An online estimation and compensation method for random errors of IMU sensors

Abstract

The present invention belongs to the technical field of automobile navigation systems, and specifically relates to a method for online estimation and compensation of random errors of IMU sensors. On the one hand, a GNSS / INS integrated navigation quality control method based on new information anti-error can effectively solve the anti-error problem of GNSS / INS integrated navigation in complex environments; on the other hand, an IMU error online estimation algorithm optimized based on the short-term stability characteristics of IMU errors effectively solves the problem of the influence of gross error observation information on the online estimation of IMU errors. The present invention mainly improves the online estimation strategy of IMU error state in the design of GNSS / INS integrated navigation system, which can improve the accuracy and reliability of GNSS / INS integrated navigation in complex environments, and enhance the navigation and positioning service capabilities of the GNSS / INS integrated navigation system in vehicle-mounted autonomous driving.

Description

Technical Field

[0001] The present invention belongs to the technical field of automobile navigation systems, and in particular relates to an online estimation and compensation method for random errors of an IMU sensor. Background Art

[0002] An inertial navigation system (INS) is a fully autonomous dead-reckoning navigation system. Its primary inertial measurement unit (IMU) consists of a 3-axis gyroscope and a 3-axis accelerometer. INS can maintain high navigation accuracy for a short period of time. However, due to sensor errors, INS gradually diverges over time without additional assistance. This is particularly true for microelectromechanical systems (MEMS) and inertial measurement units (IMUs), where navigation errors are more rapid. Therefore, INS cannot operate independently for long periods of time. Therefore, optimizing IMU error estimation and compensation is a key research topic in integrated navigation and inertial navigation technologies.

[0003] Based on IMU error characteristics, IMU sensor errors can be divided into deterministic and random errors. Deterministic errors can generally be determined in advance through laboratory calibration methods (e.g., using a turntable). After calibration, their values ​​can be directly compensated into the IMU sensor output to eliminate the effects of deterministic errors. Random errors, however, arise from the influence of certain uncertainties and cannot be estimated and compensated using deterministic error calibration methods. Instead, they are typically estimated and compensated online within the integrated navigation system through error modeling. Once IMU deterministic errors have been effectively calibrated and compensated, random errors become the primary factor affecting the navigation performance of inertial navigation systems. Accurate analysis of IMU error characteristics (or models) is crucial for improving inertial navigation accuracy. Various random error model analysis methods are commonly used, with Allan variance being the most common and simple time-domain analysis method. Using the IMU error characteristics provided by Allan variance analysis, error models can be determined and designed. The IMU error model can then be augmented into the GNSS / INS integrated navigation state vector, enabling online estimation and compensation of IMU random errors.

[0004] As a high-precision global positioning system, the Global Navigation Satellite System (GNSS) uses information assistance to estimate and correct IMU sensor errors. However, due to the poor reliability of GNSS in dynamic environments and the vulnerability to signal loss, GNSS observations may contain gross errors. Therefore, to ensure the accuracy and reliability of GNSS / INS integrated navigation, it is necessary to design a quality control scheme that effectively removes gross errors and, at the same time, reduces or eliminates the impact of gross errors on the online estimation of IMU errors (i.e., the GNSS observation corrections for that epoch are not passed to the online IMU error estimation, and the IMU error parameters retain the values ​​of the previous epoch).

[0005] The existing GNSS / INS integrated navigation algorithm design mainly uses certain quality control methods to reduce the impact of gross error observations on the overall state estimation, but does not fully utilize the good short-term stability of IMU errors to carry out targeted online estimation and optimization methods of IMU error parameters. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide an online estimation and compensation method for random errors of IMU sensors, which can reduce the influence of GNSS gross error observations on the online estimation of IMU errors, ensure the stability and reliability of the online estimation of IMU errors, and thus realize high-precision and high-reliability navigation and positioning services.

[0007] The technical solution adopted by the present invention to solve the above technical problems is:

[0008] An online estimation and compensation method for random errors of IMU sensors is used in GNSS / INS integrated navigation systems. The method combines the auxiliary information of the GNSS navigation system to estimate the random errors of the IMU sensors of the INS navigation system online, including the following strategies:

[0009] A. Before performing online IMU error estimation, GNSS / INS integrated navigation quality control is performed. Specifically, the Kalman filter-based innovation vector is used to remove gross observation errors in the GNSS navigation system auxiliary information.

[0010] B. Optimize the online estimation of the IMU error. Specifically, analyze the IMU error characteristics through the Allan variance method and control the execution status of the online estimation of the IMU error by combining the IMU error stability and GNSS / INS quality control results.

[0011] Furthermore, the GNSS / INS integrated navigation quality control includes the following steps:

[0012] S1, using Kalman filter to detect the new information sequence of the GNSS navigation system auxiliary information, generating the new information vector

[0013] S2, recording the innovation vector obtained each time, and calculating the component variance in the innovation vector;

[0014] S3, comparing each component in the innovation vector with the component variance, and eliminating components whose magnitude exceeds a preset threshold, while eliminating their corresponding rows and columns in the measurement matrix H and the observation noise matrix R.

[0015] Furthermore, the detection statistic of the Kalman filter is in The detection statistic Subject to χ with m degrees of freedom 2 distribution, i.e. m is the new information The dimension of .

[0016] Furthermore, the IMU error online estimation optimization includes the following steps:

[0017] S1, use the Allan variance method to analyze the IMU error characteristics and establish the IMU error model;

[0018] S2, extending the IMU error model to a Klaman filter for online IMU error estimation;

[0019] S3, determine whether the new information detection result passes. If passed, feedback correction of the IMU error state is performed. If not passed, feedback correction of the IMU error state is not performed.

[0020] Furthermore, the calculation formula of the Allan variance is as follows:

[0021]

[0022] Among them, N consecutive sampling points y i , i=1,2,…,N, data sampling time is τ0, any time cluster τ=(n-1)·t0(n<N / 2), according to the length of time cluster τ, y i After separation, it becomes N c data blocks,

[0023] Furthermore, the IMU error model is as follows:

[0024]

[0025] Where b and s are IMU errors, including zero bias (b g ,b a ) and the scaling factor (s g ,s a ), T b ,T s is the correlation time of the first-order Gauss-Markov process.

[0026] Furthermore, the basis for judging whether the new information detection result passes is as follows:

[0027] like The test fails;

[0028] like The test fails;

[0029] Where TD is the preset limit.

[0030] Compared with the prior art, the present invention has the following main advantages:

[0031] 1. The main purpose of the GNSS / INS integrated navigation quality control process is to reduce or eliminate the influence of gross error observations in the optimal estimation process, so as to ensure the accuracy and reliability of integrated navigation positioning. The present invention proposes a GNSS / INS integrated navigation quality control method based on innovation-based robustness, which eliminates gross errors based on the innovation-based robustness method. At the same time, the weight of gross error observations is weakened by reducing the weight, instead of directly eliminating gross error observations, to construct a highly robust quality control scheme to ensure the accuracy and reliability of GNSS / INS integrated navigation in complex environments.

[0032] 2. The present invention also proposes an IMU error online estimation algorithm based on the optimization of the short-term stability characteristics of the IMU error. It is executed after the above-mentioned quality control method, and the IMU error stability characteristics and quality control results are used to optimize the IMU error online estimation strategy, further reducing the impact of gross error observations on the IMU error estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Schematic diagram of the IMU error estimation method in the prior art;

[0034] Figure 2 Schematic diagram of the overall method for online estimation and compensation of random errors of IMU sensors of the present invention;

[0035] Figure 3 This is a flow chart of a GNSS / INS integrated navigation quality control method according to an embodiment of the present invention;

[0036] Figure 4 This is a flow chart of the IMU error online estimation algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely for the purpose of explaining the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0038] It should be pointed out that, according to the needs of implementation, the various steps / components described in this application can be split into more steps / components, or two or more steps / components or partial operations of steps / components can be combined into new steps / components to achieve the purpose of the present invention.

[0039] The error states of a GNSS (Global Navigation Satellite System) / INS (Inertial Navigation System) integrated navigation system typically include navigation error states and sensor device error states. The former can directly correct the INS navigation results, ensuring high precision in the integrated navigation output; while the latter can be fed back into the raw sensor observations for correction, mitigating the impact of random IMU errors.

[0040] like Figure 1 As shown in the figure, existing GNSS / INS integrated navigation system designs typically keep the navigation error state and the IMU error state in the same frequency feedback loop. When GNSS auxiliary information is available, the system directly enters the GNSS / INS optimal estimation phase, and simultaneously feeds the navigation error state and the IMU error state back to the INS navigation results and IMU raw observation data in the same frequency mode for correction.

[0041] While existing IMU error estimation methods can achieve online IMU error estimation, GNSS signals are highly susceptible to interference from the external environment (such as obstruction or reflection from buildings, trees, and glass), making it difficult to effectively guarantee GNSS observation quality in complex urban scenarios. Gross errors are more likely to occur in scenarios with poor GNSS observation conditions. Directly importing this gross error observation information into the GNSS / INS combination optimal estimation process will directly affect the estimated accuracy of the navigation error state and the IMU error state.

[0042] In order to ensure the accuracy and reliability of GNSS / INS integrated navigation in complex dynamic environments, this patent adds an IMU random error online estimation processing method to the quality control link of GNSS / INS integrated navigation, and designs a new method for online estimation and compensation of IMU random errors. It can reduce the impact of GNSS gross error observations on the online estimation of IMU errors, ensure the stability and reliability of the online estimation of IMU errors, and thus realize high-precision and high-reliability navigation and positioning services.

[0043] like Figure 2 As shown, the present invention provides an online estimation and compensation method for random errors of IMU sensors. On the basis of the existing technology, it adds quality control and IMU error characteristics to constrain the GNSS / INS optimal estimation link, and adopts an asynchronous IMU error state feedback strategy to reduce the impact of gross GNSS observations on the stability of IMU errors. It mainly includes two parts: a GNSS / INS integrated navigation quality control method and an IMU error online estimation algorithm. Among them, the quality control method mainly reduces or eliminates the impact of gross error observations on navigation error state estimation under the condition that navigation accuracy allows; the IMU error online estimation algorithm uses the IMU error stability characteristics and quality control results to optimize the IMU error online estimation strategy to reduce the impact of gross error observations on IMU error state estimation.

[0044] (1) GNSS / INS integrated navigation quality control method based on new information anti-error

[0045] Quality control is an important part of the design of GNSS / ISN integrated navigation algorithm. Innovation filtering can quickly detect large anomalies, while innovation sequence detection can detect smaller anomalies within a period of time. Therefore, the present invention is based on the design characteristics of the vehicle-mounted combined algorithm, such as Figure 3 As shown in FIG, based on the Kalman filter innovation vector, the combined navigation gross error detection and fault inspection are performed.

[0046] For each calculated innovation, each component of the innovation is compared with its corresponding variance, and components exceeding the threshold are removed. If a component is removed, the corresponding row and column in the measurement matrix H and the observation noise R should also be removed.

[0047]

[0048] An r value of 2 or 3 corresponds to a confidence level of 95% and 99.73%, respectively.

[0049] When a system failure occurs, the mean of the new information is no longer zero. Therefore, by testing the mean of the new information, we can also determine whether the system has failed. Make a binary hypothesis:

[0050] H0: No fault

[0051] H1: No fault

[0052] The test statistic is:

[0053] It has been shown that the test statistic Subject to χ with m degrees of freedom 2 distribution, i.e. m is the new information The fault judgment criteria are as follows: (a) If (b) If It is determined that there is no fault. The preset threshold value TD can be determined by the false alarm rate.

[0054] When the Kalman filter χ 2 If the test fails, the filter is "soft reset". In order to ensure that the Kalman filter has been contaminated and become biased, the Kalman filter χ 2 The historical state of the test results, when the Kalman filter χ 2If the test fails all the time, perform a "soft reset" on the filter.

[0055] (2) Optimizing the IMU error online estimation algorithm based on the short-term stability characteristics of the IMU error

[0056] like Figure 4 As shown in Figure 1, this section is mainly divided into two parts: IMU error characteristic analysis and IMU error online estimation optimization. IMU error characteristic analysis uses the Allan variance method to analyze the type and stability level of IMU random errors; IMU error online estimation optimization controls the execution status of IMU error online estimation based on the IMU error stability level.

[0057] Allan variance has unique advantages in error analysis. Its main feature is that it can easily characterize and identify various error sources and their contributions to the characteristics of the entire noise system. It also has the advantages of being easy to calculate and separate. Its calculation formula is as follows:

[0058]

[0059] in, N N consecutive sampling points y i , i=1,2,…,N, the data sampling time is τ0. Any time cluster τ=(n-1)·t0(n<N / 2), according to the length of the time cluster τ, y i After separation, it becomes N c data blocks.

[0060] The IMU error characteristics are determined using the Allan variance analysis method, and a reasonable error model is established. Considering the universality of the GNSS / INS integrated navigation algorithm, the GNSS / INS integrated navigation algorithm designed in this paper simultaneously estimates the IMU bias and scale factor online. Both the gyro and accelerometer bias and scale factor can be modeled as a first-order Gauss-Markov process, as shown in the following equation.

[0061]

[0062] Where b, s are IMU errors, including zero bias (b g ,b a ) and the scaling factor (s g ,s a ). b ,T s is the correlation time of the first-order Gauss-Markov process (note that T b With T sIn the design of the GNSS / INS optimal estimation algorithm, the modeled IMU error is extended to the Klaman filter for online estimation. After the optimal estimation is completed, it is necessary to introduce the new information χ 2 The judgment result of the test is: (a) If χ 2 If the test fails, the IMU error state feedback correction will not be performed; (b) If χ 2 If the test is passed, the IMU error state feedback correction will be performed.

[0063] In summary, the present invention mainly addresses the impact of GNSS observations containing gross errors on the estimated IMU error state, thereby ensuring the accuracy and reliability of the GNSS / INS integrated navigation system. The main technical problems solved by the present invention are mainly the following two parts:

[0064] (1) Solve the problem of GNSS / INS integrated navigation anti-error in complex environments

[0065] The main purpose of the GNSS / INS integrated navigation quality control process is to reduce or eliminate the influence of gross error observations in the optimal estimation process, so as to ensure the accuracy and reliability of integrated navigation positioning. The present invention proposes a GNSS / INS integrated navigation quality control method based on innovation-based robustness, which eliminates gross errors based on the innovation-based robustness method. At the same time, the weight of gross error observations is weakened by down-weighting, instead of directly eliminating gross error observations. A highly robust quality control scheme is constructed to ensure the accuracy and reliability of GNSS / INS integrated navigation in complex environments.

[0066] (2) Solve the problem of the impact of gross error observation information on the online estimation of IMU error

[0067] This paper proposes an online IMU error estimation algorithm based on the optimization of the short-term stability characteristics of IMU errors. This algorithm, executed after the aforementioned quality control method, utilizes the IMU error stability characteristics and quality control results to optimize the IMU error online estimation strategy, further reducing the impact of gross error observations on IMU error estimation. First, the IMU error characteristics are determined through Allan variance to determine their stability performance and whether the conditions for optimizing the online IMU error estimation are met. Then, based on the quality control results of the aforementioned GNSS / INS integrated navigation, it is determined whether to perform the online IMU error estimation optimization. Finally, combining the IMU error stability and GNSS / INS quality control results, online IMU error estimation is not performed at the moment of GNSS gross error (i.e., the IMU error parameters retain the estimated values ​​at the previous moment).

[0068] Based on the above method, the present invention also provides:

[0069] A vehicle navigation system includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the above-mentioned method for online estimation and compensation of random errors of an IMU sensor is implemented.

[0070] A non-transitory readable storage medium stores a program which, when executed by a vehicle-mounted navigation system, implements the above-mentioned method for online estimation and compensation of random errors of an IMU sensor.

[0071] A car comprises the above-mentioned in-car navigation system.

[0072] The above embodiments are intended only to illustrate the design concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of protection of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design concepts disclosed in the present invention are within the scope of protection of the present invention.

Claims

1. A method for online estimation and compensation of random errors of IMU sensors, used in GNSS / INS integrated navigation systems, combines auxiliary information of GNSS navigation systems to perform online estimation of random errors of IMU sensors of INS navigation systems, characterized by: These strategies include: A. Before performing online IMU error estimation, perform GNSS / INS integrated navigation quality control, specifically: S11, using Kalman filter to detect the new information sequence of the GNSS navigation system auxiliary information, generating an new information vector S12, recording the innovation vector obtained each time, and calculating the component variance in the innovation vector; S13, comparing each component in the innovation vector with the component variance, and removing components whose magnitude exceeds a preset threshold, and simultaneously removing their corresponding rows and columns in the measurement matrix H and the observation noise matrix R; B. Perform online estimation and optimization of IMU error, specifically: S21, use the Allan variance method to analyze the IMU error characteristics and establish the IMU error model; S22, extending the IMU error model to a Klaman filter to perform online IMU error estimation; S23, determine whether the new information detection result passes, if passed, then perform feedback correction of the IMU error state, if not passed, then do not perform feedback correction of the IMU error state.

2. The method for online estimation and compensation of random errors of an IMU sensor according to claim 1, characterized in that: The detection statistic of the Kalman filter is in The detection statistic Subject to χ with m degrees of freedom 2 Distribution, that is m is the new information The dimension of .

3. The method for online estimation and compensation of random errors of an IMU sensor according to claim 1, characterized in that: The calculation formula of the Allan variance is as follows: Among them, N consecutive sampling points y i , i=1,2,…,N, data sampling time is τ0, any time cluster τ=(n-1)·t0, n<N / 2, according to the length of time cluster τ, y i After separation, it becomes N c data blocks, 4. The method for online estimation and compensation of random errors of an IMU sensor according to claim 1, characterized in that: The IMU error model is as follows: Where b and s are IMU errors, including zero bias (b g ,b a ) and the scaling factor (s g ,s a ), T b ,T s is the correlation time of the first-order Gauss-Markov process.

5. The method for online estimation and compensation of random errors of an IMU sensor according to claim 1, characterized in that: The basis for judging whether the innovation detection result passes is as follows: like The test fails; like The test fails; Where TD is the preset limit.

6. A vehicle navigation system comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 5 is implemented.

7. A non-transitory readable storage medium having a program stored thereon, characterized in that: When the program is executed by a vehicle controller, the method according to any one of claims 1 to 5 is implemented.

8. An automobile, characterized in that: Including the vehicle navigation system according to claim 6.

Images