Method and device for realizing autonomous integrity monitoring of receiver

Abstract

The invention discloses a method and a device for realizing autonomous integrity monitoring of a receiver, and the method comprises the steps: carrying out the comprehensive analysis through combining a real observation value and a virtual observation value with a state estimation result of the receiver, and comprehensively considering the measurement deviation and observation noise to calculate a protection level. The precision and efficiency of integrity monitoring and protection grade evaluation of the receiver are improved, high-precision and high-safety positioning calculation is realized, and reliable navigation decisions can be made in different environments.

Description

Technical Field

[0001] The present application relates to but is not limited to integrated navigation technology, and in particular to a method and device for implementing autonomous integrity monitoring of a receiver. Background Art

[0002] For positioning and navigation applications such as aviation and autonomous driving, the receiver not only needs to provide high-precision positioning results, but also needs to meet functional safety requirements. This means that the receiver must have the ability to detect and eliminate failures, and be able to monitor and quantify the safety of positioning results in real time. This safety is usually measured by integrity indicators, which reflect the confidence level of the location information provided by the positioning system. The integrity indicator is often quantified by the protection level (PL, Protection Level). The role of the protection level is to ensure that the positioning error does not exceed a certain threshold. The size of this threshold is determined based on the integrity risk that the system can withstand. The formula is expressed as: P (positioning error > protection level) ≤ integrity risk.

[0003] In order to achieve the above functions, some GNSS receivers have introduced Receiver Autonomous Integrity Monitoring (RAIM) technology. RAIM technology was introduced into the field of satellite navigation by the aviation industry. After years of development, it can basically meet the requirements of automatic navigation, take-off and landing of aircraft. However, the RAIM method cannot meet the requirements of consumer-level applications such as automatic driving of ground vehicles and autonomous flight of drones. As a result, it cannot provide high-reliability positioning and navigation solutions for the back-end platform, nor can it quantify the integrity indicators of the provided positioning and navigation solutions, and thus cannot evaluate the reliability and safety of the navigation solutions. Summary of the invention

[0004] The present application provides a method and device for implementing autonomous integrity monitoring of a receiver, which can achieve high-precision and high-security positioning solutions and ensure that reliable navigation decisions can be made in different environments.

[0005] The present application provides a method for implementing autonomous integrity monitoring of a receiver, including:

[0006] Obtaining observation values ​​of the receiver on the satellite signal and the state estimation result of the receiver; wherein the observation values ​​include real observation values ​​and virtual observation values;

[0007] According to the observation value and the state estimation result, a state estimation error caused by a measurement deviation and a state estimation error caused by an observation noise are obtained;

[0008] The protection level of the receiver is calculated according to the state estimation error caused by the measurement bias and the state estimation error caused by the observation noise to monitor the autonomous integrity of the receiver.

[0009] An embodiment of the present application also provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute any of the above methods for implementing autonomous integrity monitoring of a receiver.

[0010] An embodiment of the present application also provides a computer device, including a memory and a processor, wherein the memory stores the following instructions executable by the processor: used to execute the steps of any of the above-mentioned methods for implementing autonomous integrity monitoring of the receiver.

[0011] The solution provided in the embodiment of the present application combines real observation values ​​and virtual observation values, performs a comprehensive analysis with the state estimation results of the receiver, and comprehensively considers measurement deviations and observation noise to calculate the protection level, thereby improving the accuracy and efficiency of receiver integrity monitoring and protection level assessment, achieving high-precision and high-security positioning solutions, and ensuring that reliable navigation decisions can be made in different environments.

[0012] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings are used to provide further understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application and do not constitute a limitation on the technical solution of the present application.

[0014] Figure 1 A flowchart of a method for implementing autonomous receiver integrity monitoring in an embodiment of the present application;

[0015] Figure 2 Schematic diagram of probability distribution of a WSSE-based detection system in a normal state and a fault state in an embodiment of the present application;

[0016] Figure 3 A schematic diagram of an experimental track of a sports car experiment in an embodiment of the present application;

[0017] FIG4( a ) is a schematic diagram of a curve showing a change in positioning error and protection level in the east direction in an embodiment of the present application;

[0018] FIG4( b ) is a schematic diagram of a curve showing a change in positioning error and protection level in the north direction in an embodiment of the present application;

[0019] FIG5( a ) is an example of an eastward Stanford diagram for evaluating the performance of a method for calculating a protection level of a navigation system in an embodiment of the present application;

[0020] FIG5( b ) is an example of a north-facing Stanford diagram for evaluating the performance of a method for calculating a protection level of a navigation system in an embodiment of the present application;

[0021] Figure 6 Schematic diagram of the structure of the device for implementing autonomous integrity monitoring of the receiver in an embodiment of the present application. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the present application more clear, the embodiments of the present application will be described in detail below with reference to the accompanying drawings. It should be noted that the embodiments and features in the embodiments of the present application can be combined with each other arbitrarily without conflict.

[0023] In order to facilitate understanding of the present application, the present application will be described more fully below with reference to the relevant drawings. Embodiments of the present application are provided in the drawings. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of this application are only for the purpose of describing specific embodiments and are not intended to limit this application.

[0025] It is understood that the terms "first" and "second" used in this application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0026] It can be understood that the “connection” in the following embodiments should be understood as “electrical connection”, “communication connection”, etc. if the connected circuits, modules, units, etc. have electrical signals or data transmission between each other.

[0027] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0028] RAIM technology uses redundant observations to perform consistency checks to detect and eliminate failures in the system. The protection level is calculated based on the statistical parameters of the signal, including the user ranging accuracy σ URA (URA, User Range Accuracy), nominal lateral deviation b nom (Nominal Bias), satellite failure probability P sat and system failure probability P const In addition, the quantification of the protection level is also subject to the integrity margin (P int ) and continuity margin (P cont ) can be expressed as: PL = f(σ URA ,b nom ,P sat ,P const ,P int ,P cont ). The quantification of the integrity parameters of the signal needs to be quantified by statistical analysis of the signal in advance, and the function f(·) depends on the design of the RAIM algorithm and the actual operating status of the system. Providing an envelope measure of the positioning error in real time is a complex task, which requires accurate quantification of the statistical characteristics of the measurement error, designing the correct filter, and relying on a rigorous integrity algorithm. Therefore, the design of the RAIM algorithm not only needs to solve the problem of failure detection, but also must deal with how to effectively calculate and monitor the protection level in a resource-constrained system.

[0029] RAIM technology is mainly divided into two categories according to the different detectors used: classic RAIM based on innovation or residual and advanced RAIM based on solution separation. Among them, innovation refers to the difference between the current observation value and the observation value predicted based on the previous state. In classic RAIM, the statistical properties of innovation (such as the norm or weighted norm of innovation) can be used to detect errors or faults in the measurement data, and then perform positioning correction or alarm processing. Residual refers to the difference between the observation value and the predicted value in a more general sense. In RAIM technology, residual is widely used to detect measurement faults. Excessively large residuals often mean that there may be satellite signal errors, environmental interference or other measurement anomalies.

[0030] Classical RAIM performs fault detection in the measurement value domain, and determines whether there is a fault by comparing the measurement innovation or residual with the prior statistical distribution of the measurement error. The error in the measurement value domain is then projected into the position domain, that is, these errors are converted into effects on the receiver position estimate, and the protection level is calculated based on this projection. Advanced RAIM detects faults by running the main estimator and multiple sub-estimators in parallel, comparing the state estimate difference between the main estimator and each sub-estimator (i.e., solution separation), and can directly calculate the protection level.

[0031] Although RAIM technology has been widely used in the field of aviation navigation and can meet the needs of automatic navigation and take-off and landing of aircraft, there are still some technical bottlenecks in the RAIM method in consumer-level applications, such as ground vehicle automatic driving and autonomous flight of drones. First, the existing RAIM algorithm has large computational and memory overheads, especially in advanced RAIM, which requires multiple Kalman filters to run in parallel, which greatly increases the computational burden and makes it difficult to apply it to resource-constrained embedded integrated navigation systems. Secondly, the existing RAIM method is difficult to adapt to various integrated navigation working modes. Under different satellite visibility conditions, the integrated navigation system will switch different sensor fusion and estimation modes (such as loose combination, tight combination, least squares method and Kalman filter), and RAIM lacks a general framework to adapt to these changes, making it difficult to achieve optimal performance in all scenarios.

[0032] In addition, the RAIM algorithm is mainly designed for satellite navigation systems. However, in complex environments, when satellite signal obstruction causes the navigation system to rely on single inertial navigation or single inertial dead reckoning, RAIM cannot provide a corresponding protection level quantification method. The calculation of protection level under single inertial dead reckoning conditions is still an unresolved problem. Finally, the RAIM algorithm is highly dependent on the error model. The calculation of the protection level depends on the statistical characteristics of the measurement error. However, in actual operation, the distribution of measurement errors is often difficult to quantify accurately. Even if the error model is known, the parameters will be adjusted due to performance optimization during actual use, which makes the existing RAIM method inflexible when dealing with these non-ideal errors and estimation models.

[0033] The present application embodiment proposes a method for implementing autonomous receiver integrity monitoring, such as Figure 1 As shown, this may include:

[0034] Step 100: Obtain the receiver's observation values ​​of satellite signals and the receiver's state estimation results; wherein the observation values ​​include real observation values ​​and virtual observation values.

[0035] In an exemplary embodiment, obtaining the observation value of the receiver on the satellite signal and the state estimation result of the receiver may include:

[0036] The observation value is obtained by performing gross error detection on the observation data of the satellite signal by the receiver;

[0037] The state of the receiver is estimated based on the observed values ​​after gross error detection to obtain the state estimation result.

[0038] In an exemplary embodiment, performing gross error detection based on observation data of satellite signals by a receiver may include:

[0039] According to the observation values ​​of the receiver on the satellite signal, such as the global navigation satellite system (GNSS) observation values, the position of the receiver is solved by, for example, the least square method and the measurement residual is obtained; and the gross error detection is performed according to the obtained measurement residual.

[0040] In one embodiment, the GNSS observation value may include but is not limited to raw observation data such as pseudorange, Doppler observation, etc.

[0041] In an exemplary embodiment, the navigation solution process includes solving the receiver position x by using the GNSS observation model as shown in formula (1):

[0042] y=Hx+b+∈ (1)

[0043] In formula (1), y represents the GNSS observation value; H represents the geometric matrix, which describes the geometric relationship between the satellite and the receiver; b and ∈ are the measurement bias and observation noise respectively, and the observation noise ∈ obeys a Gaussian distribution with a mean of zero.

[0044] In the weighted least squares method, the system state estimate is solved The equation is shown in formula (2):

[0045]

[0046] In formula (2), ∑ represents the observation noise covariance matrix, H + =(H T ∑ -1 H) -1 H T ∑ -1 is the pseudo-inverse matrix.

[0047] The estimated error is shown in formula (3):

[0048]

[0049] The estimation error reflects the deviation caused by the measurement bias b and the observation noise ∈. Therefore, the state estimation after least squares solution is It may be affected by measurement deviation (b≠0), resulting in inaccurate solution results.

[0050] Since the true state x cannot be observed directly, the estimation error is not observable. Therefore, the measurement residual To detect gross errors, the measurement residual is defined as shown in formula (4):

[0051]

[0052] By substituting the least squares estimation formula (2), since SH = 0, we can get formula (5):

[0053]

[0054] In formula (5), S = [IH (H T ∑ -1 H) -1 H T ∑ -1 ], represents the residual generation matrix. As shown in formula (5), the residual only contains the measurement bias b and observation noise ∈, that is,

[0055] In an exemplary embodiment, in order to quantitatively detect gross errors, the weighted sum of squared errors (WSSE) is defined as a detection statistic as shown in formula (6):

[0056]

[0057] As shown in formula (6), WSSE = [S(b+∈)] T ∑ -1 S(b+∈), WSSE includes measurement bias b and observation noise ∈ information and is an observable quantity.

[0058] Assuming that the observation noise ∈ follows a Gaussian distribution with a mean of zero, when there is no measurement bias, that is, b = 0, the WSSE follows a degree of freedom of N meas -N state When there is a measurement bias b≠0, WSSE follows a non-central chi-square distribution with the same degree of freedom. The probability distributions of the two scenarios are as follows: Figure 2 As shown, Figure 2 In the figure, the x-axis is WSSE, the y-axis is the probability density P of WSSE, curve 21 represents the normal state (Nominal), and curve 22 represents the faulty state (Faulty). At the same time, the figure shows two types of false detection rates: area 24 represents false alarm (False Alarm), and area 23 represents missed detection (Missed Detection). Figure 2 The distribution characteristics of the system in fault and normal states when performing detection based on WSSE are shown, and the trade-off between false positives and missed detections that are inevitable in the detection system is shown, such as Figure 2 As shown, the closer the threshold is to WSSE=20 (the intersection of curve 21 and curve 22), the lower the false alarm rate is, but the missed detection rate may increase; on the contrary, the farther the threshold is from the intersection, the missed detection rate is reduced but the false alarm rate is increased.

[0059] The goal of gross error detection based on WSSE is to determine whether there is a gross error (measurement deviation) in the observed data. In one embodiment, a threshold T can be set. When the WSSE exceeds the threshold, it is considered that there is a gross error (a fault). The fault detection rule is shown in formula (7):

[0060]

[0061] In formula (7), H0 is the null hypothesis, indicating that there is no fault, and H1 is the alternative hypothesis, indicating that there is a fault. The threshold T can be calculated based on the allowed false alarm rate (Probability of False Alarm) P fa and system degrees of freedom N meas -N state The inverse function of the cumulative distribution function F is obtained through the chi-square distribution -1 Determine: T = F -1 (P fa , N meas -N state ), where P fa Also known as continuity budget, it is a design indicator parameter.

[0062] In gross error detection, the receiver position is calculated using the least squares method Generate residuals This provides a basis for gross error detection. Gross error detection is based on residual statistics. By measuring residuals and WSSE, it is detected whether there is a gross error. In gross error detection, if a gross error is detected (such as WSSE exceeds the threshold T), the corresponding pseudorange data can be eliminated and the navigation solution can be re-performed until all observation data pass the test. Navigation solution generates residuals for gross error detection, and gross error detection improves the reliability and accuracy of navigation solution; through least squares solution and WSSE statistics, abnormal observation data can be effectively identified and eliminated, ensuring the accuracy of navigation results.

[0063] In an exemplary embodiment, the initial state of the receiver may include, but is not limited to: initial position, speed, time deviation, etc.

[0064] The initial state estimation depends on the variance distribution of the observed value, and the variance distribution of the observed value needs to be dynamically adjusted according to the residual of the initial estimation. This process is actually an iterative process. In an exemplary embodiment, obtaining the state estimation result of the receiver may include:

[0065] Estimate the state of the receiver based on the observed values ​​after gross error detection, and obtain estimated values ​​of the position, speed, etc.

[0066] Calculate residuals from state estimates and actual observations;

[0067] The variance of each observation value is adaptively adjusted according to the calculated residual, and the state is re-estimated after adjustment until the variance of the state estimate and the observation value converges to obtain a more accurate state estimate.

[0068] Based on the observations after gross error detection, the weight of each observation is dynamically adjusted through the residual, further improving the accuracy of state estimation.

[0069] In an exemplary embodiment, the real observation value is the actual measurement data; the virtual observation value is a pseudo observation value constructed based on the state prediction. The state prediction result at each moment can be a recursion based on the state estimation result at the previous moment, that is, the state estimation result at the previous moment is used to predict the state at the current moment to obtain the state prediction result at the current moment. The state estimation result at each moment can be the result obtained by state estimation based on the observation value, or it can be the result obtained by correcting the state prediction result based on the observation value.

[0070] In one embodiment, it is possible to use Calculate the virtual observation value z pred ;in, is the state prediction result at the current time k, L -1 Represents the inverse matrix of the lower triangular matrix obtained by decomposing the covariance matrix P of the state estimation result.

[0071] Step 101: According to the observation value and the state estimation result, the state estimation error caused by the measurement bias and the state estimation error caused by the observation noise are obtained.

[0072] From formula (3), we can see that the state estimation error can be caused by measurement bias and observation noise. Therefore, the position error PE (i.e., prediction error) consists of two parts: the error PE caused by measurement bias and b and the error PE caused by observation noise ∈ , as shown in formula (8):

[0073] PE=PE b +PE ∈ (8)

[0074] The position error model shown in formula (8) reflects that the state estimation error is caused by both measurement bias and observation noise.

[0075] Assuming that only one satellite j fails in the system, combined with formula (3), the i-direction error PE caused by the measurement bias is b As shown in formula (9):

[0076]

[0077] In formula (9), Denotes the observation matrix H + The element corresponding to the i-direction and the j-th observation in b j It represents the measurement deviation caused by the failure of the jth satellite.

[0078] The detection statistic WSSE caused by measurement deviation is used to measure the impact of measurement deviation on the system, as shown in formula (10):

[0079]

[0080] In formula (10), S j,j represents the weight matrix element in the system, σ j represents the standard deviation of the observation noise associated with satellite j. As shown in formula (10), the detection statistic WSSE caused by the measurement bias is calculated by the weighted residual of the observation data, reflecting the impact of the measurement bias on the system state estimation.

[0081] Combining formula (9) and formula (10), the comprehensive error PE can be obtained b,i As shown in formula (11), it represents the error in the i-direction caused by the measurement deviation of a certain observation value j:

[0082]

[0083] In an exemplary embodiment, step 101 may include:

[0084] Obtaining a state estimation error caused by the measurement deviation according to a characteristic slope of the observation value and a detection statistic caused by the measurement deviation;

[0085] The state estimation error caused by observation noise is obtained according to the covariance of the state estimation result.

[0086] In one embodiment, the characteristic slope of observation j for the i-direction is defined as The worst-case deviation effect is calculated using the characteristic slope to obtain the upper limit of the position error, that is, the state estimation error caused by the measurement deviation in step 101, as shown in formula (12):

[0087]

[0088] The state estimation error caused by observation noise obeys the normal distribution. The covariance matrix P is used to describe the probability distribution of the state estimation error, as shown in formula (13):

[0089]

[0090] For weighted least squares P = (H T ∑H) -1 , and the position-dependent term P pos It can be transformed into the form shown in formula (14), that is, the relevant part P of the covariance matrix P pos Can be used to calculate the position error:

[0091]

[0092] The integrity budget is the probability of missed alarm P. MD (The missed alarm probability indicates the probability that the system fails to detect an error in the absence of a fault) and determines the corresponding standard deviation multiplier k D As shown in formula (15):

[0093] k D =Q -1 (P MD ) (15)

[0094] In one embodiment, in formula (15), Q -1 (·) is the inverse function of the normal cumulative distribution, in 1-P MD Under the confidence level of ∈,i That is, the state estimation error caused by the observation noise in step 101 is shown in formula (16):

[0095]

[0096] In formula (16), the standard deviation multiplier k D and the probability of missed alarm P MD Correlation. The diagonal elements P of the covariance matrix P i,i represents the variance associated with the estimation error of the ith state variable (such as position, velocity, or other state).

[0097] Step 102: Calculate the protection level of the receiver according to the obtained state estimation error caused by the measurement bias and the state estimation error caused by the observation noise to monitor the autonomous integrity of the receiver.

[0098] Combined with the influence of measurement deviation and observation noise, the i-direction protection level PL i As shown in formula (17):

[0099]

[0100] In one embodiment, the protection level shown in formula (17) can be used in at least three directions, such as northeast, upper right front of the vehicle, or x, y, z directions in the ECEF coordinate system.

[0101] For the horizontal protection level (HPL), it can be based on the protection level PL in the east-west direction. E and protection level PL in the north direction N Calculate, as shown in formula (18):

[0102]

[0103] The above protection level calculation method has two assumptions: first, the observation error model used satisfies the Gaussian distribution and σ j The error standard deviation can be described in an envelope manner; second, the system model is linear. However, the above assumptions are difficult to guarantee in engineering practice, so the protection level calculated according to formula (17) may be distorted.

[0104] In order to overcome these problems and further improve the accuracy of protection level calculation, in one embodiment, adjustment coefficients k may be introduced for the deviation protection level and the noise protection level, respectively. b and k ∈ The protection level is adjusted, and the calculation formula of the adjusted protection level is shown in formula (19):

[0105]

[0106] In formula (19), k ∈ =k n k D , k b and k n The parameters can be adjusted according to the operating results of actual data, or trained through machine learning methods.

[0107] In one embodiment, k b k is the adjustment factor used to correct the protection level calculation result caused by measurement deviation. Since the measurement deviation model in the actual navigation system may not fully meet the ideal Gaussian distribution assumption, k b By adjusting the protection level, it can be more consistent with the measurement deviation of the actual system. n It is the adjustment coefficient for noise, reflecting the impact of noise on the protection level under different conditions, and is used to correct the state estimation error calculation caused by observation noise. D is a multiplier based on the probability of missed alarm P MD The calculated standard deviation multiplier represents the effect of observation noise on the state estimation error. ∈ By k n and k D Combined to correct the impact of noise on the protection level, k ∈ =k n k D Combined with the noise adjustment factor k n and the standard deviation multiplier k in statistics D .

[0108] In one embodiment, in an actual system, navigation data may be affected by a variety of unpredictable factors (such as environmental changes, noise, deviation, etc.). In order to make the calculation of the protection level more in line with the performance of the actual system, the coefficient k can be continuously adjusted based on the actual operating data of the navigation system. b and k n Optimization is performed to ensure that the calculation results provide accurate protection level estimates in all situations.

[0109] In one embodiment, the machine learning method can automatically identify the error patterns in the system through training with a large amount of historical navigation data, and automatically optimize the adjustment coefficient k according to these error patterns. b and k n For example, supervised learning methods (such as linear regression or neural networks) are used to predict the optimal adjustment coefficient k b and k n value, thereby improving the accuracy of the protection level.

[0110] In an exemplary embodiment, in order to improve the accuracy and smoothness of the navigation solution, after the system is initialized, the least squares algorithm is usually used to perform a preliminary solution of the position and attitude. The least squares method is usually used to provide the system with a fast solution result in the early stage. However, in order to improve the accuracy and smoothness of the navigation solution, after the system is initialized, it is usually switched to the Kalman filter algorithm. The Kalman filter can continuously predict and update the system state. It not only combines the current measurement values, but also comprehensively considers historical data and the system dynamic model. Therefore, compared with the least squares method, the Kalman filter is more suitable for long-term navigation applications, especially in terms of error correction and data smoothness. For the Kalman filter, the state estimate after each GNSS measurement update is equivalent to the least squares considering the state prediction. Based on the state prediction, the virtual observation value and the virtual observation value Z′ of the measurement matrix (unit matrix) can be constructed pred , as shown in formula (20):

[0111]

[0112] The covariance matrix P of the virtual observations based on formula (20) k|k-1 It is generally a non-diagonal matrix, that is, the virtual observation value Z′ pred Generally, they are correlated, which violates the assumption that the observations in the integrity algorithm are uncorrelated. In order to overcome this problem, a decorrelated virtual observation z is introduced. pred , as shown in formula (21):

[0113]

[0114] in, is the state prediction result at the current time k, which is a recursion based on the state estimation result at the previous time k-1. -1 The inverse matrix of the lower triangular matrix obtained by decomposing the covariance matrix P is represented by the covariance. x is the actual state of the system, which is usually unknown. ∈ represents the prediction error, which describes the error between the predicted state and the actual state. Formula (21) shows that since x is unobservable, the estimated value can be used in the calculation replace.

[0115] In one embodiment, the covariance matrix P may be decomposed in a manner including, but not limited to, Kholesky decomposition, upper triangular matrix and lower triangular matrix (LU) decomposition, orthogonal proof and lower triangular matrix (QR) decomposition, etc.

[0116] Formula (21) describes the relationship between state prediction and observation in Kalman filtering. State prediction constructs a virtual observation z pred , virtual observation value z pred It is constructed by Kalman state prediction, where the virtual observation value z pred is the decorrelated Kalman filter state prediction, which represents the estimate of the state vector at the current moment, derived based on the state prediction at the previous moment. In formula (21), It is the current state predicted based on the previous moment, I represents the unit matrix (i.e., the observation matrix of the system), and ∈ represents the prediction error after decorrelation.

[0117] Virtual observation value z pred The introduction of makes the dimension of the system's observation vector become N = N sat +m dimension, where N sat is the number of satellites, i.e., the number of real observations, and m represents the position estimate in three-dimensional space, i.e., the number of virtual observations. In one embodiment, m=3. In one embodiment, the m virtual observations can be obtained by the state prediction result in the Kalman filter. Construct, corresponding to the m state variables of the receiver (such as the x, y, z position or the coordinates of the east, north, and sky directions).

[0118] In one exemplary embodiment, step 102 may include:

[0119] The protection level is calculated based on the characteristic slope of the observation value j for a specific direction (such as the i-direction), the detection statistic caused by the measurement deviation, and the covariance of the state estimate; wherein the observation value j includes at least one of a real observation value and a virtual observation value. In one embodiment, the protection level is calculated as shown in formula (22).

[0120]

[0121] In formula (22), SLOPE i,j represents the characteristic slope of observation value j in the i-direction; WSSE represents the detection statistic caused by measurement bias; P i,i Represents the diagonal elements of the covariance matrix P of the state estimation result in the i-direction; N sat represents the number of real observations, m represents the number of virtual observations; k b and k ∈ is the adjustment coefficient, k b Adjustment factor for correcting the protection level caused by measurement deviation, k ∈ For noise adjustment, k ∈ =k n k D , k D is based on the probability of missed alarm P MD The calculated standard deviation multiplier, k n is the adjustment factor for noise.

[0122] In an exemplary embodiment, different navigation modes may be distinguished according to the alignment of an inertial navigation system (INS, referred to as INS in the embodiment of the present application).

[0123] In one embodiment, if the inertial navigation system has not been aligned, only the GNSS signal is used in this case, and after the GNSS observation data is subjected to Kalman filtering, the protection level calculation based on the prediction is performed on the GNSS observation data. At this time, the navigation mode of the receiver is a single GNSS navigation mode. In one embodiment, for the case of a single GNSS, the decorrelated virtual observation value z pred It can be constructed by Kalman state prediction, that is, in the case of single GNSS navigation, the predicted state Derived from the predicted state of the Kalman filter.

[0124] In one embodiment, if the inertial navigation system is aligned, the combination of GNSS and inertial navigation is used. After performing a tight combination Kalman filter on the GNSS and inertial navigation system, the protection level of the combined data of the GNSS and inertial navigation system is calculated. The navigation mode of the receiver at this time is the integrated navigation mode. In one embodiment, for the integrated navigation system, the decorrelated virtual observation value z pred It can be obtained by inertial recursion, that is, in the case of combined navigation of GNSS and inertial navigation, the predicted state It is derived from the INS inertial recursion result, which is based on the predicted state of the GNSS Kalman filter.

[0125] In an exemplary embodiment, if the satellite signal is lost (e.g., in the case of full satellite obstruction), in this case, it is possible to switch to using INS, and step 102 may also include: recursively calculating the protection level through INS according to the protection level before the GNSS signal is interrupted. In one embodiment, the protection level may be recursively calculated through INS according to the protection level of the last combined epoch before the satellite signal is interrupted. At this time, the navigation mode of the receiver is the single inertial navigation (abbreviated as single inertial navigation) mode. In one embodiment, for the single inertial navigation mode, the decorrelated virtual observation value z pred It can be obtained by inertial recursion, that is, in the single inertial navigation mode, the predicted state The inertial recursion result comes from INS, and the recursion is based on the predicted state of GNSS Kalman filter.

[0126] In the embodiment of the present application, a least squares equivalent calculation method is adopted to integrate the least squares, GNSS Kalman filtering and protection level calculation method under combined navigation into one framework, and it is carried out based on the intermediate results of the navigation solution, thus avoiding unnecessary matrix calculation and storage.

[0127] In an exemplary embodiment, in the actual operation of the integrated navigation system, there may be a scenario where satellite navigation is interrupted, such as entering an underground garage or tunnel, where the satellite signal is completely blocked, etc. At this time, the satellite signal is interrupted, and the system completely relies on INS dead reckoning. The system performs dead reckoning based on the navigation settlement of the last combined epoch before the satellite signal is interrupted. The RAIM theory based on redundancy is not applicable. Therefore, for the case where the receiver is in the single inertial navigation mode, as shown in formula (23), the subsequent protection level can be recursively deduced based on this protection level as the basis:

[0128]

[0129] In formula (23), Indicates the protection level of the last combined epoch in the i-direction before the satellite signal is interrupted; P i,i k represents the diagonal elements of the covariance matrix P of the i-direction state estimation result; b and k ∈ is the adjustment coefficient, k b For correction The adjustment factor, k ∈ For noise adjustment, k ∈ =k n k D , k D is based on the probability of missed alarm P MD The calculated standard deviation multiplier, k n is the adjustment factor for noise.

[0130] Since there is no GNSS signal at this time and fault detection cannot be performed, in the embodiments of the present application, the protection level of the last combined epoch before the satellite signal interruption is inherited. In the single inertial navigation mode, the deviation protection level of the last combined navigation epoch before the GNSS signal interruption is used As the basis of inertial navigation, through the covariance matrix P and the adjustment coefficient k ∈ Describe the noise accumulation effect of inertial navigation and use k b Adjust the deviation to obtain a protection level that more conforms to the inertial navigation error model. The method for calculating the protection level in the single inertial navigation mode provided by the embodiments of the present application ensures that in the case of GNSS signal loss, a reliable protection level can still be obtained through inertial navigation recursion, maintaining the integrity assessment of the navigation system.

[0131] Standard deviation multiplier k D (Also known as the inflation coefficient k D ) is calculated as shown in formula (24):

[0132] k D = Q -1 (P(PE < PL|H0)) (24)

[0133] In formula (23), P(PE < PL|H0) represents the probability that the prediction error PE is less than the protection level PL under the assumption of the null hypothesis H0, that is, the fault-free condition. P(PE < PL|H0) is equal to 1 - P MD ; Q -1 The operator represents the inverse function of the standard normal distribution. The calculation method of the inflation coefficient k shown in formula (24) D is based on the conditional probability P(PE < PL|H0) that the prediction error is less than the protection level, and is calculated through the inverse function Q -1 of the standard normal distribution.

[0134] For the case of satellite signal interruption, in formula (24), P(PE < PL|H0) is as shown in formula (25):

[0135]

[0136] Formula (25) represents the calculation of the error probability under the protection level, which combines the probability distributions under the assumption of the null hypothesis H0 (no fault) and the alternative hypothesis H1 (fault exists). Among them, P(PE < PL) represents the probability that the total prediction error is less than the protection level PL; t is the single inertial recursion time, τ is the fault duration, and P(H0) and P(H1) are the probabilities in the fault-free and faulty cases respectively. P fault is the probability of system failure, which is used to measure the possibility of the navigation system failing.

[0137] In single inertial navigation mode, protection level PL i When the GNSS signal is interrupted, the bias protection level of the last combined epoch before the GNSS signal interruption can be inherited and continued recursively through INS.

[0138] In an exemplary embodiment, the method for implementing autonomous integrity monitoring of a receiver provided in an embodiment of the present application may further include:

[0139] The calculated protection level is compared with the preset threshold. When the protection level does not exceed the threshold, the navigation solution and protection level are output. When the protection level exceeds the threshold, an over-limit alarm is triggered.

[0140] In an exemplary embodiment, the calculated protection level is compared with a preset threshold to determine whether the positioning solution meets the safety requirements: if the protection level is less than the threshold, it is determined that the positioning solution meets the safety requirements, and the navigation solution results (receiver position, speed, etc.) and the protection level are output to complete the navigation solution process; if the protection level exceeds the threshold, it is determined that the positioning solution does not meet the safety requirements, and an over-limit alarm is issued to alert the system or user to the potential large positioning error.

[0141] The method for realizing autonomous integrity monitoring of the receiver provided in the embodiment of the present application combines real observation values ​​and virtual observation values, performs comprehensive analysis with the state estimation results of the receiver, and comprehensively considers measurement deviations and observation noise to calculate the protection level, thereby improving the accuracy and efficiency of receiver integrity monitoring and protection level assessment, realizing high-precision and high-security positioning solutions, and ensuring that reliable navigation decisions can be made in different environments.

[0142] The present application also provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute any of the above methods for implementing autonomous integrity monitoring of a receiver.

[0143] The present application further provides a computer device, including a memory and a processor, wherein the memory stores the following instructions executable by the processor: used to execute the steps of any of the above-mentioned methods for implementing autonomous integrity monitoring of a receiver.

[0144] In order to verify the superiority of the solution provided in the embodiment of the present application, a sports car experiment was carried out in City A. In the experiment, the vehicle carrying the navigation chip module drove for a total of 4.58 hours and a mileage of 123.30 km. The experimental track is as follows Figure 3 As shown, the driving trajectory (shown by the thick black line) includes open roads, urban canyons, tunnels, underground garages, etc.

[0145] The navigation module runs the solution provided by the embodiment of the present application, and the actual positioning error and protection level output are shown in Figures 4(a) and 4(b), where Figure 4(a) is east, Figure 4(b) is north, curves 41a and 41b are positioning errors, and curves 42a and 42b are protection levels. It can be seen from Figures 4(a) and 4(b) that the solution provided by the embodiment of the present application can form an envelope for the positioning error during the entire operation process, and this envelope is relatively tight.

[0146] In order to better examine the solution provided by the embodiment of the present application in an overall manner, the test results are plotted as Stanford diagrams, as shown in Figures 5(a) and 5(b). Figure 5(a) is an eastward Stanford diagram, and Figure 5(b) is a northward Stanford diagram, which are used to evaluate the performance of the protection level calculation method of the navigation system.

[0147] As shown in Figure 5(a) and Figure 5(b), the distribution of different blocks and points intuitively shows the performance of the system under different operating conditions, where point 51 represents the system in the integrated navigation state, and point 52 represents the system in the inertial dead reckoning state. Area 53 represents the situation where the system is unavailable or misleading information exists; area 54 represents the normal operation state, the system performs well and provides accurate positioning information; area 55 represents the dangerous operation state, there may be positioning accuracy problems or the protection level fails to meet the requirements; the transition areas between other areas indicate the availability and accuracy of the system under different conditions. In Figure 5(a), most points 51 fall in the white normal operation area, indicating that the system is in normal operation for 92.91% of the epochs in the east direction, and the system is unavailable for 7.08%. In Figure 5(b), 94.53% of the epochs are in normal operation, 5.42% are in unavailable state, and no dangerous operation occurs. As shown in Figure 5(a) and Figure 5(b), the system has very few points in the misleading information and dangerous operation areas, indicating that the solution provided by the embodiment of the present application accurately predicts the positioning error in most cases and ensures the overall availability of the system. In other words, it is very effective in predicting the horizontal positioning error and has a high navigation system availability. At the same time, the autonomous integrity monitoring method has a good global performance for the calculation of the system protection level, and the overall system availability is high. The specific performance indicators are shown in Table 1.

[0148]

[0149] Table 1

[0150] In general, the solution provided in the embodiment of the present application realizes the output of protection levels in various working conditions such as loose and tight combination, single inertial dead reckoning, and single satellite navigation on the navigation chip with relatively small memory and computing power overhead. The calculated protection level achieves a tight envelope for the positioning error, and the overall availability of the system reaches a level above 90%.

[0151] Figure 6 Schematic diagram of the structure of the device for implementing autonomous receiver integrity monitoring in an embodiment of the present application. Figure 6 As shown, it includes: an acquisition module, a processing module, and a calculation module; wherein,

[0152] An acquisition module is used to acquire the observation value of the receiver on the satellite signal and the state estimation result of the receiver; wherein the observation value includes a real observation value and a virtual observation value;

[0153] The processing module is used to obtain the state estimation error caused by the measurement deviation and the state estimation error caused by the observation noise according to the observation value and the state estimation result.

[0154] The calculation module is used to calculate the protection level of the receiver according to the obtained state estimation error caused by the measurement deviation and the state estimation error caused by the observation noise calculation, so as to monitor the autonomous integrity of the receiver.

[0155] In an exemplary embodiment, the acquisition module includes:

[0156] A detection module, used for performing gross error detection based on the observation values ​​of the satellite signal by the receiver;

[0157] An estimation module, used for estimating the state of the receiver according to the observed values ​​after gross error detection;

[0158] In one illustrative example, the estimation module may be used to:

[0159] The state of the receiver is estimated based on the observations after gross error detection to obtain state estimation results such as position and speed; the residual is calculated using the state estimation results and the actual observations; the variance of each observation is adaptively adjusted based on the calculated residual, and the state estimation is performed again after adjustment until the state estimation results and the variance of the observations converge to obtain a more accurate state estimation.

[0160] In an exemplary embodiment, it also includes: a comparison module, which is used to compare the calculated protection level with a preset threshold, output the navigation solution and the protection level when the protection level does not exceed the threshold, and trigger an over-limit alarm when the protection level exceeds the threshold.

[0161] In an exemplary embodiment, the processing module can be used to: obtain the state estimation error caused by the measurement deviation according to the characteristic slope of the observation value and the detection statistic caused by the measurement deviation; obtain the state estimation error caused by the observation noise according to the covariance of the state estimation result

[0162] In an exemplary embodiment, the calculation module may also be used to: when a satellite signal is interrupted, recursively calculate the protection level through inertial navigation according to the protection level before the satellite signal is interrupted.

[0163] In an exemplary embodiment, the virtual observation value is calculated according to formula (21). In one embodiment, when the navigation mode of the receiver is GNSS, the predicted state required for calculating the virtual observation value is is the predicted state of the Kalman filter. In one embodiment, the navigation mode of the receiver is a combination of GNSS and INS or a single inertial navigation (ie, INS), and the predicted state required to calculate the virtual observation value is It is the inertial recursion result of inertial navigation.

[0164] The device for implementing autonomous receiver integrity monitoring provided by the embodiment of the present application performs a comprehensive analysis by combining real observation values ​​and virtual observation values ​​with the state estimation results of the receiver, and comprehensively considers measurement deviations and observation noise to calculate the protection level, thereby improving the accuracy and efficiency of receiver integrity monitoring and protection level assessment, achieving high-precision and high-security positioning solutions, and ensuring that reliable navigation decisions can be made in different environments.

[0165] Although the embodiments disclosed in this application are as above, the contents described are only embodiments adopted to facilitate understanding of this application and are not intended to limit this application. Any technician in the field to which this application belongs can make any modifications and changes in the form and details of implementation without departing from the spirit and scope disclosed in this application, but the scope of patent protection of this application shall still be based on the scope defined in the attached claims.

Claims

1. A method for implementing autonomous integrity monitoring of a receiver, characterized in that: include: Obtaining observation values ​​of the receiver on the satellite signal and the state estimation result of the receiver; wherein the observation values ​​include real observation values ​​and virtual observation values; According to the observation value and the state estimation result, a state estimation error caused by a measurement deviation and a state estimation error caused by an observation noise are obtained; The protection level of the receiver is calculated according to the state estimation error caused by the measurement bias and the state estimation error caused by the observation noise to monitor the autonomous integrity of the receiver.

2. The method according to claim 1, wherein: The step of obtaining the state estimation error caused by the measurement deviation and the state estimation error caused by the observation noise according to the observation value and the state estimation result includes: Obtaining a state estimation error caused by the measurement deviation according to a characteristic slope of the observation value and a detection statistic caused by the measurement deviation; The state estimation error caused by the observation noise is obtained according to the covariance of the state estimation result.

3. The method according to claim 2, wherein: The protection level is calculated using the following formula: Among them, SLOPE i,j represents the characteristic slope of observation value j in the i-direction; WSSE represents the detection statistic caused by measurement bias; P i,i Represents the diagonal elements of the covariance matrix P of the state estimation result; k b and k ∈ is the adjustment factor; N sat represents the number of real observations, and m represents the number of virtual observations.

4. The method according to claim 1, further comprising: When the satellite signal is interrupted, the protection level is recursively calculated through inertial navigation according to the protection level before the satellite signal is interrupted.

5. The method according to claim 4, wherein: The protection level is calculated recursively by inertial navigation using the following formula: in, Indicates the protection level in the i-direction before the satellite signal is interrupted; P i,i Represents the diagonal elements of the covariance matrix P of the state estimation result; k b and k ∈ is the adjustment factor.

6. The method according to claim 3 or 5, wherein: k ∈ =k n k D , k n is the adjustment factor for noise; k D =Q -1 (P(PE <PL|H0)); Among them, P(PE < PL|H0) represents the probability that the prediction error PE is less than the protection level PL under fault-free conditions; Q -1 The operator represents the inverse function of the standard normal distribution.

7. The method according to claim 6, wherein: When the satellite signal is interrupted, the P(PE) is determined by the following formula: <PL|H0): P(PE <PL)=P(H0)P(PE<PL|H0)+P(H1) =[1-(1-P fault ) t / τ ]P(PE <PL|H0)+(1-P fault ) t / τ ; Among them, P(PE < PL) represents the probability that the total prediction error is less than the protection level PL; t is the single-inertia recurrence time, τ is the fault duration, P(H0) and P(H1) are the probabilities in the case of no fault and with fault respectively; P fault is the probability of system failure.

8. The method according to any one of claims 1 to 5, wherein: The virtual observation value z is calculated using the following formula: pred : in, is the state prediction result at the current time k, L -1 Represents the inverse matrix of the lower triangular matrix obtained by decomposing the covariance matrix P of the state estimation result.

9. A computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the method for implementing receiver autonomous integrity monitoring as described in any one of claims 1-8.

10. A computer device comprising a memory and a processor, wherein: The memory stores the following instructions executable by the processor: used to execute the steps of the method for implementing autonomous integrity monitoring of the receiver as described in any one of claims 1-8.

Images