Integrated navigation information fusion method and system considering driving state deviation

Abstract

The invention provides an integrated navigation information fusion method and system considering driving state deviation, and belongs to the technical field of positioning navigation, Kalman gains generated by a first filter in past n periods are obtained, a position difference and a speed difference between an INS and a GPS in a current GPS signal interruption period are predicted by using a pre-trained LSTM, and the position difference and the speed difference between the INS and the GPS in the current GPS signal interruption period are calculated. The vehicle driving state deviation is obtained; using a second filter to update a filter gain with the predicted vehicle driving state deviation as an input; and predicting the difference between the output of the first filter and the output of the second filter by using a pre-trained TL-LSTM, and correcting the INS navigation result by taking the additionally corrected optimal estimated value as final compensation. The GPS / INS integrated navigation information fusion considering the driving state deviation combines a long short-term memory network (LSTM) and error state Kalman filtering (ESKF), and secondary correction is performed on an INS navigation result to ensure the positioning precision of the integrated navigation system in a lock losing state.

Description

Technical Field

[0001] The present invention relates to the technical field of positioning and navigation, and in particular to a combined navigation information fusion method and system taking into account driving state deviation. Background Art

[0002] Accurate positioning is the key to achieving safe and efficient vehicle driving. It enables the vehicle to determine its absolute position in the surrounding environment and operate according to the predetermined driving state. As an independent unit, the vehicle-mounted inertial navigation system (INS) can calculate navigation information such as position and speed through the acceleration and angular velocity data provided by the inertial measurement unit without relying on external information. However, the error of INS accumulates over time. It can maintain high accuracy in the short term but is difficult to maintain in the long term. In order to enable the vehicle to obtain accurate latitude and longitude data for a long time, the global positioning system (GPS) is usually integrated with INS to suppress INS divergence and provide better navigation accuracy. However, in environments where GPS signals are difficult to receive, such as tunnels and urban canyons, the vehicle cannot obtain GPS information, and the integrated system is forced to operate in pure INS mode. In this case, it is difficult for the integrated system to achieve effective information fusion, resulting in reduced positioning accuracy, and ultimately causing the vehicle's estimated driving state to deviate from the actual driving state.

[0003] At present, there have been some studies on the GPS / INS integrated navigation information fusion method during GPS interruption at home and abroad, mainly focusing on hardware redundancy and software prediction. Hardware redundancy is mostly achieved by adding additional sensors to provide missing navigation information such as position and speed to correct the state output of INS. Although this type of method provides navigation information redundancy and positioning fault tolerance for the integrated navigation system, it also brings challenges such as increased system cost, increased complexity of sensor connection and alignment, and susceptibility to environmental constraints. Software prediction focuses on establishing the relationship between INS data and GPS navigation information through neural network methods, taking advantage of the fact that INS is not affected by external conditions, predicting the pseudo-GPS position based on INS data during GPS interruption, calculating the difference between it and the INS navigation result, and correcting the INS state through a filter. However, there are several problems in this research. First, the integrated navigation system is highly nonlinear during vehicle motion, and the commonly used Kalman filter is only applicable to linear systems. In the complex dynamic environment of vehicle driving, there are problems of insufficient filtering accuracy and error accumulation. Second, most of these studies only use the INS and GPS information at the current or previous moment, and do not fully consider the impact of vehicle historical information on positioning accuracy. Third, there is an inevitable difference between the target output of the neural network during training and the actual output during prediction, which leads to a deviation between the predicted and actual trajectory of the vehicle during GPS interruption. Fourth, although increasing the complexity of the neural network can improve the prediction accuracy, it will significantly reduce the training efficiency of the algorithm. Summary of the invention

[0004] The object of the present invention is to provide a combined navigation information fusion method and system taking into account driving state deviation, so as to solve at least one technical problem existing in the above-mentioned background technology.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, the present invention provides a combined navigation information fusion method considering driving state deviation, comprising:

[0007] Get the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation;

[0008] updating the filter gain by using the second filter to take the predicted vehicle driving state deviation as input;

[0009] The pre-trained TL-LSTM is used to predict the difference between the outputs of the first filter and the second filter, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

[0010] Furthermore, the Kalman gain of the first filter in the past n cycles is used as the input of the LSTM to predict the system observation value of the current cycle, that is, the position difference and speed difference between INS and GPS, and the LSTM is trained.

[0011] Furthermore, after the LSTM training is completed, the predicted system observation value is input into the second filter to obtain a new optimal estimate. The data set for training the LSTM is used as the source data set, and the data set for training the TL-LSTM is used as the target data set for transfer learning. The training input of the TL-LSTM is the Kalman gain of the current cycle, and the output is the difference between the optimal estimates of the two filters.

[0012] Furthermore, the vehicle's speed and position information is obtained through the specific force and angular velocity data provided by the vehicle-mounted inertial measurement unit, including: through the INS strapdown solution technology, using the dynamic equations of inertial mechanization, performing conversion and calculation between different coordinate systems, and obtaining the vehicle's position and speed data at each moment of the INS.

[0013] Furthermore, the LSTM model is trained, including: obtaining the INS navigation result at each moment through INS strapdown solution, and then calculating the position difference and speed difference between INS and GPS, that is, the vehicle driving state deviation; inputting the vehicle driving state deviation as the observation vector into the first error state Kalman filter ESKF-1, fusing the GPS and INS navigation information, and obtaining the Kalman gain at each moment and the optimal estimate of ESKF-1; then taking the Kalman gain of the previous moment and the previous m moments as the model input, and taking the position difference and speed difference between INS and GPS at the current moment as the model output.

[0014] Furthermore, the training of the second LSTM model includes: predicting the driving state deviation at the current moment through the trained first LSTM model, inputting it as an observation vector into the second error state Kalman filter ESKF-2, obtaining the Kalman gain at the current moment and the optimal estimate of ESKF-2, using the Kalman gain at the current moment as the model input, and using the difference between the optimal estimate values ​​generated by ESKF-1 and ESKF-2 as the model output.

[0015] In a second aspect, the present invention provides a combined navigation information fusion system considering driving state deviation, comprising:

[0016] An acquisition module is used to obtain the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between the INS and the GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation;

[0017] An updating module, for updating a filter gain by using the second filter to take the predicted vehicle driving state deviation as input;

[0018] The correction module is used to use the pre-trained TL-LSTM to predict the difference between the outputs of the first filter and the second filter, and use the additional corrected optimal estimate as the final compensation to correct the INS navigation result.

[0019] In a third aspect, the present invention provides a non-transitory computer-readable storage medium, which is used to store computer instructions. When the computer instructions are executed by a processor, the combined navigation information fusion method considering driving state deviation as described in the first aspect is implemented.

[0020] In a fourth aspect, the present invention provides a computer device comprising a memory and a processor, wherein the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, and the processor calls the program instructions to execute the combined navigation information fusion method considering driving state deviation as described in the first aspect.

[0021] In a fifth aspect, the present invention provides an electronic device comprising: a processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the combined navigation information fusion method considering driving state deviations as described in the first aspect.

[0022] The beneficial effects of the present invention are as follows: the GPS / INS integrated navigation information is fused considering the driving state deviation, a long short-term memory network (LSTM) is combined with an error state Kalman filter (ESKF), and the INS navigation result is corrected twice to ensure the positioning accuracy of the integrated navigation system in the lost-lock state.

[0023] Additional advantages of the present invention will be more clearly given in the following description or learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0025] Figure 1 The present invention is a flowchart of the INS strapdown solution according to an embodiment of the present invention.

[0026] Figure 2 This is a flowchart of the working process when the GPS signal is available according to an embodiment of the present invention.

[0027] Figure 3 This is a flowchart of the process when the GPS signal is interrupted according to an embodiment of the present invention.

[0028] Figure 4 This is a graph showing how the performance of the LSTM model described in an embodiment of the present invention varies with input delay length.

[0029] Figure 5 This is a graph showing how the training time varies with the input delay length according to an embodiment of the present invention.

[0030] Figure 6 This is a comparison result diagram described in an embodiment of the present invention. DETAILED DESCRIPTION

[0031] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below by the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be interpreted as limiting the present invention.

[0032] It should be understood by those skilled in the art that unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art to which this invention belongs.

[0033] It should also be understood that terms, such as those defined in commonly used dictionaries, should be understood to have a meaning consistent with that in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless as defined herein.

[0034] Those skilled in the art will appreciate that, unless otherwise stated, the singular forms "a", "an", "said" and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present invention refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements and / or groups thereof.

[0035] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. Different embodiments or examples described in this specification and features of different embodiments or examples may be combined and combined by those skilled in the art without contradiction.

[0036] To facilitate understanding of the present invention, the present invention is further explained below with reference to specific embodiments in conjunction with the accompanying drawings, and the specific embodiments do not constitute a limitation on the embodiments of the present invention.

[0037] Those skilled in the art should understand that the drawings are merely schematic diagrams of embodiments, and the components in the drawings are not necessarily necessary for implementing the present invention.

[0038] In order to improve the information fusion effect of the GPS / INS integrated navigation system when the GPS signal is interrupted, correct the driving state deviation of the vehicle, improve the positioning accuracy of the integrated navigation and ensure the training efficiency of the method, the present invention constructs an error state Kalman filter (ESKF)-long short-term memory (LSTM) neural network structure that considers the historical dynamic information of the vehicle. Through the combination of two LSTMs and ESKF, the correction of the vehicle driving state deviation and the further optimization of the correction value are realized. Among them, ESKF is used to fuse the data of INS and GPS, and output the navigation correction value of the integrated system; LSTM is used to establish the relationship between the internal parameters of the filter and the navigation information. In addition, the second LSTM is obtained by transfer learning of the first LSTM, thereby improving the training efficiency and robustness of the method.

[0039] Example 1

[0040] In this embodiment 1, a combined navigation information fusion system considering driving state deviation is first provided, including: an acquisition module, used to obtain the Kalman gain generated by the first filter in the past n cycles, and use a pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation; an update module, used to use the second filter to update the filter gain by taking the predicted vehicle driving state deviation as input; a correction module, used to use a pre-trained TL-LSTM to predict the difference between the outputs of the first filter and the second filter, and use the additional corrected optimal estimate as the final compensation to correct the INS navigation result.

[0041] In this embodiment, the above system is used to implement a combined navigation information fusion method that takes into account driving state deviation. The GPS / INS combined navigation information fusion method that takes into account driving state deviation is divided into two situations: GPS signal available and GPS interrupted.

[0042] When GPS signals are available, the proposed method operates in training mode, which consists of two steps: initial optimization and additional corrections.

[0043] In the initial optimization step, the core idea of ​​this step is to use the Kalman gain of ESKF-1 in the past n cycles (n represents the input delay length m plus 1) as the input of LSTM to predict the system observation value of the current cycle (i.e., the position difference O between INS and GPS t (p) and speed difference O t (v)). After the LSTM training is completed, the predicted system observations are input into ESKF-2 to obtain a new optimal estimate. In the additional correction step, the work starts after the initial optimization step. The dataset used to train the LSTM in the initial optimization step is used as the source dataset, and the dataset used to train the TL-LSTM in the additional correction step is used as the target dataset for transfer learning. The training input of the TL-LSTM is the Kalman gain of the current cycle, and the output is the difference between the optimal estimates of the two filters. In this way, the errors generated by the filters can be effectively modeled and compensated. When GPS is interrupted, the proposed method operates in the prediction mode, and ESKF-1 no longer works. LSTM takes the Kalman gain generated by ESKF-1 in the past n cycles as input to predict the necessary system observations in the current GPS signal interruption cycle. Subsequently, the prediction results are input into ESKF-2 to update the filter gains. Finally, the difference between the outputs of the two filters ESKF-1 and ESKF-2 is predicted by TL-LSTM, and the optimal estimate after additional correction is used as the final compensation to correct the INS navigation results.

[0044] In this embodiment, the speed and position information of the vehicle is obtained through the specific force and angular velocity data provided by the vehicle-mounted inertial measurement unit. Specifically, through the INS strapdown solution technology, the dynamic equations of inertial mechanization are used to convert and calculate between different coordinate systems (such as inertial coordinate system, earth-centered earth-fixed coordinate system, navigation coordinate system and carrier coordinate system), so that INS can provide the position and speed data of the vehicle at each moment. This process provides a data basis for the subsequent fusion of GPS and INS data.

[0045] For the construction of LSTM-ESKF framework, the INS navigation result P at each moment is obtained through INS strapdown solution. INS (t) and V INS (t), and then further calculate the position difference between INS and GPSt (p) and speed difference O t (v), i.e., the vehicle driving state deviation. It is used as the observation vector and input into the first error state Kalman filter (ESKF-1) to fuse the GPS and INS navigation information to obtain the Kalman gain K at each moment. t (p) and K t (v) and the optimal estimate of ESKF-1δ(p final ) t and δ(v final ) t Then the Kalman gain (k t-m-1 (p),k t-m-1 (v); k t-m (p),k t-m (v); ...; k t-1 (p),k t-1 (v)) is used as the model input, and the current position difference between INS and GPS is taken as t (p) and speed difference O t (v) is used as the model output to train the LSTM model.

[0046] For the application of the dual LSTM-ESKF framework, the current driving state deviation is predicted by the trained LSTM model and input into the second error state Kalman filter (ESKF-2) as the observation vector to obtain the current Kalman gain K t (p) and K t (p) and the optimal estimate of ESKF-2δ(p final ) t ' and δ(v final ) t '. The Kalman gain at the current moment is used as the model input, and the difference between the optimal estimates generated by ESKF-1 and ESKF-2 is used as the model output. Based on the pre-trained LSTM model, a second LSTM model is further trained through transfer learning, called TL-LSTM.

[0047] Example 2

[0048] like Figures 1 to 6 As shown, in this embodiment, a GPS / INS integrated navigation information fusion method considering driving state deviation is provided. In complex situations, when GPS may lose lock, resulting in a significant decrease in navigation accuracy, a long short-term memory network (LSTM) is combined with an error state Kalman filter (ESKF) to perform a secondary correction on the INS navigation result to ensure the positioning accuracy of the integrated navigation system in the lost lock state, thereby solving the problems existing in the background technology.

[0049] The method specifically comprises the following steps:

[0050] 1. Based on the measurement data of the inertial measurement unit (IMU), the INS strapdown solution is performed to obtain the position and speed information of the INS at each moment.

[0051] 2. An LSTM-ESKF framework that considers historical vehicle dynamic information is established, which makes full use of the linearization characteristics of ESKF in dealing with nonlinear problems and the powerful ability of LSTM in time series data modeling to make initial corrections to INS navigation results.

[0052] 3. Based on step 2, two LSTM-ESKF frameworks are used to realize the correction of vehicle driving state deviation and further optimization of correction value. In addition, the second LSTM is obtained by transfer learning from the first LSTM, which improves the training efficiency and robustness of the method.

[0053] 4. Determine whether the GPS / INS integrated navigation information fusion method taking into account driving state deviation is in training mode or prediction mode based on whether the GPS signal is interrupted.

[0054] In step 1, the INS strapdown solution uses the dynamic equations of inertial mechanization to calculate the vehicle's attitude, speed and position, including three parts: common coordinate system, coordinate system conversion and inertial navigation system update algorithm.

[0055] (1) Common coordinate systems

[0056] Common coordinate systems include inertial coordinate system, Earth-centered Earth-fixed coordinate system, navigation coordinate system and carrier coordinate system, as follows:

[0057] ①Inertial coordinate system (i system)

[0058] The origin of the inertial coordinate system is the center of mass of the earth, the x-axis corresponds to the vernal equinox of the earth, the z-axis points to the North Pole, the y-axis is perpendicular to the xoz plane, and it satisfies the right-hand rule. The inertial coordinate system is stationary relative to the stars, and the observed quantities measured by the inertial measurement device are relative to the inertial coordinate system.

[0059] ②Earth-centered Earth-fixed coordinate system (e system)

[0060] The origin of the Earth-centered Earth-fixed coordinate system is the Earth's center of mass, the x-axis points to the zero meridian, the z-axis coincides with the Earth's rotation axis, the y-axis points to the 90 degrees east longitude, and the xoy plane coincides with the Earth's equatorial plane.

[0061] ③Navigation coordinate system (n system)

[0062] The navigation coordinate system is a geographic coordinate system. Typical geographic coordinate systems include the Northeast Heaven (ENU) coordinate system and the North East Heaven (NED) coordinate system. The Northeast Heaven coordinate system uses the center of mass of the carrier as the coordinate origin, the local east direction as the x-axis, the north direction as the y-axis, and the zenith direction as the z-axis.

[0063] ④Carrier coordinate system (b system)

[0064] The carrier coordinate system takes the center of mass of the carrier as its origin, the y-axis points forward along the longitudinal axis of the carrier, the x-axis points to the right along the transverse direction of the carrier, and the z-axis points upwards. The three axes together form the right-hand rule.

[0065] (2) Coordinate system conversion

[0066] ①Convert the inertial coordinate system to the Earth-centered Earth-fixed coordinate system

[0067] The rotation matrix of the inertial coordinate system converted to the Earth-centered Earth-fixed coordinate system can be expressed as:

[0068]

[0069] Where t is time, ω ie is the Earth's rotation speed.

[0070] ②Conversion from Earth-centered Earth-fixed coordinate system to navigation coordinate system

[0071] The rotation matrix of the Earth-centered Earth-fixed coordinate system to the navigation coordinate system can be expressed as:

[0072]

[0073] Among them, λ represents the latitude of the carrier origin, and L represents the longitude of the carrier origin.

[0074] ③Convert the carrier coordinate system to the navigation coordinate system

[0075] The relative relationship between the carrier coordinate system b and the navigation coordinate system n is the carrier's attitude. γ, θ, ψ are defined as the carrier's roll angle, pitch angle, and yaw angle respectively. The navigation coordinate system rotates around the z-axis by an angle of γ, then around the x-axis by an angle of θ, and finally around the y-axis by an angle of ψ to transform the body coordinate system. The rotation matrices of the two can be expressed as:

[0076]

[0077] (3) Inertial navigation system update algorithm

[0078] The data output by the IMU sensor is solved by the INS strapdown to obtain attitude information, speed information and position information. The specific solution method is as follows:

[0079] ①Calculation of attitude information

[0080] Using the quaternion method, the quaternion is defined as:

[0081] Q=q0+q1i+q2j+q3k (4) The quaternion describing the carrier’s attitude is The differential equation is expressed as follows:

[0082]

[0083] in is the projection of the angular velocity of the carrier in the b system converted to the n system, and its expression is:

[0084]

[0085] In the formula, is the angular velocity of the carrier, is the posture matrix, is the projection of the earth's rotation angular velocity in the navigation coordinate system, is the rotation angular velocity of the navigation coordinate system relative to the Earth-centered Earth-fixed coordinate system, satisfying:

[0086]

[0087] Among them, v n and v e Respectively represent the northward and eastward velocities of the carrier, R m Represents the meridian curvature radius, R n represents the radius of curvature of the circle, and h represents the elevation.

[0088] Solve the differential equation (6) and update Get the rotation matrix of the vector:

[0089]

[0090] ②Calculation of velocity information

[0091] Assume that the velocity calculated by the carrier in the navigation coordinate system is v n , then the derivative of velocity with respect to time in the navigation coordinate system and the derivative with respect to time in the inertial coordinate system have the following relationship:

[0092]

[0093] The local gravity vector can be expressed as: So we have:

[0094]

[0095] Among them, f bis the specific force information output by the accelerometer. The differential equation for velocity update can be obtained from equations (10) and (11):

[0096]

[0097] Among them, f n is the specific acceleration in the navigation coordinate system, is the Columbia correction term.

[0098] Discretizing equation (12) yields the velocity update equation:

[0099]

[0100] ③Calculation of position information

[0101] In the navigation coordinate system, the position update equation can be obtained by integrating the velocity:

[0102]

[0103] in, They represent the eastward speed, northward speed and celestial speed in the navigation coordinate system respectively.

[0104] In step 2, the LSTM-ESKF framework considering historical vehicle dynamic information first obtains the INS navigation result P at each moment through INS strapdown solution. INS (t) and V INS (t), and then further calculate the position difference between INS and GPS t (p) and speed difference O t (v), i.e., the vehicle driving state deviation. It is input as the observation vector into ESKF-1 to fuse the GPS and INS navigation information to obtain the Kalman gain K at each moment. t (p) and K t (v) and the optimal estimate of ESKF-1δ(p final ) t and δ(v final ) t . The Kalman gain (k t-m-1 (p),k t-m-1 (v); k t-m (p),k t-m (v); ...; k t-1 (p),k t-1 (v)) is used as the model input, and the current position difference between INS and GPS is taken as t (p) and speed difference O t (v) is used as the model output to train the LSTM model.

[0105] The Error State Kalman Filter (ESKF) is an extension of the classic Kalman Filter (KF). It considers INS as a motion model, GPS as an observation model, and defines the true state variable as the sum of the nominal state variable and the error state variable. During the filtering process, only the error state variable is predicted and updated. The principle is as follows:

[0106] (1) System state variables

[0107] ESKF divides the system state into nominal state and error state, which together constitute the true state. The nominal state does not consider system noise and various disturbances and is directly obtained by integrating IMU data, while the error state contains various noises and disturbances and is predicted and updated by the extended Kalman filter (EKF). Nominal state x, error state δx and true state is defined as follows:

[0108]

[0109] In formula (17), and Represent position, velocity, attitude quaternion, accelerometer bias and gyroscope bias respectively, and Represents the vehicle's position error, velocity error, attitude error, accelerometer bias error, and gyroscope bias error. represents generalized addition, Represents a quaternion product operation.

[0110] Since quaternions have parameter redundancy when representing attitude, and the covariance matrix is ​​prone to singularity, the minimized parameter is used to represent the attitude in the error state, that is, δθ is used to express the rotation error. The relationship between the rotation matrix δR, the axis angle vector δθ, and the quaternion δq in the error state is:

[0111]

[0112] In formula (18), Exp(·) represents the exponential mapping function, [·] × Represents the antisymmetric matrix of a vector.

[0113] (2) System kinematic model

[0114] During the ESKF process, the real state kinematic model of the IMU is as follows:

[0115]

[0116] where a m and Represent the acceleration and angular velocity measurements, a n and Represent the acceleration and angular velocity measurement white noise, a ω and Represent the acceleration and angular velocity bias vectors respectively. The nominal state kinematic model is as follows:

[0117]

[0118] The kinematic model of the error state can be obtained from equations (19) and (20):

[0119]

[0120] Since equations (19)-(21) are all continuous time equations, they need to be discretized. After discretization, we can get the IMU at t k The nominal state kinematic model at the moment is:

[0121]

[0122] In formula (22), Δt is the sampling interval of IMU. k The error state kinematic model at the moment is:

[0123]

[0124] In formula (23), w vk represents the Gaussian random pulse noise at speed, w θk represents the Gaussian random pulse noise under the attitude, w ak represents Gaussian random pulse noise under acceleration bias, w ωk represents Gaussian random impulse noise under angular velocity bias. Their covariance is defined as follows:

[0125]

[0126] In formula (24), and represent the standard deviation of Gaussian white noise of acceleration and angular velocity, respectively, and Represent the standard deviation of the Gaussian white noise of acceleration and angular velocity bias, respectively.

[0127] (3) ESKF prediction process

[0128] In discrete terms, the nominal state, error state, IMU measurement, and noise are defined as follows:

[0129]

[0130] Substituting the vector defined above into equation (22), the recursive formula of the nominal state vector at this moment is:

[0131] x k+1 =f(x k ,u mk ) (26)

[0132] In formula (26), f(·) represents the recursive function of the nominal state variable. Combined with formula (23), linearization is performed according to Taylor's formula to obtain t k+1 The recursive formula of the moment error state vector is:

[0133]

[0134] In formula (27), f δ (·) represents the recursive function of the error state variable, F xk and F wk are the Jacobian matrices corresponding to the error state and noise state respectively. The error state and covariance update process are obtained:

[0135]

[0136] Where Q w Represents the noise matrix, and the specific form is as follows:

[0137]

[0138] (4) ESKF observation process

[0139] Accelerometers and gyroscopes have bias errors, and simply integrating them through the IMU will cause drift in the pose estimation, which will gradually increase over time. This paper uses the position and velocity information obtained from GPS to correct the error state estimation of the ESKF. When the GPS signal is available, its observation data will be promptly incorporated into the filter for update.

[0140] z k =h(x tk )+w mk (29)

[0141] Where z k is the measurement signal vector, w mk represents the random white noise of the measurement signal, and h(·) represents the nonlinear mapping function of the system state. The a posteriori calibration update equation of the error state is:

[0142]

[0143] According to the chain rule, the Jacobian matrix is ​​calculated as follows:

[0144]

[0145] in,

[0146]

[0147] (5) ESKF error state injection and zeroing

[0148] According to the recursion of the prior system state and the calibration of the error state, the updated state is obtained:

[0149]

[0150] Each system status corresponds to the following:

[0151]

[0152] After the error state is injected into the nominal state, the prior error state and the corresponding covariance need to be reset. The specific operation is as follows:

[0153]

[0154] Where G is the reset function The linearized Jacobian matrix is ​​approximately the unit matrix, so it is usually replaced by the unit matrix in practical applications.

[0155] In step 3, the current driving state deviation is predicted by the trained LSTM model and input into ESKF-2 as the observation vector to obtain the Kalman gain K at the current moment. t (p) and K t (p) and the optimal estimate of ESKF-2δ(p final ) t ' and δ(v final ) t '. The Kalman gain at the current moment is used as the model input, and the difference between the optimal estimates generated by ESKF-1 and ESKF-2 is used as the model output. Based on the pre-trained LSTM model, a second LSTM model is further trained through transfer learning, called TL-LSTM.

[0156] In step 4, the GPS / INS integrated navigation information fusion method considering the driving state deviation is divided into two cases: GPS signal is available and GPS is interrupted.

[0157] (1) When GPS signals are available, the proposed method operates in training mode, which consists of two steps: initial optimization and additional correction.

[0158] ① Initial optimization step

[0159] The core idea of ​​this step is to use the Kalman gain of ESKF-1 in the past n cycles (n represents the input delay length m plus 1) as the input of LSTM to predict the system observation value of the current cycle (i.e., the position difference O between INS and GPS) t (p) and speed difference O t (v)). After LSTM training is completed, the predicted system observations are input into ESKF-2 to obtain a new optimal estimate.

[0160] ② Additional correction steps

[0161] The additional correction step starts after the initial optimization step. The dataset used to train the LSTM in the initial optimization step is used as the source dataset, and the dataset used to train the TL-LSTM in the additional correction step is used as the target dataset for transfer learning. The training input of the TL-LSTM is the Kalman gain of the current cycle, and the output is the difference between the two filter optimal estimates. In this way, the error generated by the filter can be effectively modeled and compensated.

[0162] (2) When GPS is interrupted, the proposed method operates in prediction mode, and ESKF-1 no longer works.

[0163] LSTM takes the Kalman gains generated by ESKF-1 in the past n cycles as input and predicts the necessary system observations in the current GPS signal interruption cycle. The prediction results are then input into ESKF-2 to update the filter gain. Finally, the difference between the outputs of the two filters ESKF-1 and ESKF-2 is predicted by TL-LSTM, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

[0164] In this embodiment, the proposed navigation strategy is verified by road testing. The experimental equipment uses INS570D (a high-precision vehicle-mounted GPS / INS combined navigation and positioning system), including IMU570D with a sampling frequency of 100Hz and a satellite navigation board with an operating frequency of 10Hz. The specific specifications of the sensors used are shown in Table 1.

[0165] Table 1 Specifications of sensors used in the experiment

[0166]

[0167]

[0168] Firstly, the proposed method is trained using a 1800-second period of available GPS signal data, and then a 120-second GPS signal outage is simulated to verify the performance of the proposed scheme.

[0169] like Figure 4 As shown in Figure 1, when the input delay length m is 20, the RMSE position error reaches a minimum value of 5.85 meters, which is 35% higher than the positioning accuracy when there is no input delay. Figure 5 It is shown that when m exceeds 20, the improvement of model prediction accuracy is not significant, but the training time is significantly prolonged (increased by 30%). Therefore, in this embodiment, the optimal value of the input delay length is determined to be 20 to ensure a high positioning accuracy and control the training time within a reasonable range.

[0170] like Figure 6 As shown in the figure, within 120 seconds of GPS interruption, the positioning error of pure INS due to integral drift reached 23.80m, and the standard deviation of positioning error was 44.45m. The proposed method effectively limits the integral drift of INS by using two LSTM-ESKF frameworks, and the positioning error is 0.62m, which is reduced by 97.39%, and the standard deviation is 0.33m, which is reduced by 99.26%.

[0171] Example 3

[0172] This embodiment 3 provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the combined navigation information fusion method considering driving state deviation as described above is implemented. The method includes:

[0173] Get the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation;

[0174] updating the filter gain by using the second filter to take the predicted vehicle driving state deviation as input;

[0175] The pre-trained TL-LSTM is used to predict the difference between the outputs of the first filter and the second filter, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

[0176] Example 4

[0177] This embodiment 4 provides a computer device, including a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions executable by the processor, the processor calls the program instructions to execute the above-mentioned combined navigation information fusion method considering driving state deviation, the method comprising:

[0178] Get the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation;

[0179] updating the filter gain by using the second filter to take the predicted vehicle driving state deviation as input;

[0180] The pre-trained TL-LSTM is used to predict the difference between the outputs of the first filter and the second filter, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

[0181] Example 5

[0182] This embodiment 5 provides an electronic device, including: a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory, so that the electronic device executes instructions for implementing the above-mentioned combined navigation information fusion method considering driving state deviation, the method comprising:

[0183] Get the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation;

[0184] updating the filter gain by using the second filter to take the predicted vehicle driving state deviation as input;

[0185] The pre-trained TL-LSTM is used to predict the difference between the outputs of the first filter and the second filter, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

[0186] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0187] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0188] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0189] These computer program instructions can also be loaded onto a computer or other programmable data processing device, and a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0190] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative work on the basis of the technical solution disclosed in the present invention should be included in the scope of protection of the present invention.

Claims

1. A combined navigation information fusion method considering driving state deviation, characterized in that: include: Get the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between INS and GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation; updating the filter gain by using the second filter to take the predicted vehicle driving state deviation as input; The pre-trained TL-LSTM is used to predict the difference between the outputs of the first filter and the second filter, and the additional corrected optimal estimate is used as the final compensation to correct the INS navigation result.

2. The integrated navigation information fusion method considering driving state deviation according to claim 1 is characterized in that: The Kalman gain of the first filter in the past n cycles is used as the input of LSTM to predict the system observation value of the current cycle, that is, the position difference and speed difference between INS and GPS, and train LSTM.

3. The integrated navigation information fusion method considering driving state deviation according to claim 2 is characterized in that: After LSTM training is completed, the predicted system observation value is input into the second filter to obtain a new optimal estimate. The data set for training LSTM is used as the source data set, and the data set for training TL-LSTM is used as the target data set for transfer learning. The training input of TL-LSTM is the Kalman gain of the current cycle, and the output is the difference between the optimal estimates of the two filters.

4. The integrated navigation information fusion method considering driving state deviation according to claim 1 is characterized in that: The vehicle's speed and position information is obtained through the specific force and angular velocity data provided by the on-board inertial measurement unit, including: through INS strapdown solution technology, using the dynamic equations of inertial mechanization, converting and calculating between different coordinate systems, and obtaining the vehicle's position and speed data at each moment of the INS.

5. The integrated navigation information fusion method considering driving state deviation according to claim 3 is characterized in that: The LSTM model is trained, including: obtaining the INS navigation result at each moment through INS strapdown solution, and then calculating the position difference and speed difference between INS and GPS, that is, the vehicle driving state deviation; inputting the vehicle driving state deviation as the observation vector into the first error state Kalman filter ESKF-1, fusing the GPS and INS navigation information, and obtaining the Kalman gain at each moment and the optimal estimate of ESKF-1; then taking the Kalman gain of the previous moment and the previous m moments as the model input, and taking the position difference and speed difference between INS and GPS at the current moment as the model output.

6. The integrated navigation information fusion method considering driving state deviation according to claim 5 is characterized in that: The training of the second LSTM model includes: predicting the current driving state deviation through the trained first LSTM model, inputting it as the observation vector into the second error state Kalman filter ESKF-2, obtaining the current Kalman gain and the optimal estimate of ESKF-2, using the current Kalman gain as the model input, and taking the difference between the optimal estimate values ​​generated by ESKF-1 and ESKF-2 as the model output.

7. A combined navigation information fusion system considering driving state deviation, characterized in that: include: An acquisition module is used to obtain the Kalman gain generated by the first filter in the past n cycles, and use the pre-trained LSTM to predict the position difference and speed difference between the INS and the GPS in the current GPS signal interruption cycle, that is, the vehicle driving state deviation; An updating module, for updating a filter gain by using the second filter to take the predicted vehicle driving state deviation as input; The correction module is used to use the pre-trained TL-LSTM to predict the difference between the outputs of the first filter and the second filter, and use the additional corrected optimal estimate as the final compensation to correct the INS navigation result.

8. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by the processor, the combined navigation information fusion method considering driving state deviation as described in any one of claims 1-6 is implemented.

9. A computer device, characterized in that: It includes a memory and a processor, the processor and the memory communicate with each other, the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the combined navigation information fusion method considering driving state deviation as described in any one of claims 1-6.

10. An electronic device, characterized in that: include: A processor, a memory and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory so that the electronic device executes instructions for implementing the combined navigation information fusion method considering driving state deviation as described in any one of claims 1-6.

Images