in that, The method for forming the multi-source error superposition relationship includes: Based on the navigation state quantities under a unified time series, a state deviation sequence corresponding to each sampling time is constructed. The state deviation sequence is calculated from the difference of the navigation state quantities at adjacent sampling times. The navigation state quantities are statistically calculated within a sliding time window of a fixed length to obtain the local mean sequence corresponding to each time window. The local mean sequence is used as a slow-varying trend component, and the slow-varying trend component is defined as zero bias error. Based on the input-output relationship of navigation state variables in the integral evolution process, for any sampling moment, the equivalent acceleration in the navigation coordinate system is taken as the driving quantity, and the velocity obtained by the corresponding integration is taken as the output quantity. A proportional mapping relationship between the output quantity and the driving quantity is constructed. By calculating the deviation between the velocity increment at the current moment and the corresponding acceleration integration result, the deviation is determined as the proportional factor error. Using triaxial angular velocity, triaxial acceleration, and attitude parameters as basic variables, a multivariate expression relationship containing the cross product term between triaxial angular velocity and triaxial acceleration is constructed during the transformation of triaxial acceleration from the carrier coordinate system to the navigation coordinate system. By calculating the difference between the coordinate transformation results at the current time and the previous time, the acceleration additional term caused by the change in angular velocity is extracted, and this additional term is determined as the dynamic coupling error. After detrending the state deviation sequence, the remaining residual sequence is extracted, and the part of the residual sequence that does not meet the expression form of zero bias error, scale factor error and dynamic coupling error is defined as environmental disturbance error. The zero bias error, scale factor error, dynamic coupling error and environmental disturbance error are superimposed according to a unified time index to form a multi-source error superposition relationship of navigation state quantity.

5. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 4, characterized in that, The method for generating the error sequence that evolves over time includes: Based on a unified time index, the zero bias error, scaling factor error, dynamic coupling error, and environmental disturbance error are aligned at each time point to construct the time series corresponding to each type of error component. The error components at the same time point are superimposed to obtain the total error vector at the corresponding time point. The total error vector at each time point is organized and arranged in chronological order to form an error sequence that evolves over time.

6. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 5, characterized in that, The method for dividing the error sequence into steady-state intervals and transition intervals includes: After obtaining the error sequence that evolves over time, the error values ​​corresponding to each time point are extracted according to a unified time index to construct an evolution analysis sequence of error over time. Second-order difference calculation is performed on the error sequence, and gradient change information between multi-source errors is extracted based on the spatial distribution relationship of various error components at the same time point to characterize the degree of coupling change between different error components. The second-order difference result is combined with the gradient change information to construct an error structure criterion function, and the value of the error structure criterion function is obtained. Based on the comparison result between the error structure criterion function value and the preset interval division threshold, the error state at each time is determined. When the error structure criterion function value is less than the preset interval division threshold, the corresponding time is divided into a steady state interval. When the preset interval division threshold is greater than or equal to the preset interval division threshold, the corresponding time is divided into a transition interval.

7. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 6, characterized in that, The method for obtaining continuous error prediction results includes: After dividing the error sequence into steady-state and transition intervals, the intervals are organized in chronological order, and corresponding error prediction functions are constructed for the differences in error evolution characteristics in different intervals. In the steady-state interval, an evolution prediction function containing time polynomial terms is constructed, and motion constraint terms related to navigation state changes are introduced. By coupling error changes with changes in carrier motion state, the prediction of the error change process can be achieved. Within the transition interval, an enhanced prediction function driven by the error magnitude and its variation characteristics is constructed. The transition process is characterized piecewise by setting basis functions, and the range of action of the basis functions is adaptively adjusted according to the error variation intensity at each time step. After constructing prediction functions for different intervals, a continuity constraint is applied to adjacent intervals to ensure that the function values ​​at the interval boundary points remain consistent and that the first-order change trend of the function at that point remains continuous. This ensures a smooth transition in the numerical values ​​and change trends of the prediction results for different intervals, ultimately yielding error prediction results that change continuously throughout the entire time series.

8. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 7, characterized in that, The method for outputting the compensated navigation and positioning results includes: After obtaining the error prediction results at each time point, the error prediction values ​​at the corresponding time point are extracted and quantized according to the error type to obtain a set of error components corresponding to the navigation state variables. Based on the set of error components, the error components are mapped to the position, velocity and attitude parameters in the navigation state variables to construct an error vector that corresponds one-to-one with each state parameter. The compensation amount is generated based on the error vector. The direction of the compensation amount is opposite to that of the corresponding error component, and its magnitude is determined by the error prediction value, so as to cancel the error. For position parameters and velocity parameters, the compensation amount is superimposed on the original position parameter and velocity parameter state variables in the corresponding coordinate system for correction, so as to obtain the compensated position parameters and velocity parameters. For attitude parameters, quaternion-based compensation is used. By constructing a compensation quaternion corresponding to the attitude error and combining it with the current attitude quaternion, the compensated attitude parameters are obtained. After completing the compensation and correction of position, velocity, and attitude, the various state parameters are integrated according to a unified time index to form the compensated navigation state variables, and the corresponding navigation and positioning results are output.

9. The inertial navigation positioning calibration method based on multi-source error compensation according to claim 8, characterized in that, The method for achieving continuous dynamic calibration of inertial navigation positioning includes: After the compensation and correction of the navigation state quantity is completed at the current sampling time, the compensated navigation state quantity is obtained and used as the initial state input for the navigation solution at the next sampling time. At the next sampling time, the updated navigation state quantity is obtained by integrating the acceleration and angular velocity data output by the inertial measurement unit and the initial state. The updated navigation state variables are repeatedly subjected to error characterization and decomposition processing to extract zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, and generate corresponding error sequences. Based on the error sequences, error evolution characteristic analysis and interval division are further performed to determine the steady state interval and transition interval of the error state. Based on the interval determination results, the corresponding evolution prediction function or enhancement prediction function is called to perform error prediction, obtain the error prediction result at the current time, and generate a new compensation amount based on the error prediction result; the new compensation amount is applied to the navigation state quantity at the current time to complete the compensation correction; by repeating the process at each time, the compensated navigation state quantity is continuously used as the navigation state quantity input at the next time, thereby forming a closed-loop iteration of error representation, prediction modeling and compensation correction, realizing the continuous dynamic calibration of inertial navigation and positioning results in the time dimension.

10. An inertial navigation positioning calibration system based on multi-source error compensation, used to implement the inertial navigation positioning calibration method based on multi-source error compensation as described in any one of claims 1 to 9, characterized in that, include: The inertial navigation construction module is used to acquire the acceleration and angular velocity data output by the inertial measurement unit, and obtain position, velocity and attitude information through integration calculations to construct navigation state variables under a unified time series. The multi-source error characterization module is used to characterize and decompose the errors in navigation state variables. It divides the errors into zero bias error, scaling factor error, dynamic coupling error and environmental disturbance error, forms a multi-source error superposition relationship, and generates an error sequence that evolves over time. The segmented prediction modeling module is used to perform error evolution prediction modeling based on the error sequence, and divide the error sequence into steady-state intervals and transition intervals; within the steady-state interval, an evolution prediction function constrained by motion is constructed, and within the transition interval, an error-driven enhancement prediction function is constructed, and continuity constraints are applied to adjacent intervals to obtain continuous error prediction results. The error compensation and correction module is used to generate corresponding compensation amounts based on the error prediction results, and to compensate and correct the position, velocity and attitude in the navigation state quantities, and output the compensated navigation and positioning results. The calibration feedback update module is used to take the compensated navigation and positioning results as the input for the next moment, drive the cyclic execution of error characterization, prediction modeling and compensation correction, thereby realizing continuous dynamic calibration of inertial navigation and positioning.

Images