Dynamic constraint method and device for vehicle-mounted GNSS / INS combination

Abstract

The invention discloses a dynamic constraint method and device for a vehicle-mounted GNSS / INS (Global Navigation Satellite System / Inertial Navigation System) combination, and relates to the technical field of vehicle-mounted navigation. According to the method, the motion mode represented by the vehicle-mounted real-time monitoring data can be identified and captured through the optimal displacement prediction model, so that more accurate inertial navigation center prediction displacement data can be obtained, and the inaccuracy caused by directly using IMU data for constraint in the prior art is avoided; according to the method, a constraint type more suitable for the vehicle-mounted GNSS / INS combination can be identified through predicted inertial navigation center displacement data, then error accumulation and divergence of the vehicle-mounted GNSS / INS combination are better limited through a measurement equation, the anti-interference capability of the vehicle-mounted GNSS / INS combination and the re-initialization speed after interference are improved, and thus the accuracy of the vehicle-mounted GNSS / INS combination is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle-mounted navigation, and particularly relates to a dynamic constraint method and device for vehicle-mounted GNSS / INS combination. Background Art

[0002] With the rapid development of fields such as autonomous driving and intelligent driverless driving, the integrated navigation system based on GNSS / INS combination has become the main way to realize vehicle-mounted navigation. The GNSS / INS combination is the combination of the GNSS system (Global Navigation Satellite System) and the INS system (Inertial Navigation System). In a vehicle-mounted environment, satellite signals are vulnerable to interference and occlusion, and there are errors in the measurement accuracy of the IMU (Inertial Measurement Unit) in the inertial navigation system, both of which lead to a reduction in the accuracy of the GNSS / INS combination.

[0003] Currently, the motion information of the vehicle is usually used as a virtual measurement value to constrain the GNSS / INS system, that is, the integrated navigation method assisted by motion constraints. Common ones include zero-velocity constraint, lateral velocity constraint, vertical velocity constraint, etc. However, due to the influence of the accuracy of motion state monitoring, for example, in the zero-velocity constraint, affected by IMU error drift, inaccurate installation position and angle calibration in the vehicle body, road environment, etc., the IMU data is inaccurate, which affects the accuracy of the constraint, and further reduces the accuracy of the GNSS / INS combination. Therefore, there is an urgent need for a vehicle-mounted GNSS / INS combination dynamic constraint method and device to solve the defects of the existing technology. Summary of the Invention

[0004] The present invention aims to provide a vehicle-mounted GNSS / INS combination dynamic constraint method and device to solve the above technical problems. By predicting the displacement of the inertial navigation center, the dynamic constraint of the vehicle-mounted GNSS / INS combination is realized, and the accuracy of the vehicle-mounted GNSS / INS combination is improved.

[0005] To solve the above technical problems, an embodiment of the present invention provides a vehicle-mounted GNSS / INS combination dynamic constraint method, including:

[0006] Collect the actual motion data of the vehicle, and perform calculation on the actual motion data to obtain vehicle-mounted real-time monitoring data;

[0007] Train a preset displacement prediction model according to the vehicle-mounted real-time monitoring data to obtain an optimal displacement prediction model, and obtain inertial navigation center predicted displacement data based on the optimal displacement prediction model;

[0008] Obtain the constraint type according to the predicted displacement data of the inertial navigation center, and construct a measurement equation according to the constraint type;

[0009] Update the vehicle-mounted GNSS / INS combination according to the measurement equation to complete the dynamic constraint of the vehicle-mounted GNSS / INS combination.

[0010] It can be understood that, compared with the prior art, the present invention calculates the measured motion data of the vehicle to obtain the vehicle-mounted real-time monitoring data, and then trains the preset displacement prediction model with the vehicle-mounted real-time monitoring data, so as to obtain the predicted displacement data of the inertial navigation center with the optimal displacement prediction model. Furthermore, a measurement equation is constructed based on the predicted displacement data of the inertial navigation center, realizing the dynamic constraint of the vehicle-mounted GNSS / INS combination. The present invention can identify and capture the motion patterns shown by the vehicle-mounted real-time monitoring data through the optimal displacement prediction model, and thus can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art; the predicted displacement data of the inertial navigation center can identify the constraint type more suitable for the vehicle-mounted GNSS / INS combination, and then better limit the error accumulation and divergence of the vehicle-mounted GNSS / INS combination through the measurement equation, improving the anti-interference ability and the re-initialization speed after interference of the vehicle-mounted GNSS / INS combination, thereby improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0011] As a preferred solution, collecting the measured motion data of the vehicle and calculating the measured motion data to obtain the vehicle-mounted real-time monitoring data specifically includes:

[0012] Collect the measured motion data of the vehicle, and the measured motion data includes: IMU data and GNSS data;

[0013] Perform state recursion on the IMU data to obtain the INS velocity variance at the current moment;

[0014] Calculate the phase residual at the current moment and the pseudorange residual at the current moment according to the GNSS data;

[0015] Determine the vehicle-mounted real-time monitoring data according to the IMU data, GNSS data, the INS velocity variance at the current moment, the phase residual at the current moment, and the pseudorange residual at the current moment.

[0016] This preferred solution calculates the vehicle-mounted real-time monitoring data by calculating the measured motion data of the vehicle, so as to accurately represent the real-time motion data of the vehicle through the vehicle-mounted real-time monitoring data, improving the prediction accuracy of the subsequent predicted displacement data of the inertial navigation center, and further improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0017] As a preferred solution, training the preset displacement prediction model based on the vehicle-mounted real-time monitoring data to obtain an optimal displacement prediction model, and obtaining the predicted displacement data of the inertial navigation center based on the optimal displacement prediction model specifically includes:

[0018] Obtaining the position sequence of the vehicle, and converting the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center north-east-down displacement vector sequence;

[0019] Using the IMU center north-east-down displacement vector sequence as the reference output of the preset displacement prediction model, and using the vehicle-mounted real-time monitoring data as the input parameter of the preset displacement prediction model to train the preset displacement prediction model to obtain an optimal displacement prediction model;

[0020] Inputting the vehicle-mounted real-time monitoring data into the optimal displacement prediction model to obtain the predicted displacement data of the inertial navigation center.

[0021] This preferred solution trains the preset displacement prediction model, so that the optimal displacement prediction model can identify and capture the motion patterns shown by the vehicle-mounted real-time monitoring data, and then can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art and improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0022] As a preferred solution, the obtaining the position sequence of the vehicle, and converting the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center north-east-down displacement vector sequence specifically includes:

[0023] Obtaining the position sequence of the vehicle, and converting the position sequence according to a preset attitude rotation matrix and the position of the reference point of the preset vehicle body coordinate system relative to the IMU center to obtain the IMU center position vector sequence; wherein, the formula for converting the position sequence to obtain the IMU center position vector sequence is specifically:

[0024]

[0025] wherein, p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, and l is the position of the reference point of the vehicle body coordinate system relative to the IMU center;

[0026] Performing a time dimension calculation on the IMU center position vector sequence to obtain a calculation result, and performing a coordinate system conversion on the calculation result to obtain the IMU center north-east-down displacement vector sequence.

[0027] This preferred solution obtains the IMU center's northeast-down displacement vector sequence by transforming the position sequence, thereby realizing the training of the preset displacement prediction model, obtaining the optimal displacement prediction model, and further being able to obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art and improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0028] As a preferred solution, obtaining the constraint type according to the predicted displacement data of the inertial navigation center and constructing the measurement equation according to the constraint type specifically includes:

[0029] The predicted displacement data of the inertial navigation center includes: the predicted northeast-down displacement of the inertial navigation center;

[0030] Judge the predicted northeast-down displacement of the inertial navigation center;

[0031] When the judgment result conforms to the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination, construct the observation residual corresponding to the straight-line constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the straight-line constraint;

[0032] When the judgment result conforms to the second preset result, perform a zero-velocity constraint on the vehicle-mounted GNSS / INS combination; and construct the observation residual corresponding to the zero-velocity constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the zero-velocity constraint.

[0033] This preferred solution can identify a more suitable constraint type for the vehicle-mounted GNSS / INS combination through the predicted displacement data of the inertial navigation center, and then better limit the error accumulation and divergence of the vehicle-mounted GNSS / INS combination through the measurement equation, improving the anti-interference ability of the vehicle-mounted GNSS / INS combination and the re-initialization speed after interference, thereby improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0034] As a preferred solution, when the judgment result conforms to the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination, construct the observation residual corresponding to the straight-line constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the straight-line constraint, specifically including:

[0035] When the judgment result conforms to the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination and obtain the straight-line constraint observation variance according to the straight-line constraint;

[0036] Obtain the position of the inertial navigation center in the Earth-centered Earth-fixed (ECEF) coordinate system at adjacent times, the position error of the inertial navigation center in the ECEF coordinate system, and the transformation matrix from the ECEF coordinate system of the inertial navigation center to the local coordinate system. Then, combine the predicted north-east-down displacement of the inertial navigation center and the observation variance of the straight-line constraint to construct the observation residual corresponding to the straight-line constraint, so as to determine the measurement equation corresponding to the straight-line constraint. Among them, the observation residual corresponding to the straight-line constraint is specifically:

[0037]

[0038] where l LC is the observation residual corresponding to the straight-line constraint, de, dn, and du are the predicted north-east-down displacements of the inertial navigation center, and are the positions of the inertial navigation center in the ECEF coordinate system at time k and time k - 1 respectively, is the transformation matrix from the ECEF coordinate system of the inertial navigation center to the local coordinate system, and are the position errors of the inertial navigation center in the ECEF coordinate system, and R LC is the observation variance of the straight-line constraint.

[0039] In this preferred solution, the observation residual corresponding to the straight-line constraint is constructed through the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the straight-line constraint is determined, which better restricts the error accumulation and divergence of the vehicle-mounted GNSS / INS combination, improves the anti-interference ability and the re-initialization speed after interference of the vehicle-mounted GNSS / INS combination, and thus improves the accuracy of the vehicle-mounted GNSS / INS combination.

[0040] As a preferred solution, when the judgment result meets the second preset result, perform zero-velocity constraint on the vehicle-mounted GNSS / INS combination; and construct the observation residual corresponding to the zero-velocity constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the zero-velocity constraint, which specifically includes:

[0041] When the judgment result meets the second preset result, perform zero-velocity constraint on the vehicle-mounted GNSS / INS combination;

[0042] Construct the observation variance corresponding to the zero-velocity constraint according to the predicted north-east-down displacement of the inertial navigation center;

[0043] Obtain the inertial navigation center's recursive vehicle speed and the speed error of the inertial navigation center in the ECEF coordinate system;

[0044] Construct the observation residual corresponding to the zero-velocity constraint according to the observation variance corresponding to the zero-velocity constraint, the inertial navigation center's recursive vehicle speed, and the speed error of the inertial navigation center in the ECEF coordinate system, and then determine the measurement equation corresponding to the zero-velocity constraint; among them, the observation residual corresponding to the zero-velocity constraint is specifically:

[0045]

[0046] where l ZUPT is the observation residual corresponding to the zero-velocity constraint, ν ins is the velocity of the carrier recursively estimated at the inertial center, and is the velocity error of the inertial center in the Earth-centered Earth-fixed coordinate system, and R ZUPT is the observation variance corresponding to the zero-velocity constraint.

[0047] In this preferred solution, the observation residual corresponding to the zero-velocity constraint is constructed from the predicted displacement data of the inertial center, and then the measurement equation corresponding to the zero-velocity constraint is determined, which better restricts the error accumulation and divergence of the vehicle-mounted GNSS / INS combination, improves the anti-interference ability of the vehicle-mounted GNSS / INS combination and the re-initialization speed after interference, and thus improves the accuracy of the vehicle-mounted GNSS / INS combination.

[0048] Correspondingly, an embodiment of the present invention provides a dynamic constraint device for a vehicle-mounted GNSS / INS combination, including: a vehicle-mounted real-time monitoring data acquisition module, an inertial center predicted displacement data acquisition module, a measurement equation calculation module, and a dynamic constraint module;

[0049] where the vehicle-mounted real-time monitoring data acquisition module is used to collect the actual measured data of the vehicle's movement and perform calculations on the actual measured data of the movement to obtain vehicle-mounted real-time monitoring data;

[0050] The inertial center predicted displacement data acquisition module is used to train a preset displacement prediction model according to the vehicle-mounted real-time monitoring data, obtain an optimal displacement prediction model, and obtain inertial center predicted displacement data based on the optimal displacement prediction model;

[0051] The measurement equation calculation module is used to obtain a constraint type according to the inertial center predicted displacement data and construct a measurement equation according to the constraint type;

[0052] The dynamic constraint module is used to update the vehicle-mounted GNSS / INS combination according to the measurement equation to complete the dynamic constraint of the vehicle-mounted GNSS / INS combination.

[0053] As a preferred solution, the inertial center predicted displacement data acquisition module includes: an inertial center predicted displacement data acquisition unit;

[0054] The inertial center predicted displacement data acquisition unit is used to obtain the position sequence of the vehicle and perform conversion on the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center north-east-down displacement vector sequence;

[0055] Taking the IMU center northeast-up displacement vector sequence as the reference output of the preset displacement prediction model and the vehicle real-time monitoring data as the input parameters of the preset displacement prediction model, training the preset displacement prediction model to obtain the optimal displacement prediction model;

[0056] Inputting the vehicle real-time monitoring data into the optimal displacement prediction model to obtain the predicted displacement data of the inertial navigation center.

[0057] As a preferred solution, the predicted displacement data acquisition unit of the inertial navigation center includes: an IMU center northeast-up displacement vector sequence acquisition subunit;

[0058] The IMU center northeast-up displacement vector sequence acquisition subunit is used to obtain the position sequence of the vehicle, and convert the position sequence according to the preset attitude rotation matrix and the position of the reference point of the preset vehicle coordinate system relative to the IMU center to obtain the IMU center position vector sequence; wherein, the formula for converting the position sequence to obtain the IMU center position vector sequence is specifically:

[0059]

[0060] where p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, and l is the position of the reference point of the vehicle coordinate system relative to the IMU center;

[0061] Performing a solution in the time dimension on the IMU center position vector sequence to obtain a solution result, and performing coordinate system conversion on the solution result to obtain the IMU center northeast-up displacement vector sequence.

[0062] It can be understood that, compared with the prior art, the device in this application performs a solution on the actual measurement data of the vehicle's movement to obtain the vehicle real-time monitoring data, and then trains the preset displacement prediction model with the vehicle real-time monitoring data, so as to obtain the predicted displacement data of the inertial navigation center with the optimal displacement prediction model, and then constructs a measurement equation based on the predicted displacement data of the inertial navigation center, realizing the dynamic constraint on the vehicle GNSS / INS combination. The device in this application can identify and capture the motion patterns shown by the vehicle real-time monitoring data through the optimal displacement prediction model, and then can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art; the predicted displacement data of the inertial navigation center can identify more suitable constraint types for the vehicle GNSS / INS combination, and then better limit the error accumulation and divergence of the vehicle GNSS / INS combination through the measurement equation, improving the anti-interference ability of the vehicle GNSS / INS combination and the re-initialization speed after interference, thereby improving the accuracy of the vehicle GNSS / INS combination. Brief Description of the Drawings

[0063] Figure 1 : It is a flowchart of steps of a vehicle-mounted GNSS / INS combined dynamic constraint method provided by an embodiment of the present invention;

[0064] Figure 2 : It is a schematic structural diagram of a preset displacement prediction model provided by an embodiment of the present invention;

[0065] Figure 3 : It is a training flowchart of a preset displacement prediction model provided by an embodiment of the present invention;

[0066] Figure 4 : It is a flowchart of steps of a vehicle-mounted GNSS / INS combined dynamic constraint method based on LSTM provided by an embodiment of the present invention;

[0067] Figure 5 : It is a schematic structural diagram of a dynamic constraint device for a vehicle-mounted GNSS / INS combination provided by an embodiment of the present invention;

[0068] Among them, 201: Vehicle-mounted real-time monitoring data acquisition module; 202: Inertial navigation center predicted displacement data acquisition module; 203: Measurement equation calculation module; 204: Dynamic constraint module. Detailed Embodiment

[0069] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0070] It should be noted that the relevant technical terms involved in the subsequent embodiments of the present invention are explained herein:

[0071] (1) Vehicle-mounted GNSS / INS combination: A combined navigation system of the GNSS system and the INS system. The GNSS system (Global Navigation Satellite System) is a system that uses artificial earth satellites for high-precision radio navigation and positioning; the INS system (Inertial Navigation System, or inertial system) is an autonomous navigation system that does not rely on external information and does not radiate energy to the outside.

[0072] (2) Earth-Centered, Earth-Fixed System: Also known as the Earth-Centered, Earth-Fixed coordinate system (abbreviated as ECEF), and also called the Earth-Centered, Earth-Fixed rectangular coordinate system, it is a geodetic coordinate system with the Earth's center (the center of mass of the Earth) as the origin.

[0073] (3) Integrated Equipment Body Coordinate System: A coordinate system constructed based on the integrated equipment. The origin of this coordinate system is usually selected at a fixed point of the integrated equipment (such as the center of gravity of the integrated equipment or the center of a key component), and the coordinate axes are defined according to the structure and functional requirements of the integrated equipment.

[0074] (4) Local Coordinate System: A coordinate system constructed with the center position of the IMU in the vehicle-mounted GNSS / INS combination as the origin.

[0075] (5) East-North-Up Displacement: Displacement described in the East-North-Up (ENU) coordinate system. The East-North-Up coordinate system, also known as the East-North-Up coordinate system or the local rectangular coordinate system, is a reference coordinate system widely used in geographic navigation and positioning.

[0076] (6) Laser / VIO Integrated Equipment (Visual-Inertial Odometry): Also known as Visual-Inertial Odometry, it is a navigation and positioning device that integrates laser sensor and visual inertial odometry technologies.

[0077] Embodiment 1

[0078] Please refer to Figure 1 , which is a flowchart of the steps of a vehicle-mounted GNSS / INS combination dynamic constraint method provided by an embodiment of the present invention, including steps S101 to S104.

[0079] Step S101: Collect the actual measured motion data of the vehicle, and perform calculations on the actual measured motion data to obtain vehicle-mounted real-time monitoring data.

[0080] In this embodiment, the collection of the actual measured motion data of the vehicle and the calculation of the actual measured motion data to obtain vehicle-mounted real-time monitoring data specifically include:

[0081] Collect the actual measured motion data of the vehicle, and the actual measured motion data includes: IMU data and GNSS data;

[0082] Perform state recursion on the IMU data to obtain the INS velocity variance at the current moment;

[0083] Calculate the phase residual at the current moment and the pseudorange residual at the current moment according to the GNSS data;

[0084] Determine the vehicle real-time monitoring data based on the IMU data, GNSS data, INS speed variance at the current moment, phase residual at the current moment, and pseudorange residual at the current moment.

[0085] In an optional embodiment, read the IMU data and GNSS data from the vehicle-mounted GNSS / INS combination; the IMU data is 6-dimensional IMU data, including 3-dimensional gyroscope data and 3-dimensional accelerometer data; input the 6-dimensional IMU data into the inertial navigation system for state recursion to obtain the INS speed variance at the current moment; then calculate the GNSS data according to the GNSS system to obtain the phase residual at the current moment and the pseudorange residual at the current moment.

[0086] In this embodiment, by resolving the measured motion data of the vehicle, the vehicle real-time monitoring data is obtained, so that the real-time motion data of the vehicle can be accurately characterized by the vehicle real-time monitoring data, improving the prediction accuracy of the subsequent predicted displacement data of the inertial navigation center, and further improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0087] Step S102: Train a preset displacement prediction model according to the vehicle real-time monitoring data, obtain an optimal displacement prediction model, and obtain the predicted displacement data of the inertial navigation center based on the optimal displacement prediction model.

[0088] In this embodiment, the training of the preset displacement prediction model according to the vehicle real-time monitoring data, obtaining an optimal displacement prediction model, and obtaining the predicted displacement data of the inertial navigation center based on the optimal displacement prediction model specifically include:

[0089] Obtain the position sequence of the vehicle, and convert the position sequence according to the preset position sequence conversion algorithm to obtain the IMU center north-east-down displacement vector sequence;

[0090] Use the IMU center north-east-down displacement vector sequence as the reference output of the preset displacement prediction model, and use the vehicle real-time monitoring data as the input parameters of the preset displacement prediction model to train the preset displacement prediction model to obtain an optimal displacement prediction model;

[0091] Input the vehicle real-time monitoring data into the optimal displacement prediction model to obtain the predicted displacement data of the inertial navigation center.

[0092] In this embodiment, by training the preset displacement prediction model, the optimal displacement prediction model can identify and capture the motion pattern shown by the vehicle real-time monitoring data, and thus more accurate predicted displacement data of the inertial navigation center can be obtained, avoiding the inaccuracy of directly using IMU data for constraint in the prior art, and improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0093] In this embodiment, obtaining the position sequence of the vehicle and converting the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center north-east-down displacement vector sequence specifically includes:

[0094] Obtaining the position sequence of the vehicle and converting the position sequence according to a preset attitude rotation matrix and the position of the reference point of the preset vehicle coordinate system relative to the IMU center to obtain the IMU center position vector sequence; wherein, the formula for converting the position sequence to obtain the IMU center position vector sequence is specifically:

[0095]

[0096] wherein, p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, and l is the position of the reference point of the vehicle coordinate system relative to the IMU center;

[0097] Performing a solution in the time dimension on the IMU center position vector sequence to obtain a solution result, and performing a coordinate system conversion on the solution result to obtain the IMU center north-east-down displacement vector sequence.

[0098] In an alternative embodiment, That is, the IMU center position vector sequence, which includes the IMU center positions at different times; subtracting the IMU center position at the previous time from the IMU center position at the current time to obtain a difference (i.e., the solution result), and converting this difference to the local coordinate system to obtain the IMU center north-east-down displacement vector sequence.

[0099] It should be noted that in the vehicle-mounted GNSS / INS combination, usually a relatively large number of data acquisition devices are carried. The vehicle-mounted GNSS / INS combination described in the embodiments of the present invention carries a laser / VIO integrated device.

[0100] In an alternative embodiment, first calibrate the relative spatial relationship between the laser / VIO integrated device and the vehicle (i.e., the position of the reference point of the vehicle coordinate system relative to the IMU center), then obtain the position sequence of the vehicle output by the laser / VIO integrated device, and then convert this vehicle position sequence to the IMU center position sequence according to this relative spatial relationship. The specific conversion process is shown in formula (1) above.

[0101] It should be noted that in formula (1), R groundtruth is the attitude rotation matrix, specifically the attitude rotation matrix from the earth-centered earth-fixed system to the integrated device vehicle coordinate system of the position sequence output by the laser / VIO integrated device, p groundtruthIt is the position sequence output by the laser / VIO integrated device, and l is the position of the reference point of the vehicle system relative to the center of the IMU.

[0102] In this embodiment, the IMU center northeast-down displacement vector sequence is obtained by converting the position sequence, thereby realizing the training of the preset displacement prediction model, obtaining the optimal displacement prediction model, and further being able to obtain more accurate inertial center predicted displacement data, avoiding the inaccuracy of directly using IMU data for constraint in the prior art, and improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0103] In an alternative embodiment, please refer to Figure 2 , which is a schematic structural diagram of a preset displacement prediction model provided by an embodiment of the present invention. The preset displacement prediction model is an LSTM model (Long Short Term Memory network), and the preset displacement prediction model includes three parts. The first part is a bilinear layer for bilinearly transforming the input data to obtain a new tensor and then passing it to the next part. The second part is a forward LSTM layer and a backward LSTM layer. Both the forward LSTM layer and the backward LSTM layer contain three layers of LSTM. The third part is a linear layer for receiving the output tensors from the forward LSTM layer and the backward LSTM layer, thereby outputting the prediction result.

[0104] In an alternative embodiment, please refer to Figure 3 , which is a training flowchart of a preset displacement prediction model provided by an embodiment of the present invention. As shown in Figure 3 , a target vehicle equipped with an IMU and GNSS antennas travels in multiple scenarios, collects IMU raw data and GNSS data, then performs INS solution to obtain the INS speed variance, performs RTK solution on the GNSS data to obtain the vehicle position, phase residual, and pseudorange residual, and then obtains the reference speed sequence output by the laser / VIO integrated device (obtained by solving the IMU center northeast-down displacement vector sequence). During the training process, the Adam optimizer (Adaptive Moment Estimation) is used, the number of neurons in a single layer of LSTM is set to 120, and the loss function uses MSE (Mean Squared Error), and finally the optimal displacement prediction model is output.

[0105] Step S103: Obtain the constraint type according to the inertial center predicted displacement data, and construct a measurement equation according to the constraint type.

[0106] In this embodiment, the obtaining the constraint type according to the inertial center predicted displacement data and constructing a measurement equation according to the constraint type specifically includes:

[0107] The predicted displacement data of the inertial navigation center includes: the predicted north-east-down displacement of the inertial navigation center;

[0108] Judge the predicted north-east-down displacement of the inertial navigation center;

[0109] When the judgment result conforms to the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination, construct an observation residual corresponding to the straight-line constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the straight-line constraint;

[0110] When the judgment result conforms to the second preset result, perform a zero-velocity constraint on the vehicle-mounted GNSS / INS combination; and construct an observation residual corresponding to the zero-velocity constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the zero-velocity constraint.

[0111] In this embodiment, the predicted displacement data of the inertial navigation center can identify a more suitable constraint type for the vehicle-mounted GNSS / INS combination, and then better limit the error accumulation and divergence of the vehicle-mounted GNSS / INS combination through the measurement equation, improving the anti-interference ability and the re-initialization speed after interference of the vehicle-mounted GNSS / INS combination, thereby improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0112] It should be noted that obtaining the constraint type according to the predicted displacement data of the inertial navigation center in this embodiment specifically refers to judging the predicted displacement data of the inertial navigation center (i.e., the predicted north-east-down displacement of the inertial navigation center) to obtain the constraint type required by the vehicle-mounted GNSS / INS combination, where the constraint types required by the vehicle-mounted GNSS / INS combination include: straight-line constraint and zero-velocity constraint.

[0113] In this embodiment, when the judgment result conforms to the first preset result, performing a straight-line constraint on the vehicle-mounted GNSS / INS combination, constructing an observation residual corresponding to the straight-line constraint according to the predicted displacement data of the inertial navigation center, and then determining the measurement equation corresponding to the straight-line constraint specifically includes:

[0114] When the judgment result conforms to the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination, and obtain the straight-line constraint observation variance according to the straight-line constraint;

[0115] Obtain the position of the inertial navigation center in the Earth-centered Earth-fixed coordinate system at adjacent moments, the position error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and the transformation matrix from the Earth-centered Earth-fixed coordinate system of the inertial navigation center to the local coordinate system, and then combine the predicted north-east-down displacement of the inertial navigation center and the straight-line constraint observation variance to construct an observation residual corresponding to the straight-line constraint, thereby determining the measurement equation corresponding to the straight-line constraint, where the observation residual corresponding to the straight-line constraint is specifically:

[0116]

[0117] where l LC is the observation residual corresponding to the straight-line constraint, de, dn, and du are the predicted north-east-down displacements of the inertial navigation center, and are the positions of the inertial navigation center in the Earth-centered Earth-fixed (ECEF) coordinate system at time k and time k-1, respectively, is the transformation matrix from the ECEF coordinate system of the inertial navigation center to the local coordinate system, and are the position errors of the inertial navigation center in the ECEF coordinate system, R LC is the observation variance of the straight-line constraint.

[0118] It should be noted that is the transformation matrix from the ECEF coordinate system of the inertial navigation center to the local coordinate system, specifically the transformation matrix from the local coordinate system to the ECEF coordinate system calculated based on the position of the inertial navigation center at time k-1, and are the position errors of the inertial navigation center in the ECEF coordinate system, specifically the position errors of the inertial navigation center in the x, y, and z directions in the ECEF coordinate system.

[0119] In an alternative embodiment, there are already relatively mature means for obtaining the observation variance of the straight-line constraint according to the straight-line constraint, and it can also be defined according to the actual work requirements of technicians, which will not be elaborated here.

[0120] It should be noted that after constructing the observation residual corresponding to the straight-line constraint, formula (2) is used as the measurement equation corresponding to the straight-line constraint.

[0121] In an alternative embodiment, the first preset result is set such that the predicted displacement data of the inertial navigation center is a straight line. Specifically, the sliding window method can be used for judgment. The predicted displacement data of the inertial navigation center is put into the window for summation and standard deviation calculation. If the standard deviations of the predicted displacements of the inertial navigation center in the three directions in the north-east-down coordinate system are all less than 0.01 m, and the total displacement within the window is greater than 0.5, it is determined that straight-line motion has occurred during this period. In addition, when the number of satellites is less than the preset threshold, it can also be regarded as straight-line motion. The preset threshold can be determined according to the satellite system number and the ratio of the total received satellites. In this embodiment, the preset threshold is defined as 0.8.

[0122] In this embodiment, the observation residual corresponding to the straight-line constraint is constructed through the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the straight-line constraint is determined, which better restricts the error accumulation and divergence of the vehicle-mounted GNSS / INS combination, improves the anti-interference ability of the vehicle-mounted GNSS / INS combination and the speed of re-initialization after interference, and thus improves the accuracy of the vehicle-mounted GNSS / INS combination.

[0123] In this embodiment, when the judgment result meets the second preset result, zero velocity constraint is imposed on the vehicle-mounted GNSS / INS combination; and an observation residual corresponding to the zero velocity constraint is constructed based on the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the zero velocity constraint is determined, which specifically includes:

[0124] When the judgment result meets the second preset result, zero velocity constraint is imposed on the vehicle-mounted GNSS / INS combination;

[0125] An observation variance corresponding to the zero velocity constraint is constructed based on the predicted north-east-down displacement of the inertial navigation center;

[0126] The recursive carrier velocity of the inertial navigation center and the velocity error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system are obtained;

[0127] An observation residual corresponding to the zero velocity constraint is constructed based on the observation variance corresponding to the zero velocity constraint, the recursive carrier velocity of the inertial navigation center, and the velocity error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and then the measurement equation corresponding to the zero velocity constraint is determined; where the observation residual corresponding to the zero velocity constraint is specifically:

[0128]

[0129] where, l ZUPT is the observation residual corresponding to the zero velocity constraint, ν ins is the recursive carrier velocity of the inertial navigation center, and are the velocity errors of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and R ZUPT is the observation variance corresponding to the zero velocity constraint.

[0130] In an alternative embodiment, the construction formula for constructing the observation variance corresponding to the zero velocity constraint based on the predicted north-east-down displacement of the inertial navigation center is specifically:

[0131]

[0132] where, R ZUPT is the observation variance corresponding to the zero velocity constraint, de, dn, and du are the predicted north-east-down displacements of the inertial navigation center, dt is the time interval between adjacent moments, diag represents taking the diagonal matrix, is the transformation matrix from the local coordinate system to the Earth-centered Earth-fixed coordinate system.

[0133] It should be noted that after constructing the observation residual corresponding to the zero velocity constraint, formula (3) is used as the measurement equation corresponding to the zero velocity constraint.

[0134] In an alternative embodiment, the second preset result is set such that the displacements of the predicted displacement data of the inertial navigation center in three directions are all less than 0.5. Specifically, it can be that the predicted displacements of the inertial navigation center in three directions in the northeast-up coordinate system are all less than 0.5.

[0135] In this embodiment, the observation residual corresponding to the zero-velocity constraint is constructed from the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the zero-velocity constraint is determined, which better restricts the error accumulation and divergence of the vehicle-mounted GNSS / INS combination, improves the anti-interference ability of the vehicle-mounted GNSS / INS combination and the ability to re-initialize the speed after interference, and thus improves the accuracy of the vehicle-mounted GNSS / INS combination.

[0136] Step S104: Update the vehicle-mounted GNSS / INS combination according to the measurement equation to complete the dynamic constraint on the vehicle-mounted GNSS / INS combination.

[0137] In an alternative embodiment, please refer to Figure 4 FIG. [FIGURE NUMBER], which is a flowchart of the steps of a method for dynamically constraining a vehicle-mounted GNSS / INS combination based on LSTM provided by an embodiment of the present invention. As Figure 4 shown, in each epoch, the speed variance, vehicle position, phase observation residuals (i.e., pseudorange residuals and phase residuals), and 6D IMU data obtained through INS update and GNSS calculation are input into a trained neural network model (i.e., an LSTM model) to obtain the predicted center displacement result of the IMU carrier. According to the predicted center displacement result of the IMU carrier, the speed of the IMU carrier is obtained, and it is judged whether the speed of the IMU carrier is less than a certain threshold. If so, corresponding dynamic constraints are performed.

[0138] In this embodiment, the measured motion data of the vehicle is solved to obtain vehicle-mounted real-time monitoring data, and then the preset displacement prediction model is trained with the vehicle-mounted real-time monitoring data, so as to obtain the predicted displacement data of the inertial navigation center with the optimal displacement prediction model. Furthermore, a measurement equation is constructed based on the predicted displacement data of the inertial navigation center, realizing the dynamic constraint on the vehicle-mounted GNSS / INS combination. In this embodiment, the optimal displacement prediction model can identify and capture the motion patterns exhibited by the vehicle-mounted real-time monitoring data, and thus can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art; the predicted displacement data of the inertial navigation center can identify more suitable constraint types for the vehicle-mounted GNSS / INS combination, and then better restricts the error accumulation and divergence of the vehicle-mounted GNSS / INS combination through the measurement equation, improves the anti-interference ability of the vehicle-mounted GNSS / INS combination and the ability to re-initialize the speed after interference, and thus improves the accuracy of the vehicle-mounted GNSS / INS combination.

[0139] Embodiment 2

[0140] Please refer to Figure 5 , which is a schematic structural diagram of a dynamic constraint device for an in-vehicle GNSS / INS combination provided by an embodiment of the present invention, including: an in-vehicle real-time monitoring data acquisition module 201, an inertial navigation center predicted displacement data acquisition module 202, a measurement equation calculation module 203, and a dynamic constraint module 204;

[0141] Among them, the in-vehicle real-time monitoring data acquisition module 201 is used to collect the actual motion data of the vehicle and perform calculations on the actual motion data to obtain in-vehicle real-time monitoring data.

[0142] In this embodiment, the in-vehicle real-time monitoring data acquisition module 201 includes: an in-vehicle real-time monitoring data acquisition unit;

[0143] The in-vehicle real-time monitoring data acquisition unit is used to collect the actual motion data of the vehicle, and the actual motion data includes: IMU data and GNSS data;

[0144] Perform state recursion on the IMU data to obtain the INS velocity variance at the current moment;

[0145] Calculate the phase residual at the current moment and the pseudorange residual at the current moment according to the GNSS data;

[0146] Determine the in-vehicle real-time monitoring data according to the IMU data, GNSS data, the INS velocity variance at the current moment, the phase residual at the current moment, and the pseudorange residual at the current moment.

[0147] The inertial navigation center predicted displacement data acquisition module 202 is used to train a preset displacement prediction model according to the in-vehicle real-time monitoring data, obtain an optimal displacement prediction model, and obtain inertial navigation center predicted displacement data based on the optimal displacement prediction model.

[0148] In this embodiment, the inertial navigation center predicted displacement data acquisition module 202 includes: an inertial navigation center predicted displacement data acquisition unit;

[0149] The inertial navigation center predicted displacement data acquisition unit is used to obtain the position sequence of the vehicle and perform conversion on the position sequence according to a preset position sequence conversion algorithm to obtain an IMU center north-east-up displacement vector sequence;

[0150] Use the IMU center north-east-up displacement vector sequence as the reference output of the preset displacement prediction model, and use the in-vehicle real-time monitoring data as the input parameters of the preset displacement prediction model to train the preset displacement prediction model to obtain an optimal displacement prediction model;

[0151] Input the vehicle-mounted real-time monitoring data into the optimal displacement prediction model to obtain the predicted displacement data of the inertial navigation center.

[0152] In this embodiment, the predicted displacement data acquisition unit of the inertial navigation center includes: an IMU center north-east-down displacement vector sequence acquisition subunit;

[0153] The IMU center north-east-down displacement vector sequence acquisition subunit is used to obtain the position sequence of the vehicle, and convert the position sequence according to the preset attitude rotation matrix and the position of the reference point of the preset vehicle coordinate system relative to the IMU center to obtain the IMU center position vector sequence; wherein, the formula for converting the position sequence to obtain the IMU center position vector sequence is specifically:

[0154]

[0155] wherein, p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, and l is the position of the reference point of the vehicle coordinate system relative to the IMU center;

[0156] Perform a solution in the time dimension on the IMU center position vector sequence to obtain a solution result, and perform a coordinate system conversion on the solution result to obtain the IMU center north-east-down displacement vector sequence.

[0157] The measurement equation calculation module 203 is used to obtain the constraint type according to the predicted displacement data of the inertial navigation center, and construct a measurement equation according to the constraint type.

[0158] In this embodiment, the measurement equation calculation module 203 includes: a measurement equation calculation unit;

[0159] In the measurement equation calculation unit, the predicted displacement data of the inertial navigation center includes: the predicted north-east-down displacement of the inertial navigation center;

[0160] The measurement equation calculation unit is used to judge the predicted north-east-down displacement of the inertial navigation center;

[0161] When the judgment result meets the first preset result, perform a straight-line constraint on the vehicle-mounted GNSS / INS combination, construct an observation residual corresponding to the straight-line constraint according to the predicted displacement data of the inertial navigation center, and further determine the measurement equation corresponding to the straight-line constraint;

[0162] When the judgment result meets the second preset result, perform a zero-velocity constraint on the vehicle-mounted GNSS / INS combination; and construct an observation residual corresponding to the zero-velocity constraint according to the predicted displacement data of the inertial navigation center, and further determine the measurement equation corresponding to the zero-velocity constraint.

[0163] In this embodiment, the measurement equation calculation unit includes: a straight-line constraint subunit;

[0164] When the judgment result meets the first preset result, the straight-line constraint subunit is used to perform straight-line constraint on the vehicle-mounted GNSS / INS combination, and obtain the straight-line constraint observation variance according to the straight-line constraint;

[0165] Obtain the position of the inertial navigation center in the Earth-centered Earth-fixed coordinate system at adjacent moments, the position error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and the transformation matrix from the Earth-centered Earth-fixed coordinate system of the inertial navigation center to the local coordinate system. Then, combine the predicted north-east-down displacement of the inertial navigation center and the straight-line constraint observation variance to construct the observation residual corresponding to the straight-line constraint, so as to determine the measurement equation corresponding to the straight-line constraint. Among them, the observation residual corresponding to the straight-line constraint is specifically:

[0166]

[0167] where l LC is the observation residual corresponding to the straight-line constraint, de, dn, and du are the predicted north-east-down displacements of the inertial navigation center, and are the positions of the inertial navigation center in the Earth-centered Earth-fixed coordinate system at the k-th moment and the (k - 1)-th moment respectively, is the transformation matrix from the Earth-centered Earth-fixed coordinate system of the inertial navigation center to the local coordinate system, and are the position errors of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and R LC is the straight-line constraint observation variance.

[0168] In this embodiment, the measurement equation calculation unit includes: a zero-velocity constraint subunit;

[0169] When the judgment result meets the second preset result, the zero-velocity constraint subunit is used to perform zero-velocity constraint on the vehicle-mounted GNSS / INS combination;

[0170] Construct the observation variance corresponding to the zero-velocity constraint according to the predicted north-east-down displacement of the inertial navigation center;

[0171] Obtain the inertial navigation center recursive vehicle velocity and the velocity error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system;

[0172] Construct the observation residual corresponding to the zero-velocity constraint according to the observation variance corresponding to the zero-velocity constraint, the inertial navigation center recursive vehicle velocity, and the velocity error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, and then determine the measurement equation corresponding to the zero-velocity constraint; among them, the observation residual corresponding to the zero-velocity constraint is specifically:

[0173]

[0174] where lZUPT is the observation residual corresponding to the zero-velocity constraint, v ins is the carrier velocity recursively calculated by the inertial navigation center, and is the velocity error of the inertial navigation center in the Earth-centered Earth-fixed coordinate system, R ZUPT is the observation variance corresponding to the zero-velocity constraint.

[0175] The dynamic constraint module 204 is configured to update the vehicle-mounted GNSS / INS combination according to the measurement equation to complete the dynamic constraint on the vehicle-mounted GNSS / INS combination.

[0176] In this embodiment, the measured motion data of the vehicle is solved to obtain vehicle-mounted real-time monitoring data, and then the preset displacement prediction model is trained with the vehicle-mounted real-time monitoring data, so as to obtain the predicted displacement data of the inertial navigation center with the optimal displacement prediction model. Furthermore, a measurement equation is constructed based on the predicted displacement data of the inertial navigation center, realizing the dynamic constraint on the vehicle-mounted GNSS / INS combination. Through the optimal displacement prediction model, this embodiment can identify and capture the motion patterns shown by the vehicle-mounted real-time monitoring data, and thus can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art; through the predicted displacement data of the inertial navigation center, the constraint type more suitable for the vehicle-mounted GNSS / INS combination can be identified, and then the error accumulation and divergence of the vehicle-mounted GNSS / INS combination are better restricted through the measurement equation, improving the anti-interference ability of the vehicle-mounted GNSS / INS combination and the ability to re-initialize the speed after interference, thereby improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0177] In summary, in the embodiment of the present invention, the measured motion data of the vehicle is solved to obtain vehicle-mounted real-time monitoring data, and then the preset displacement prediction model is trained with the vehicle-mounted real-time monitoring data, so as to obtain the predicted displacement data of the inertial navigation center with the optimal displacement prediction model. Furthermore, a measurement equation is constructed based on the predicted displacement data of the inertial navigation center, realizing the dynamic constraint on the vehicle-mounted GNSS / INS combination. Through the optimal displacement prediction model, this embodiment can identify and capture the motion patterns shown by the vehicle-mounted real-time monitoring data, and thus can obtain more accurate predicted displacement data of the inertial navigation center, avoiding the inaccuracy of directly using IMU data for constraint in the prior art; through the predicted displacement data of the inertial navigation center, the constraint type more suitable for the vehicle-mounted GNSS / INS combination can be identified, and then the error accumulation and divergence of the vehicle-mounted GNSS / INS combination are better restricted through the measurement equation, improving the anti-interference ability of the vehicle-mounted GNSS / INS combination and the ability to re-initialize the speed after interference, thereby improving the accuracy of the vehicle-mounted GNSS / INS combination.

[0178] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, it is pointed out that for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A dynamic constraint method for a vehicle-mounted GNSS / INS combination, applicable to a vehicle-mounted GNSS / INS combination, characterized in that: include: Collecting the measured motion data of the vehicle, and solving the measured motion data to obtain the on-board real-time monitoring data; Training a preset displacement prediction model according to the on-board real-time monitoring data to obtain an optimal displacement prediction model, and obtaining the predicted displacement data of the inertial navigation center based on the optimal displacement prediction model; Obtaining a constraint type according to the predicted displacement data of the inertial navigation center, and constructing a measurement equation according to the constraint type; The vehicle-mounted GNSS / INS combination is updated according to the measurement equation to complete the dynamic constraint on the vehicle-mounted GNSS / INS combination.

2. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 1, characterized in that: The method of collecting the measured motion data of the vehicle and solving the measured motion data to obtain the on-board real-time monitoring data specifically includes: Collecting the measured motion data of the vehicle, wherein the measured motion data includes: IMU data and GNSS data; Perform state recursion on the IMU data to obtain the INS velocity variance at the current moment; Calculate the phase residual and the pseudorange residual at the current moment according to the GNSS data; The vehicle-mounted real-time monitoring data is determined according to the IMU data, the GNSS data, the INS velocity variance at the current moment, the phase residual at the current moment, and the pseudo-range residual at the current moment.

3. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 1, characterized in that: The method of training a preset displacement prediction model according to the on-board real-time monitoring data to obtain an optimal displacement prediction model, and obtaining the predicted displacement data of the inertial navigation center based on the optimal displacement prediction model, specifically includes: Obtaining a position sequence of the vehicle, and converting the position sequence according to a preset position sequence conversion algorithm to obtain an IMU center northeast celestial displacement vector sequence; The IMU center northeast sky displacement vector sequence is used as a reference output of a preset displacement prediction model, and the vehicle-mounted real-time monitoring data is used as an input parameter of the preset displacement prediction model, and the preset displacement prediction model is trained to obtain an optimal displacement prediction model; The vehicle-mounted real-time monitoring data is input into the optimal displacement prediction model to obtain the inertial navigation center predicted displacement data.

4. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 3, characterized in that: The acquiring of the vehicle position sequence and converting the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center northeast sky displacement vector sequence specifically includes: The position sequence of the vehicle is obtained, and the position sequence is transformed according to the preset attitude rotation matrix and the position of the reference point of the preset carrier system relative to the IMU center to obtain the IMU center position vector sequence; wherein the formula for transforming the position sequence to obtain the IMU center position vector sequence is specifically: Among them, p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, l is the position of the reference point of the carrier system relative to the center of the IMU; The IMU center position vector sequence is solved in the time dimension to obtain a solution result, and the solution result is converted into a coordinate system to obtain the IMU center northeast celestial displacement vector sequence.

5. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 1, characterized in that: The step of obtaining the constraint type according to the predicted displacement data of the inertial navigation center and constructing the measurement equation according to the constraint type specifically includes: The predicted displacement data of the inertial navigation center include: the predicted northeast celestial displacement of the inertial navigation center; Determining the predicted northeast celestial displacement of the inertial navigation center; When the judgment result meets the first preset result, a straight line constraint is performed on the vehicle-mounted GNSS / INS combination, and an observation residual corresponding to the straight line constraint is constructed according to the predicted displacement data of the inertial navigation center, and then a measurement equation corresponding to the straight line constraint is determined; When the judgment result meets the second preset result, the vehicle-mounted GNSS / INS combination is subjected to a zero-speed constraint; and the observation residual corresponding to the zero-speed constraint is constructed according to the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the zero-speed constraint is determined.

6. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 5, characterized in that: When the judgment result meets the first preset result, the vehicle-mounted GNSS / INS combination is subjected to a straight line constraint, and the observation residual corresponding to the straight line constraint is constructed according to the predicted displacement data of the inertial navigation center, and then the measurement equation corresponding to the straight line constraint is determined, which specifically includes: When the judgment result meets the first preset result, a straight line constraint is performed on the vehicle-mounted GNSS / INS combination, and a straight line constraint observation variance is obtained according to the straight line constraint; The position of the inertial navigation center in the earth-centered earth-fixed system at adjacent moments, the position error of the inertial navigation center in the earth-centered earth-fixed system, and the conversion matrix of the earth-centered earth-fixed system to the local system of the inertial navigation center are obtained, and then the observation residual corresponding to the straight line constraint is constructed in combination with the predicted northeast sky displacement of the inertial navigation center and the straight line constraint observation variance, so as to determine the measurement equation corresponding to the straight line constraint, wherein the observation residual corresponding to the straight line constraint is specifically: Among them, l LC is the observation residual corresponding to the straight line constraint, de, dn and dn are the predicted northeast celestial displacements of the inertial navigation center, and are the positions of the inertial navigation center at the earth-fixed system at the time k and k-1 respectively, is the conversion matrix from the Earth-centered Earth-fixed system to the local system at the inertial navigation center, and is the position error of the inertial navigation center in the Earth-fixed system, R LC is the linearly constrained observation variance.

7. A dynamic constraint method for a vehicle-mounted GNSS / INS combination as claimed in claim 5, characterized in that: When the judgment result meets the second preset result, the vehicle-mounted GNSS / INS combination is subjected to zero speed constraint; And construct the observation residual corresponding to the zero speed constraint according to the predicted displacement data of the inertial navigation center, and then determine the measurement equation corresponding to the zero speed constraint, which specifically includes: When the judgment result meets the second preset result, the vehicle-mounted GNSS / INS combination is subjected to zero speed constraint; constructing the observation variance corresponding to the zero-speed constraint according to the predicted northeast celestial displacement of the inertial navigation center; Obtain the recursive carrier velocity of the inertial navigation center and the velocity error of the inertial navigation center in the earth-centered fixed system; The observation residual corresponding to the zero speed constraint is constructed according to the observation variance corresponding to the zero speed constraint, the recursive carrier velocity of the inertial navigation center, and the velocity error of the inertial navigation center in the earth-fixed system at the center of the earth, and then the measurement equation corresponding to the zero speed constraint is determined; wherein the observation residual corresponding to the zero speed constraint is specifically: Among them, l ZUPT is the observation residual corresponding to the zero-velocity constraint, ν ins is the recursive carrier velocity of the inertial navigation center, and is the velocity error of the inertial navigation center in the Earth-fixed system, R ZUPT is the observation variance corresponding to the zero velocity constraint.

8. A vehicle-mounted GNSS / INS combined dynamic restraint device, characterized in that: include: On-board real-time monitoring data acquisition module, inertial navigation center predicted displacement data acquisition module, measurement equation calculation module and dynamic constraint module; The vehicle-mounted real-time monitoring data acquisition module is used to collect the vehicle's actual motion data and calculate the actual motion data to obtain the vehicle-mounted real-time monitoring data; The inertial navigation center predicted displacement data acquisition module is used to train a preset displacement prediction model according to the vehicle-mounted real-time monitoring data, obtain an optimal displacement prediction model, and obtain the inertial navigation center predicted displacement data based on the optimal displacement prediction model; The measurement equation calculation module is used to obtain the constraint type according to the predicted displacement data of the inertial navigation center, and construct the measurement equation according to the constraint type; The dynamic constraint module is used to update the vehicle-mounted GNSS / INS combination according to the measurement equation to complete the dynamic constraint of the vehicle-mounted GNSS / INS combination.

9. A vehicle-mounted GNSS / INS combined dynamic restraint device as claimed in claim 8, characterized in that: The inertial navigation center predicted displacement data acquisition module includes: an inertial navigation center predicted displacement data acquisition unit; The inertial navigation center predicted displacement data acquisition unit is used to acquire the position sequence of the vehicle, and convert the position sequence according to a preset position sequence conversion algorithm to obtain the IMU center northeast celestial displacement vector sequence; The IMU center northeast sky displacement vector sequence is used as a reference output of a preset displacement prediction model, and the vehicle-mounted real-time monitoring data is used as an input parameter of the preset displacement prediction model, and the preset displacement prediction model is trained to obtain an optimal displacement prediction model; The vehicle-mounted real-time monitoring data is input into the optimal displacement prediction model to obtain the inertial navigation center predicted displacement data.

10. A vehicle-mounted GNSS / INS combined dynamic restraint device as claimed in claim 9, characterized in that: The inertial navigation center predicted displacement data acquisition unit includes: an IMU center northeast sky displacement vector sequence acquisition subunit; The IMU center northeast celestial displacement vector sequence acquisition subunit is used to acquire the vehicle position sequence, and transform the position sequence according to the preset attitude rotation matrix and the position of the reference point of the preset carrier system relative to the IMU center to obtain the IMU center position vector sequence; wherein the formula for transforming the position sequence to obtain the IMU center position vector sequence is specifically: Among them, p ins is the inertial navigation center position, R groundtruth is the attitude rotation matrix, p groundtruth is the position sequence, l is the position of the reference point of the carrier system relative to the center of the IMU; The IMU center position vector sequence is solved in the time dimension to obtain a solution result, and the solution result is converted into a coordinate system to obtain the IMU center northeast celestial displacement vector sequence.

Images