Vehicle-mounted combined navigation trusted access method and system for GNSS recapturing

Abstract

The application discloses a vehicle-mounted combined navigation credible access method and system for GNSS reacquisition, and belongs to the technical field of vehicle-mounted combined navigation. During GNSS interruption, a bridging state and a prediction uncertainty are generated according to historical multi-source sequences, interruption duration and candidate observation-independent safe reference states; after GNSS re-output candidate observation, a multi-epoch reacquisition event window is constructed, and admission evidence is recursively generated according to the whitening bias of the candidate state relative to the safe reference state and the bridging state, cross-epoch consistency, motion accessibility and observation quality; the candidate observation is controlled to access through a four-state admission state machine with hysteresis, rollback, isolation and cooling functions, and the state injection increment is limited in the soft access state, and a safe navigation state is output. The application can reduce the risk of abnormal candidate observation entering the main fusion chain too early, and inhibit navigation state mutation and repeated switching of the admission state.

Description

Technical Field

[0001] This invention belongs to the field of vehicle-mounted integrated navigation technology, specifically relating to a trusted access method and system for vehicle-mounted integrated navigation oriented towards GNSS reacquisition. Background Technology

[0002] In-vehicle integrated navigation systems typically consist of a satellite navigation system, an inertial navigation system, and auxiliary sensors. They continuously output navigation status information such as position, velocity, and attitude during vehicle operation. The satellite navigation system provides position and velocity observations with absolute reference value, the inertial navigation system performs short-term continuous recursion based on acceleration and angular velocity output from the inertial measurement unit, and the auxiliary sensors provide supplementary information such as altitude, wheel speed, or vehicle motion constraints. Through multi-source information fusion, in-vehicle integrated navigation systems can balance absolute positioning accuracy and navigation continuity to a certain extent, and are therefore widely used in intelligent vehicles, autonomous driving, vehicle mapping, road inspection, and vehicle positioning scenarios in complex road environments.

[0003] During actual vehicle operation, when GNSS transitions from an interrupted state to re-outputting observation data, although the receiver has reacquired satellite signals, the observation state in the initial recovery phase is not necessarily stable immediately. Especially near tunnel exits, under viaducts, urban canyons, and densely built-up areas, the reacquired GNSS observations may still be affected by multipath propagation, non-line-of-sight errors, satellite geometric changes, and the receiver relocking process. Therefore, re-outputting GNSS observation data and ensuring that GNSS observation data can reliably participate in integrated navigation updates are two distinct issues.

[0004] Existing integrated navigation systems typically determine the usability of GNSS observations based on the observation residuals at the current epoch, the number of satellites, the carrier-to-noise ratio, or the satellite geometrical accuracy factor. This type of method primarily assesses quality at a single epoch, failing to fully utilize the changes between multiple consecutive candidate epochs after GNSS reacquisition. When a candidate epoch temporarily meets a preset threshold, but subsequent candidate epochs still exhibit significant fluctuations, single-epoch assessment may lead to unstable GNSS observations prematurely participating in integrated navigation updates.

[0005] Furthermore, existing technologies typically apply a fixed residual threshold directly to state components with different dimensions, such as position and velocity, or determine the reliability of a candidate state solely based on the difference between the current state and the GNSS candidate state. Such methods struggle to simultaneously reflect the uncertainties of the current safe state, the predicted state, and the individual GNSS candidate states. When the duration of GNSS outages varies, the level of inertial recursion error differs, or the accuracy of GNSS observations changes, using a fixed state difference may fail to produce consistent reliability assessment results.

[0006] Some existing methods utilize inertial recursion results, auxiliary sensor information, or neural network predictions from the GNSS outage period to compare candidate states after GNSS recovery. However, if the current GNSS candidate state is involved in both the generation of the reference state and the consistency check of the candidate state, information coupling may occur between the reference state and the state being checked, reducing the independence of the consistency judgment. Therefore, the bridging prediction state used to check the GNSS recapture candidate state needs to be independently generated from historical information already formed during the GNSS outage.

[0007] Existing technologies typically handle GNSS reacquisition states by directly accepting, directly rejecting, or adjusting observation weights. When the judgment result fluctuates around a threshold, GNSS observations may frequently switch between access and exit. While some methods can temporarily reject abnormal observations, they lack mechanisms for state rollback when evidence deteriorates, isolation mechanisms after detecting dangerous candidate states, and re-observation processes after isolation ends.

[0008] Meanwhile, when a GNSS candidate state initially meets the access conditions but its reliability has not yet been fully confirmed, injecting the complete state difference into the integrated navigation system at once may still cause abrupt changes in the navigation state. Existing technologies lack a mechanism to constrain the single injection increment of GNSS reacquisition candidate states by determining the upper limit of the state increment based on the GNSS outage duration and bridging prediction uncertainty.

[0009] Therefore, existing GNSS reacquisition processing methods have the following shortcomings: they rely heavily on single-epoch observation information and lack the accumulation of temporal evidence from multiple consecutive candidate epochs; uncertainties in different states are not fully incorporated into candidate state deviation evaluation; there may be information coupling between the bridging prediction reference and the current GNSS candidate state; the access decision lacks a state control process with hysteresis, backoff, isolation, and cooling functions; and there is a lack of risk constraints on injecting incremental states when candidate states are not yet fully reliable. These problems may affect the continuity and security of integrated navigation state updates during the GNSS reacquisition phase. Summary of the Invention

[0010] The purpose of this invention is to address the problems of unstable candidate observation quality in the early stages of GNSS reacquisition, repeated switching of the admission state caused by single-epoch judgment, and the direct impact of abnormal candidate observations on the integrated navigation output. This invention provides a time-series evidence-driven trusted access method for vehicle-mounted integrated navigation systems oriented towards GNSS reacquisition events. This method treats the re-output of candidate observations by GNSS and the safe entry of these candidate observations into the main fusion chain of integrated navigation as two separate issues. Through bridging prediction decoupled from candidate observations, cross-epoch evidence accumulation, an admission state machine with hysteresis and backoff capabilities, and restricted injection in soft access states, it achieves phased trusted access to GNSS reacquisition candidate states.

[0011] To achieve the above objectives, the present invention provides the following solution: a trusted access method for vehicle-mounted integrated navigation oriented to GNSS reacquisition, comprising the following steps: S1: Acquire multi-source data from the vehicle integrated navigation system, identify GNSS reacquisition events based on the multi-source data, and establish a reacquisition event window containing several candidate epochs; the reacquisition event window includes: GNSS reacquisition candidate status of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction status, and bridging prediction covariance matrix. S2: Based on the reacquisition event window, obtain the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, as well as the change direction consistency evaluation quantity and vehicle motion reachability index; S3: Generate GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtain single-epoch access evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle motion accessibility index. S4: Obtain the temporal admission evidence for candidate epochs based on the single epoch admission evidence, and control the GNSS reacquisition admission state machine based on the temporal admission evidence. S5: Output a safe navigation state based on the access state of the admission state machine; the safe navigation state is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation state recursion.

[0012] More preferably, the recapture event window include: ; In the formula, X r,k Indicates the first k Each candidate state is captured repeatedly; Q k Indicates the normalized GNSS observation quality; T out Indicates the duration of a fixed interruption; X b,k Indicates the first k The bridging prediction state corresponding to each candidate epoch; Σ b,k Indicates the first k The bridging prediction covariance matrix of each candidate epoch; k 0 indicates the start epoch of the recapture event.

[0013] More preferably, the method for obtaining the bridging prediction state includes: The historical multi-source navigation sequence, the duration of the GNSS outage, and the current safety status are input into the bridging prediction model to obtain the bridging height correction Δ. h b,k and bridging speed correction Δ v b,k ; The bridging prediction state is obtained based on the bridging height correction and the bridging speed correction. ; ; ; In the formula, the superscript T indicates transpose; h b,k and v b,k The first k The bridging prediction height and bridging prediction vertical velocity for each candidate epoch; h s,k and v s,k The first k The current safe altitude and current safe vertical velocity for each epoch.

[0014] More preferably, the method for obtaining the bridging prediction covariance matrix includes: ; in, ; ; In the formula, Σ b,k For the first k The bridging prediction covariance matrix of each candidate epoch; T 0 represents the normalized scale for GNSS outage duration; The uncertainty adjustment factor is greater than zero; Represents the basic prediction covariance matrix; r b,j For the first j One bridging prediction residual; This represents the sample mean of the bridging prediction residuals; M This represents the number of residual samples; ε To prevent the covariance matrix from becoming singular, regularization coefficients are used; I is the identity matrix.

[0015] More preferably, the first whitening deviation include: ; In the formula, Σ s,k Let be the covariance matrix of the current safe state. Σ r,k The covariance matrix of the GNSS recapture candidate states; Indicates the first k The current security status of each epoch; Second whitening deviation include: ; The consistency evaluation quantity of the direction of change include: ; In the formula, and These are the dimensionless increments of the recaptured candidate states and the increments of the neural network bridging states, respectively. The vehicle mobility index include: ; wherein The maximum permissible vertical acceleration is set according to the vehicle's range of motion; This represents the vertical acceleration between epochs.

[0016] More preferably, the single-epoch access evidence includes: ; ; In the formula, e k For the first k Single-epoch admission evidence for each candidate epoch; , , , , These are the non-negative evidence weights corresponding to the first whitening bias, the second whitening bias, the consistency evaluation of the direction of change, the vehicle mobility accessibility index, and the GNSS observation quality index, respectively, and the sum of each weight is 1.

[0017] More preferably, the admission state machine includes: an observation state, a soft access state, a full access state, and an isolation state; The soft access entry threshold, soft access exit threshold, full access entry threshold, and full access exit threshold must satisfy the following: ; In the formula, , , , These represent the soft access exit threshold, soft access entry threshold, full access exit threshold, and full access entry threshold, respectively. When the timing admission evidence is continuous N S When a candidate epoch is not lower than the soft access entry threshold and no dangerous transition is detected, the admission state machine transitions from the observation state to the soft access state. When the timing admission evidence is continuous N F When the number of candidate epochs is not lower than the full access threshold and no dangerous transition is detected, the admission state machine transitions from the soft access state to the full access state. When the timing admission evidence in the full access state is lower than the full access exit threshold, the admission state machine reverts to the soft access state. When the timing admission evidence in the soft access state is lower than the soft access exit threshold, the admission state machine reverts to the observation state.

[0018] More preferably, the method for determining whether the dangerous transition has occurred includes: The first whitening deviation exceeds the first isolation threshold; The second whitening deviation exceeds the second isolation threshold; The vehicle mobility accessibility evaluation value is lower than the lower limit of accessibility; The normalized GNSS observation quality is below the lower limit of observation quality; When a dangerous transition is detected, the admission state machine transitions from the current state to the isolation state and prohibits the GNSS reacquisition candidate state from entering the main fusion chain of the integrated navigation system. Continuous operation begins after the danger transition condition is resolved. N C A cooling window is set for each candidate epoch, during which the isolation state is maintained. After the cooling window ends, the timing admission evidence is reset to a preset initial value, and the isolation state is switched back to the observation state.

[0019] More preferably, in S5, when the admission state machine is in a soft access state, the upper limit of the state increment is determined based on the GNSS interruption duration and the bridging prediction uncertainty: ; in: ; In the formula, tr[ ] represents the trace of a matrix; B k This is the upper limit vector of the state increment allowed to be injected into the main fusion chain of the integrated navigation system under soft access conditions; B0 is the upper bound vector of the base state increment; u k This is a dimensionless representation of the uncertainty of the bridging prediction covariance after the state scaling matrix is ​​made dimensionless. k B This is the positive adjustment coefficient corresponding to the duration of the GNSS interruption; k U This is the positive adjustment coefficient corresponding to the bridging prediction uncertainty; T out Indicates the duration of a fixed interruption; T 0 represents the normalized scale for GNSS outage duration; S x This is the state scale matrix.

[0020] This invention also provides a trusted access system for vehicle-mounted integrated navigation for GNSS reacquisition, comprising: Data acquisition module: acquires multi-source data from the vehicle integrated navigation system, identifies GNSS reacquisition events based on the multi-source data, and establishes a reacquisition event window containing several candidate epochs; the reacquisition event window includes: GNSS reacquisition candidate status of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction status, and bridging prediction covariance matrix; Data processing module: Based on the reacquisition event window, obtain the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, as well as the change direction consistency evaluation quantity and vehicle motion reachability index; Admission evidence generation module: Generates GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtains single-epoch admission evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle motion accessibility index; Admission state control module: Based on the single epoch admission evidence, obtain the temporal admission evidence of the candidate epoch, and control the GNSS reacquisition admission state machine based on the temporal admission evidence; Navigation status output module: Outputs safe navigation status based on the access status of the admission state machine; the safe navigation status is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation status recursion.

[0021] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention treats the GNSS re-access process after it becomes available as an independent reacquisition event, constructing a reacquisition event window containing multiple consecutive candidate epochs, and directly restoring access without relying on whether a single GNSS epoch meets a fixed threshold. This allows for the utilization of changes in the reacquisition candidate state across multiple consecutive epochs, reducing the risk of erroneous access caused by single-epoch gross errors, short-term multipath effects, or sporadic observation jumps.

[0022] 2. This invention generates a bridging prediction state using historical multi-source data during GNSS outages, the duration of the outage, and the current security status, while explicitly excluding the current GNSS recapture candidate state from the input of the bridging prediction model. This allows the bridging prediction result to serve as a reference independent of the current GNSS candidate state, avoiding the coupling of judgment criteria caused by the recapture candidate state simultaneously participating in reference state generation and consistency verification, thus improving the independence of recapture reliability judgment.

[0023] 3. This invention calculates the whitening deviation of the recapture candidate state relative to the current safe state and the bridging prediction state, and introduces the corresponding state covariance matrix into the deviation calculation. Compared with directly comparing altitude difference and velocity difference, the whitening deviation can simultaneously reflect the dimensional differences of different state components and the uncertainty of state estimation, so that the deviation evaluation under different GNSS outage durations and different prediction uncertainty conditions has better consistency.

[0024] 4. This invention comprehensively utilizes whitening bias, candidate state change direction, vehicle mobility reachability, and GNSS observation quality to generate single-epoch access evidence, and recursively accumulates single-epoch access evidence from multiple candidate epochs. This transforms the GNSS reacquisition access decision from a single-epoch threshold judgment to a time-series evidence judgment, suppressing the direct impact of transient abnormal observations on access decisions from a methodological mechanism perspective, without relying on actual experimental results as access conditions.

[0025] 5. This invention establishes an admission state machine comprising observation, soft access, full access, and isolation states, and employs separate upgrade and rollback conditions to form hysteresis control. When the sequential admission evidence continuously improves, access is completed step-by-step according to the observation, soft access, and full access states; when the evidence deteriorates, state rollback is executed; when a dangerous candidate state is detected, the system enters the isolation state and initiates a cooling window. This reduces frequent switching of admission states near critical conditions and avoids direct restoration of access without re-verification after isolation ends.

[0026] 6. In soft access mode, this invention determines the upper limit of state increment based on the GNSS interruption duration and bridging prediction uncertainty, and uses the risk budget to limit the single injection increment of GNSS reacquisition candidate states. This avoids direct abrupt changes in the integrated navigation state caused by a large difference between the reacquisition candidate state and the current safe state, achieving restricted and gradual GNSS state access even before sufficient timing evidence is established.

[0027] 7. This invention integrates bridging prediction, timing evidence generation, admission state control, and state-restricted injection into a collaborative design, enabling the navigation state update process after GNSS reacquisition to have an observable, rollback, and isolable control path. It is suitable for vehicle-mounted integrated navigation scenarios where GNSS is reacquisitioned after an interruption but the reliability of the observation has not yet been fully confirmed. Attached Figure Description

[0028] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a schematic diagram of the overall process of a trusted access method for GNSS reacquisition in vehicle-mounted integrated navigation provided by an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the principle of bridging prediction state generation and recapture multi-epoch admission evidence generation provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of an admission state machine with hysteresis, rollback, isolation, and cooling control, and a restricted state injection method, provided for embodiments of the present invention. Detailed Implementation

[0030] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0032] Example 1: like Figure 1 As shown, this embodiment provides a trusted access method for vehicle-mounted integrated navigation for GNSS reacquisition, including the following steps: S1: Acquire multi-source data from the vehicle integrated navigation system, identify GNSS reacquisition events based on the multi-source data, and establish a reacquisition event window containing several candidate epochs; the reacquisition event window includes: GNSS reacquisition candidate status of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction status, and bridging prediction covariance matrix.

[0033] Specifically, in this embodiment, the multi-source data includes: GNSS data, SINS / IMU data, barometer data, and integrated navigation status data. The acquired multi-source data undergoes time synchronization, outlier removal, and coordinate unification processing. The GNSS data includes at least: GNSS position, GNSS velocity, carrier-to-noise ratio, number of available satellites, and satellite geometrical accuracy factor; the SINS / IMU data includes at least: vehicle acceleration, angular velocity, and inertial recursive state; the barometer data includes at least: barometric altitude or barometric altitude change; and the integrated navigation status data includes at least: current safe position, current safe velocity, and the corresponding state covariance matrix.

[0034] The GNSS reacquisition event begins when GNSS transitions from an interrupted or restricted state to a reacquisition state and outputs the first reacquisition candidate state. The reacquisition event window is updated according to the GNSS candidate epochs, and at least the following are recorded within the reacquisition event window: GNSS reacquisition candidate state, GNSS observation quality information, GNSS interruption duration, bridging prediction status, and the corresponding bridging prediction covariance matrix. The current reacquisition event ends when the admission state machine maintains full access for a consecutive preset epoch, or when the consecutive missing time of GNSS candidate observations exceeds a preset time. When the admission state machine enters the isolation state, the current reacquisition event does not immediately end; instead, it continues to monitor subsequent GNSS candidate epochs within a cooling window.

[0035] During GNSS outages or restrictions, construct historical multi-source navigation sequences, record the duration of lock-out, and maintain the current safe state. k The current security state of each epoch is represented as: (1) In the formula, h s,k For the first k The current safe height of each epoch; v s,k For the first k The current safe vertical velocity of each epoch; the superscript T indicates transpose.

[0036] In this embodiment, the current safe speed and current safe altitude are calculated recursively as follows: (2) (3) In the formula, a z,k The vertical acceleration is the result of attitude transformation and error compensation; Δ t The time interval between adjacent epochs; h baro,k To assist in determining the barometer's altitude; K b This is the pressure altitude correction factor.

[0037] The above recursive form is only used to illustrate one method of generating the current safe state. In other implementations, Kalman filtering, factor graph optimization, or other multi-source integrated navigation methods can also be used to maintain the current safe state.

[0038] It should be noted that the current safety status during the entire GNSS reacquisition event... X s,k Maintained by an independent, safe recursive chain that does not use GNSS reacquisition candidate observations. Safe reacquisition state generated under soft access conditions. X safe,k It serves only as an output state and is not fed back to the independent safety recursion chain, nor is it used as the current safety state in the bridging prediction model or whitening bias calculation. This avoids GNSS recapture candidate observations indirectly participating in the generation of the bridging reference state through the safety output state.

[0039] During GNSS interruption, the current safe state is maintained based on SINS / IMU data, barometer data, and the combined navigation status before the GNSS interruption, and historical multi-source data sequences generated during the GNSS interruption are preserved: (4) In the formula, L The length of the history window; F i Indicates the first i Historical characteristics of each era.

[0040] Among them, the i The historical characteristics of each epoch are represented as follows: (5) In the formula, h imu,i and v imu,i These are the inertial recursive height and the inertial recursive vertical velocity, respectively. e h,i and e v,i These are the highly consistent residuals and the velocity consistent residuals, respectively.

[0041] When GNSS is unavailable, the consistency residual can be the most recent valid residual before the GNSS interruption, the internal filtering residual of the integrated navigation system, or the auxiliary consistency residual between the inertial sensor and the barometer.

[0042] Record the duration of the GNSS outage: (6) In the formula, Indicates the first k The duration of GNSS interruption in each epoch.

[0043] When the GNSS outputs the first reacquisition candidate state, the accumulating GNSS interrupt duration is stopped, and the interrupt duration corresponding to the start epoch of the reacquisition event is used as the fixed interrupt duration for the current reacquisition event.

[0044] The fixed interrupt duration used within the current recapture event is: (7) In this embodiment, when GNSS is detected to have transitioned from an unavailable state to re-outputting observation data in a valid format, the corresponding epoch is determined as the starting epoch of the reacquisition event. k 0. The recapture event window includes: (8) In the formula, X r,k Indicates the first k Each candidate state is captured repeatedly; Q k Indicates the normalized GNSS observation quality; T out Indicates the duration of a fixed interruption; X b,k Indicates the first k The bridging prediction state corresponding to each candidate epoch; Σ b,k Indicates the first k The bridging prediction covariance matrix of candidate epochs.

[0045] Among them, the GNSS reacquisition candidate states are: (9) In the formula, , These represent the GNSS vertical altitude and GNSS vertical velocity of the current candidate epoch, respectively.

[0046] After the GNSS recapture event begins, the system does not directly replace the current safe state with the first GNSS candidate state. Instead, it continuously acquires candidate epochs in the recapture event window and enters the subsequent bridging prediction and admission evidence generation process.

[0047] Historical multi-source data sequences during GNSS outages H k GNSS interruption duration T out and current security status X s,k Input the bridging prediction model to obtain the bridging height correction and bridging speed correction.

[0048] It should be noted that the current GNSS reacquisition candidate state X r,k It is not used as input to the bridging prediction model. The bridging prediction model generates bridging prediction results only based on historical information already formed during the GNSS outage, so that the bridging prediction state can serve as a reference state independent of the current GNSS candidate state.

[0049] In this embodiment, the bridging prediction model employs a TCN-GRU-Attention temporal network. TCN is used to extract local change features of multi-source states during GNSS outages, GRU is used to extract temporal evolution features of the GNSS outage process, and Attention is used to determine the contribution of historical features at different times to the bridging prediction results.

[0050] The bridging prediction model outputs a bridging height correction Δ. h b,k and bridging speed correction Δ v b,k The bridging prediction state is generated according to the following formula: (10) (11) (12) In this embodiment, TCN-GRU-Attention is only a preferred implementation. In other implementations, the bridging prediction model can employ TCN, GRU, LSTM, Transformer, or other prediction models capable of handling multi-source time-series data.

[0051] To characterize the uncertainty of the bridging prediction results, a basic prediction covariance matrix is ​​constructed based on the prediction residuals of the bridging prediction model on historical available data. : (13) (14) In the formula, r b,j For the first j One bridging prediction residual; This represents the sample mean of the bridging prediction residuals; M This represents the number of residual samples; ε To prevent the covariance matrix from becoming singular, regularization coefficients are used; I is the identity matrix.

[0052] Adjust the base prediction covariance matrix based on the duration of GNSS outage: (15) In the formula, Σ b,k For the first k The bridging prediction covariance matrix of each candidate epoch; T 0 represents the normalized scale for GNSS outage duration; This is a positive zero uncertainty adjustment factor.

[0053] As the duration of GNSS outages increases, the bridging prediction covariance matrix increases accordingly to reflect the growing uncertainty of historical prediction information as the outage duration increases. The bridging prediction covariance matrix is ​​not a mandatory direct output of the bridging prediction model; it can also be determined jointly by historical residual statistics and the GNSS outage duration. In this embodiment, the GNSS reacquisition candidate states and the states updated from them do not directly or indirectly participate in the generation of the bridging prediction states and the bridging prediction covariance matrix.

[0054] S2: Based on the reacquisition event window, obtain the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, as well as the change direction consistency evaluation quantity and the vehicle motion reachability index.

[0055] For the first recapture event window k For each candidate epoch, the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state is calculated. : (16) In the formula, Σ s,k Let be the covariance matrix of the current safe state. Σ r,k This is the covariance matrix of the GNSS recapture candidate states.

[0056] Calculate the second whitening bias of the GNSS reacquisition candidate state relative to the bridging prediction state. : (17) Calculate the consistency evaluation metric between the change directions of adjacent GNSS reacquisition candidate states and the change directions of adjacent neural network bridging states. : (18) In the formula, and These are the dimensionless recapture candidate state increments and the neural network bridging state increments, respectively.

[0057] in, (19) (20) ;(twenty one) ;(twenty two) ;(twenty three) In the formula, , These represent the adjacent epoch state increments of the GNSS recapture candidate state and the adjacent epoch state increments of the neural network bridging state, respectively. S x This is the state scaling matrix; s h and s v These are the positive normalized scales for the height component and the vertical velocity component, respectively.

[0058] Calculate the inter-epoch vertical acceleration of the candidate state based on the vertical velocities of adjacent GNSS candidate epochs: .(twenty four) Vehicle mobility reachability indices are generated based on vertical acceleration between epochs: (25) wherein The maximum permissible vertical acceleration is set according to the vehicle's range of motion.

[0059] When the epochal changes of the candidate state satisfy the vehicle motion constraints. C a,k Approaching 1; when the GNSS candidate state exhibits an inter-epochal abrupt change that does not conform to the normal motion law of the vehicle, The corresponding decrease.

[0060] S3: Generate GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtain single-epoch access evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle motion accessibility index.

[0061] GNSS observation quality indices are generated based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor. : (26) (27) In the formula, Normalized carrier-to-noise ratio; To normalize the number of available satellites; This is the normalized satellite geometric precision factor; w C , w N and w D It is a non-negative quality weight.

[0062] Based on the first whitening bias, the second whitening bias, the consistency evaluation of the direction of change, the motion accessibility index, and the GNSS observation quality index, single-epoch access evidence is generated: (28) (29) In the formula, e k For the first k Single-epoch admission evidence for each candidate epoch; , , , , These are the non-negative evidence weights corresponding to the first whitening bias, the second whitening bias, the consistency evaluation of the direction of change, the vehicle mobility accessibility index, and the GNSS observation quality index, respectively, and the sum of each weight is 1.

[0063] S4: Obtain the temporal admission evidence for candidate epochs based on the single epoch admission evidence, and control the GNSS reacquisition admission state machine based on the temporal admission evidence.

[0064] like Figure 2 As shown, to avoid making admission decisions based on a single GNSS candidate epoch, single-epoch admission evidence from multiple candidate epochs is recursively accumulated: (30) (31) In the formula, Z k As of the date k Temporal admission evidence for each candidate epoch; λ This is the evidence memory coefficient.

[0065] At the start of the recapture event, the initial temporal admission evidence is set to the single-epoch admission evidence of the first candidate epoch, i.e.: (32) The initial timing admission evidence can also be set to a preset initial value according to security requirements. Through timing recursion accumulation, the timing admission evidence gradually meets the state upgrade conditions only when the GNSS candidate state maintains a high degree of consistency over multiple consecutive epochs; occasional anomalies in a single candidate epoch will not directly cause the GNSS to enter the full access state.

[0066] Based on time-series access evidence Z k The GNSS reacquisition admission state machine is controlled. In this embodiment, the admission state machine includes: observation state, soft access state, full access state, and isolation state, denoted as follows: O, S, F and R .

[0067] No. k Admission status corresponding to each candidate epoch M k Represented as: (33) Among them, observation status O Used to continuously acquire GNSS reacquisition candidate states and accumulate timing admission evidence, but not to allow GNSS reacquisition candidate states to directly replace the current safe state; soft access state S Used to introduce GNSS candidate states into integrated navigation updates in a restricted manner when the temporal admission evidence initially meets the access conditions; full access state F This is used to allow GNSS observations to participate in state updates in a normal integrated navigation manner when the temporal admission evidence continues to meet the credibility conditions; isolated state. R This is used to prevent GNSS candidate states from participating in integrated navigation updates when there are obvious anomalies in the GNSS candidate states or when the timing admission evidence continues to deteriorate.

[0068] Set soft access entry threshold, soft access exit threshold, full access entry threshold, and full access exit threshold, and satisfy the following: (34) In the formula, , , , These represent the soft access exit threshold, soft access entry threshold, full access exit threshold, and full access entry threshold, respectively.

[0069] like Figure 3 As shown, when the order of admission evidence is continuous N S Each candidate epoch is not lower than the soft access entry threshold. When no dangerous transition is detected, the admission state machine transitions from the observation state to the soft access state. When admission evidence is continuous... N F Each candidate epoch is not lower than the full access threshold. If no dangerous transition is detected, the admission state machine transitions from the soft access state to the full access state. When the admission evidence in the full access state is below the full access exit threshold... When the admission state machine reverts to the soft access state, the admission evidence in the soft access state falls below the soft access exit threshold. When the entry state machine reverts to the observation state, it switches to the isolation state when a dangerous transition is detected. After the dangerous transition condition is resolved, a cooling window is initiated, and the machine re-enters the observation state after the cooling window ends.

[0070] At the start of the GNSS reacquisition event, the admission state machine is initialized to the observation state: (35) Furthermore, a dangerous transition is determined to have been detected when any of the following conditions are met: First whitening deviation D s,k Exceeding the first isolation threshold; second whitening deviation D b,k Exceeding the second isolation threshold; vehicle mobility accessibility index C a,k Below the accessibility lower limit; normalized GNSS observation quality Q k The observation quality is below the lower limit. When a dangerous transition is detected, regardless of whether the admission state machine is currently in observation state, soft access state, or full access state, it will switch to isolation state and prohibit the recapture of candidate states from entering the main fusion chain of integrated navigation.

[0071] Once the danger transition condition is resolved, the continuous operation will begin. N C A cooling window is established for each candidate epoch. During this cooling window, the system remains isolated, continuing to monitor subsequent GNSS candidate epochs, but access to the GNSS candidate states is not restored. After the cooling window ends, if the GNSS continues to output candidate states, the admission state machine returns to the observation state and resets the timing admission evidence to preset initial values.

[0072] After the cooling window ends, if the GNSS is still able to output candidate states, the admission state machine returns from the isolated state to the observation state and reinitializes the timing admission evidence. (36) (37) In the formula, Z 0 represents the preset initial value for time-series admission evidence.

[0073] After the isolation state ends, the system cannot directly enter the soft access state or the full access state. Instead, it needs to undergo multi-epoch evidence accumulation again under the observation state. This prevents the system from immediately restoring GNSS access without re-verification after the disappearance of GNSS abnormal candidate states.

[0074] It should be noted that the cooling window is a time-constrained process after the isolation state, and is not a fifth admission state alongside the observation state, soft access state, full access state, and isolation state.

[0075] S5: Output safe navigation state based on the access state machine; the safe navigation state is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation state recursion.

[0076] When the admission state machine is in soft access state, the upper limit of the state increment is determined based on the GNSS outage duration and the bridging prediction uncertainty: (38) in: (39) In the formula, tr[ The ] symbol represents the trace of a matrix, which is the sum of the elements on the main diagonal of the matrix; B k This is the upper limit vector of the state increment allowed to be injected into the main fusion chain of the integrated navigation system under soft access conditions; B 0 is the upper limit vector of the basic state increment, which includes the upper limit of the height increment and the upper limit of the vertical velocity increment; u k This is a dimensionless representation of the uncertainty of the bridging prediction covariance after the state scaling matrix is ​​made dimensionless. k B This is the positive adjustment coefficient corresponding to the duration of the GNSS interruption; k U This represents the positive adjustment coefficient corresponding to the bridging prediction uncertainty. The longer the GNSS interruption duration or the greater the bridging prediction uncertainty, the higher the upper limit of the state increment. B k The smaller.

[0077] As the duration of GNSS outages increases and the uncertainty of bridging predictions grows, the risk budget for state updates decreases accordingly.

[0078] First, generate the target access status: (40) in: (41) In the formula, X tar,k The target access status; μ k represents the proximity coefficient of the candidate state.

[0079] Then, state increment limiting is performed: (42) In the formula, Δ X lim,k The state increment after being constrained by the upper limit of state increment; X safe,k 1 indicates the safe recapture state output by the previous candidate epoch; sat ( ) indicates according to B k The included upper limits for height increment and vertical velocity increment limit the amplitude of the corresponding state components of the target access state.

[0080] Last update to secure recapture status: (43) When the admission state machine is in the observation state, the GNSS reacquisition candidate state is not allowed to directly replace the current safe state. The system continues to maintain safe navigation output based on SINS / IMU data, barometer data and bridging prediction results.

[0081] When the admission state machine is in full access state, the upper limit constraint on state increment in soft access state is removed, allowing GNSS observations to participate in state estimation according to the system's original integrated navigation update method. The original integrated navigation update method can employ extended Kalman filtering, unscented Kalman filtering, factor graph optimization, or other multi-source integrated navigation methods.

[0082] When the admission state machine is in isolation, it blocks GNSS reacquisition candidate states from participating in integrated navigation updates and continues to output safe navigation results maintained by SINS / IMU data, barometer data, and the current safe state. The navigation states obtained under different admission states are unified as the output safe navigation state.

[0083] Through the above process, this invention divides the access process after GNSS reacquisition into four states: observation, soft access, full access, and isolation. A GNSS reacquisition candidate state can only participate in integrated navigation updates step by step after sufficient temporal admission evidence has been formed within multiple consecutive candidate epochs. When the evidence deteriorates or a dangerous candidate state is detected, the system can perform state rollback or isolation. In the soft access state, risk budgeting and state increment limiting are used to suppress navigation state abrupt changes caused by GNSS candidate states.

[0084] Example 2: This embodiment provides a trusted access system for vehicle-mounted integrated navigation for GNSS reacquisition, including: a data acquisition module: acquiring multi-source data from the vehicle integrated navigation system, identifying GNSS reacquisition events based on the multi-source data, and establishing a reacquisition event window containing several candidate epochs; the reacquisition event window includes: the GNSS reacquisition candidate state of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction state, and bridging prediction covariance matrix; and a data processing module: obtaining, based on the reacquisition event window, the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, the consistency evaluation quantity of the change direction, and the vehicle motion reachability. The system comprises the following modules: an admission evidence generation module, which generates GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtains single-epoch admission evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle mobility reachability indicator; an admission state control module, which obtains temporal admission evidence for candidate epochs based on the single-epoch admission evidence, and controls the GNSS reacquisition admission state machine based on the temporal admission evidence; and a navigation state output module, which outputs a safe navigation state based on the access state of the admission state machine. The safe navigation state is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation state recursion.

[0085] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A trusted access method for vehicle-mounted integrated navigation systems oriented towards GNSS reacquisition, characterized in that, Includes the following steps: S1: Acquire multi-source data from the vehicle integrated navigation system, identify GNSS re-acquisition events based on the multi-source data, and establish a re-acquisition event window containing several candidate epochs; The recapture event window includes: the GNSS recapture candidate status of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction status, and bridging prediction covariance matrix. S2: Based on the reacquisition event window, obtain the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, as well as the change direction consistency evaluation quantity and vehicle motion reachability index; S3: Generate GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtain single-epoch access evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle motion accessibility index. S4: Obtain the temporal admission evidence for candidate epochs based on the single epoch admission evidence, and control the GNSS reacquisition admission state machine based on the temporal admission evidence. S5: Output a safe navigation state based on the access state of the admission state machine; the safe navigation state is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation state recursion.

2. The trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition as described in claim 1, characterized in that, The recapture event window include: ； In the formula, X r,k Indicates the first k Each candidate state is captured repeatedly; Q k Indicates the normalized GNSS observation quality; T out Indicates the duration of a fixed interruption; X b,k Indicates the first k The bridging prediction state corresponding to each candidate epoch; Σ b,k Indicates the first k The bridging prediction covariance matrix of each candidate epoch; k 0 indicates the start epoch of the recapture event.

3. The trusted access method for vehicle-mounted integrated navigation oriented to GNSS reacquisition according to claim 2, characterized in that, The method for obtaining the bridging prediction state includes: The historical multi-source navigation sequence, the duration of the GNSS outage, and the current safety status are input into the bridging prediction model to obtain the bridging height correction Δ. h b,k and bridging speed correction Δ v b,k ; The bridging prediction state is obtained based on the bridging height correction and the bridging speed correction. ； ； ； In the formula, the superscript T indicates transpose; h b,k and v b,k The first k The bridging prediction height and bridging prediction vertical velocity for each candidate epoch; h s,k and v s,k The first k The current safe altitude and current safe vertical velocity for each epoch.

4. The trusted access method for vehicle-mounted integrated navigation oriented to GNSS reacquisition according to claim 2, characterized in that, The methods for obtaining the bridging prediction covariance matrix include: ； in, ； ； In the formula, Σ b,k For the first k The bridging prediction covariance matrix of each candidate epoch; T 0 represents the normalized scale for GNSS outage duration; The uncertainty adjustment factor is greater than zero; Represents the basic prediction covariance matrix; r b,j For the first j One bridging prediction residual; This represents the sample mean of the bridging prediction residuals; M This represents the number of residual samples; ε To prevent the covariance matrix from becoming singular, regularization coefficients are used; I is the identity matrix.

5. A trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition according to claim 4, characterized in that, First whitening deviation include: ； In the formula, Σ s,k Let be the covariance matrix of the current safe state. Σ r,k The covariance matrix of the GNSS recapture candidate states; Indicates the first k The current security status of each epoch; Second whitening deviation include: ； The consistency evaluation quantity of the direction of change include: ； In the formula, and These are the dimensionless increments of the recaptured candidate states and the increments of the neural network bridging states, respectively. The vehicle mobility index include: ； In the formula, The maximum permissible vertical acceleration is set according to the vehicle's range of motion; This represents the vertical acceleration between epochs.

6. A trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition according to claim 5, characterized in that, The single-epoch access evidence includes: ； ； In the formula, e k For the first k Single-epoch admission evidence for each candidate epoch; , , , , These are the non-negative evidence weights corresponding to the first whitening bias, the second whitening bias, the consistency evaluation of the direction of change, the vehicle mobility accessibility index, and the GNSS observation quality index, respectively, and the sum of each weight is 1.

7. A trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition according to claim 1, characterized in that, The admission state machine includes: observation state, soft access state, full access state, and isolation state; The soft access entry threshold, soft access exit threshold, full access entry threshold, and full access exit threshold must satisfy the following: ； In the formula, , , , These represent the soft access exit threshold, soft access entry threshold, full access exit threshold, and full access entry threshold, respectively. When the timing admission evidence is continuous N S When a candidate epoch is not lower than the soft access entry threshold and no dangerous transition is detected, the admission state machine transitions from the observation state to the soft access state. When the timing admission evidence is continuous N F When the number of candidate epochs is not lower than the full access threshold and no dangerous transition is detected, the admission state machine transitions from the soft access state to the full access state. When the timing admission evidence in the full access state is lower than the full access exit threshold, the admission state machine reverts to the soft access state. When the timing admission evidence in the soft access state is lower than the soft access exit threshold, the admission state machine reverts to the observation state.

8. A trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition according to claim 7, characterized in that, Methods for determining whether a dangerous transition has occurred include: The first whitening deviation exceeds the first isolation threshold; The second whitening deviation exceeds the second isolation threshold; The vehicle mobility accessibility evaluation value is lower than the lower limit of accessibility; The normalized GNSS observation quality is below the lower limit of observation quality; When a dangerous transition is detected, the admission state machine transitions from the current state to the isolation state and prohibits the GNSS reacquisition candidate state from entering the main fusion chain of the integrated navigation system. Continuous operation begins after the danger transition condition is resolved. N C A cooling window is set for each candidate epoch, during which the isolation state is maintained. After the cooling window ends, the timing admission evidence is reset to a preset initial value, and the isolation state is switched back to the observation state.

9. A trusted access method for vehicle-mounted integrated navigation based on GNSS reacquisition as described in claim 8, characterized in that, In S5, when the admission state machine is in soft access state, the upper limit of the state increment is determined based on the GNSS interruption duration and the bridging prediction uncertainty: ； in: ； In the formula, tr[ ] represents the trace of a matrix; B k This is the upper limit vector of the state increment allowed to be injected into the main fusion chain of the integrated navigation system under soft access conditions; B 0 is the upper bound vector of the base state increment; u k This is a dimensionless representation of the uncertainty of the bridging prediction covariance after the state scaling matrix is ​​made dimensionless. k B This is the positive adjustment coefficient corresponding to the duration of the GNSS interruption; k U This is the positive adjustment coefficient corresponding to the bridging prediction uncertainty; T out Indicates the duration of a fixed interruption; T 0 represents the normalized scale for GNSS outage duration; S x This is the state scale matrix.

10. A trusted access system for vehicle-mounted integrated navigation oriented to GNSS reacquisition, the system being used to implement the method according to any one of claims 1-9, characterized in that, include: Data acquisition module: acquires multi-source data from the vehicle integrated navigation system, identifies GNSS re-acquisition events based on the multi-source data, and establishes a re-acquisition event window containing several candidate epochs; The recapture event window includes: the GNSS recapture candidate status of the candidate epoch, GNSS observation quality information, GNSS outage duration, bridging prediction status, and bridging prediction covariance matrix. Data processing module: Based on the reacquisition event window, obtain the first whitening deviation of the GNSS reacquisition candidate state relative to the current safe state, the second whitening deviation relative to the bridging prediction state, as well as the change direction consistency evaluation quantity and vehicle motion reachability index; Admission evidence generation module: Generates GNSS observation quality indicators based on normalized carrier-to-noise ratio, normalized number of available satellites, and normalized satellite geometric accuracy factor, and obtains single-epoch admission evidence based on the GNSS observation quality indicators, the first whitening deviation, the second whitening deviation, the consistency evaluation quantity of the change direction, and the vehicle motion accessibility index; Admission state control module: Based on the single epoch admission evidence, obtain the temporal admission evidence of the candidate epoch, and control the GNSS reacquisition admission state machine based on the temporal admission evidence; Navigation status output module: Outputs safe navigation status based on the access status of the admission state machine; the safe navigation status is used for vehicle positioning, path tracking, vehicle control, and subsequent combined navigation status recursion.

Images