Fault cutting-based robust method

Abstract

The invention discloses a robust method based on fault cutting, which is applied to a GNSS / SINS (Global Navigation Satellite System / Strapdown Inertial Navigation System) double navigation sensor integrated navigation system, and comprises the following steps: carrying out improvement and optimization on the basis of a residual chi-square test method and a fading sequential probability ratio detection method; fault types are recognized according to the two optimized and improved detection methods, processing factors are set according to the fault types, and cutting processing of the faults is achieved in the mode that the processing factors and robust processing factors are reset to alternately respond; the robust effect of a traditional robust mode is effectively improved.

Description

Technical Field

[0001] The invention relates to the technical field of fault cutting, in particular to a fault cutting-based robustness method. Background Art

[0002] In the process of integrated navigation system, timely fault detection and fault handling of measurement information has become a common way to ensure the stability of integrated navigation system. In terms of fault detection, the commonly used fault detection methods are: 2 Method and sequential probability ratio test (SPRT), where χ 2 The method includes the state χ 2 Method and residual χ 2 Method. Residual χ 2 The test is a commonly used detection method due to its simple calculation and sensitivity to sudden fault detection. However, it is slow to respond to small slow faults. 2 and dual state χ 2 It can produce a good detection effect on slowly changing faults, but as time goes by, the recursive device is easily contaminated by accumulated errors; SPRT method, as a residual-based inspection method, is independent of the state change statistics and has become a commonly used software fault detection method in integrated navigation systems. However, the traditional SPRT method has problems such as detection time delay, negative statistical value, and delay in the end of fault detection. Further improvement and optimization are needed. Some scholars have combined the residual χ 2 The fault end time of the test is reset to eliminate the time delay problem after the fault ends. In the actual integrated navigation process, a single moment fault will cause the combined positioning result to have a fault tailing phenomenon due to the mutual coupling relationship between system states. This method can only detect the residual χ 2 The fault size of the test is used as the fault end time, and the end time of the slowly varying fault in the tailing process cannot be detected. In addition, by introducing a gradual fading factor to enhance the influence of the new epoch on the test statistic change rate and statistics, the detection rate of the traditional SPRT algorithm is improved, and the defect of the traditional SPRT method that it cannot detect the end time of the slowly varying fault is improved.

[0003] When a system measurement fault is detected, the fault needs to be processed accordingly to ensure the stability of the system. Some scholars isolate the faulty sensor in the multi-navigation sensor system and do not use the faulty sensor measurement information in the federal filter, which has become a common method to ensure the robustness of the system in the multi-sensor combined navigation process. In the dual navigation sensor system, some scholars use the method of setting the faulty measurement information to zero to eliminate the influence of the faulty measurement information on the positioning results. However, with the long-term application of the fault, the inertial navigation state is not corrected in time by the measurement information, and the positioning divergence phenomenon will appear. Most scholars use the anti-error method to process the fault. By assigning a weight of 0-1 to the covariance of the fault, the weight of the fault measurement information in the state update is reduced to achieve the isolation effect. However, this method is only effective for single mutation faults, and will produce tailing and insignificant processing for continuous mutation faults or slow-changing faults, respectively. Summary of the invention

[0004] In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a robustness method based on fault cutting, which effectively improves the robustness effect of traditional robustness methods.

[0005] To achieve the above object, the technical solution adopted by the present invention is: a fault-cutting-based robustness method, applied to a GNSS / SINS dual navigation sensor combined navigation system, specifically comprising the following steps:

[0006] Step 1: Improve and optimize the residual chi-square test method and the gradual disappearance sequential probability ratio test method;

[0007] Step 2: Identify the fault type according to the two optimized and improved detection methods, set the processing factor according to the fault type, and implement fault cutting processing by resetting the processing factor and the robustness processing factor in an alternating manner.

[0008] As a further improvement of the present invention, the step 1 is specifically as follows:

[0009] The sequential idea is used to optimize the residual chi-square test. Assuming that only the latitude and longitude three-dimensional position information of the integrated navigation is observed, let ε i,k is the position measurement residual at the kth moment, where i = 1, 2, 3; the chi-square statistic λ corresponding to the residual of the i-th dimension at this time i,k for:

[0010]

[0011] Where P (i,i),k / k-1 P k,k-1 The diagonal elements of row i and column i, R (i,i),k For R k The diagonal elements of row i and column i;

[0012] In the gradually disappearing sequential probability ratio detection method, the covariance P of the measurement dimension corresponding to the residual statistical results exceeding the detection threshold in the process of calculating the statistic is calculated. (i,i)k,k-1 Replace with P at the previous moment (i,i)k-1 , used to prevent the calculation of P (i,i)k,k-1 As the fault continues to be applied, it continues to grow, thereby weakening the growth trend of the detection statistics; at the same time, in order to ensure that the statistics are not negative and affect the growth rate of the statistical results, the statistics are taken

[0013] As a further improvement of the present invention, in step 2, the fault type is identified according to the two optimized and improved detection methods as follows:

[0014] Calculate the current measurement residual ε k , and count the residuals at each moment to obtain the residual group vector Assume the residual chi-square test statistic is The corresponding residual chi-square test threshold is The test statistic of the progressive sequential probability ratio test The corresponding test threshold is Two tests are performed on the residual of the i-th dimension at the k-th moment in a sequential manner. The tests include:

[0015] 1) When and At this time, the inspection result is no fault;

[0016] 2) When and When , the test result is a slow-changing fault;

[0017] 3) When Must exist At this time, the test result is a sudden failure;

[0018] The detection situation corresponding to the i-th latitude at the k-th time is defined as FT i,k .

[0019] As a further improvement of the present invention, in step 2, the processing factor is set according to the fault type as follows:

[0020] By obtaining different detection situations FT i,k Update the processing factor. Different processing factors correspond to different processing methods. NT i,k is the robustness factor, NR i,k To reset the processing factor; a processing factor value of 1 indicates that this processing is performed, and a processing factor value of 0 indicates that this processing is not performed.

[0021] As a further improvement of the present invention, in step 2, the manner in which the resetting processing factor and the robustness processing factor alternately respond is specifically as follows:

[0022] The corresponding relationship between the three treatment factor combination modes and the treatment methods is as follows:

[0023]

[0024] Anti-error processing: Use the IGGⅢ method to obtain the anti-error weight factor t i,k , the chi-square statistic λ i,k R (i,i),k Replace R (i,i),k / t i,k , realize the anti-error processing of the current latitude;

[0025] Reset processing: First, the residual statistics of the previous moment The residual median of the previous k-1 moments is used to replace and reset, and the residual statistics of the previous moment are processed Prevent the influence of the fault residual in the previous step of robustness processing on the calculation of the residual median at the current moment; at the same time, the residual vector ε that needs to be reset is i,k Reset with replacement using the median residual.

[0026] The beneficial effects of the present invention are:

[0027] The present invention proposes a fault cutting robustness method for continuous mutation faults and slow-changing faults of measurement information in navigation systems. First, the sequential idea is introduced to improve the residual chi-square test to achieve accurate positioning of mutation faults, and the covariance and statistics of the gradually disappearing SPRT are optimized to improve the slow-changing fault detection sensitivity of the gradually disappearing SPRT method. The fault types are divided by the improved double test method, the processing factors of the fault types are updated, and the reset and robustness response states are allocated to achieve cutting of continuous faults. The simulation results show that the algorithm of the present invention performs fault cutting robustness processing on faults that reach the residual chi-square test threshold, and resets the faults that reach the SPRT threshold but are not satisfied with the residual chi-square test threshold. It effectively improves the robustness effect of traditional robustness methods and has certain engineering reference significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a schematic diagram of the changes in various parameters of IGGⅢ during a slow-changing fault;

[0029] Figure 2 This is a schematic diagram of the changes in various parameters of IGGⅢ during continuous mutation failure;

[0030] Figure 3 It is a flow chart of double inspection method detection in an embodiment of the present invention;

[0031] Figure 4 This is a flowchart of processing factor update in case 3 in an embodiment of the present invention;

[0032] Figure 5 A schematic diagram of a flight trajectory in an embodiment of the present invention;

[0033] Figure 6 is the GNSS position measurement error change in the embodiment of the present invention;

[0034] Figure 7 Schematic diagram of the change of combined navigation positioning error without robustness processing in an embodiment of the present invention;

[0035] Figure 8 It is a schematic diagram comparing the detection results of the traditional chi-square test and the chi-square test in the embodiment of the present invention;

[0036] Fig. 9 It is a schematic diagram of the detection result of the latitude-imposed slow-varying fault in an embodiment of the present invention;

[0037] Fig.10 This is a schematic diagram of latitude statistics results during fault cutting in an embodiment of the present invention;

[0038] Fig.11 Schematic diagram of longitude statistics during fault cutting in an embodiment of the present invention;

[0039] Fig.12 This is a global change diagram of the processing factors in the embodiment of the present invention;

[0040] Fig.13 Schematic diagram of the change of the robustness weight factor in an embodiment of the present invention;

[0041] Fig.14 2 is a diagram showing a change in positioning error in an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0043] Example

[0044] A fault-cutting-based robustness method specifically includes:

[0045] 1. SINS / GPS integrated navigation system:

[0046] This embodiment uses the standard Kalman filter as the data fusion algorithm based on the SINS / GPS integrated navigation system to carry out simulation and verification of the fault cutting algorithm.

[0047] 1.1, State equation:

[0048] The system state equation of SINS / GNSS is described as follows:

[0049] X k =F k,k-1 X k-1 +Γ k W k (1)

[0050] In formula (1): X k is the error state at time k, F k / k-1 is the state transition matrix from time k-1 to time k, X k-1 is the error state at the previous moment, Γ k1 is the noise driving array, W k is the system noise.

[0051] The state of the system consists of the misalignment angle, velocity error, position error, gyroscope error and accelerometer error of SINS. The error state of SINS is selected as 15 dimensions, as follows:

[0052]

[0053] In formula (2): Φ, δv n , δp represent the misalignment angle, velocity error, and position error of SINS in the navigation coordinate system (N system), ε b is the projection of the gyroscope zero bias in the carrier coordinate system (b system), ▽ b is the projection of the accelerometer zero bias in the carrier coordinate system (b system).

[0054] The state transfer matrix F k / k-1 It can be expressed as shown in formula (3):

[0055]

[0056] 1.2. Measurement equation:

[0057] In the GNSS / SINS integrated navigation mode, the positioning information output by SINS is corrected by high-precision GNSS positioning information to suppress the accumulation of inertial errors over time. The measurement equation of the system can be expressed as follows:

[0058] Z k =H k X k +V k (4)

[0059] In the formula, Z k is the measured value; H k is the measurement matrix; V k To measure noise; in this embodiment H k It can be expressed by formula (5).

[0060]

[0061] 1.3. Standard Kalman filter equation:

[0062] The standard Kalman filter algorithm mainly includes two parts: time update and measurement update. The time update includes the state and its covariance update, and the measurement update includes the filter gain, state estimation, and variance estimation:

[0063] X k / k-1 =F k X k-1 / k-1 (6)

[0064]

[0065]

[0066]

[0067] P k =(IK k H k ) k / k-1 (10)

[0068] 2. Introduction to fault detection and robustness methods:

[0069] 2.1. Fault model:

[0070] Faults can be divided into sudden faults and slow faults, and their fault models are as follows:

[0071]

[0072] In formula (11), is the fault observation value, z k is the true observed value, δz k is the mutation fault value, α z is the fault change rate, k is the current time, and k0 is the fault starting time.

[0073] 2.2. Residual chi-square test method:

[0074] The residual chi-square test is a sensitive test for sudden failures, with a simple structure and easy calculation. k , which is expressed as follows:

[0075]

[0076] In formula (12), Z k is the measured value, One-step state prediction.

[0077] According to the orthogonality principle, the square of the residual Mahalanobis distance obeys the chi-square distribution, and its chi-square test statistic is as follows:

[0078]

[0079] If the measured value is abnormal, the square of the residual Mahalanobis distance will not conform to the chi-square distribution with n degrees of freedom, where n is the measurement latitude. The residual ε can be determined by setting the chi-square test significance level α. k The credibility of The chi-square test statistic is obtained as follows:

[0080]

[0081] If k >T D , then the residual is judged to be abnormal, otherwise the opposite is true.

[0082] Since the residual chi-square test relies only on the state estimate from the previous step Information can effectively detect sudden faults. However, when faced with slowly changing faults, The statistical characteristics of the residuals of adjacent epochs change little. The residual chi-square test will have a detection delay phenomenon and is not sensitive to slowly varying faults.

[0083] 2.3. Gradually disappearing SPRT test method:

[0084] The traditional SPRT algorithm is a sensitive detection method for slowly varying faults and is widely used in the slowly varying fault detection process of integrated navigation. The following is a brief introduction to the algorithm.

[0085] A binary hypothesis is made for the measured values ​​of the integrated navigation system: normal state H0 and abnormal state H1. When the measured value is fault-free, ε k According to the Gaussian distribution with a mean of zero, the residual expectation and variance can be expressed as follows:

[0086]

[0087] When the measured value fails, ε k It does not obey a Gaussian distribution with a mean of zero. The corresponding residual expectation and variance can be expressed as follows:

[0088]

[0089] In formula (16), δv k It is fault noise.

[0090] The expectation of the residual in equations (15) and (16) can be expressed by the mean:

[0091]

[0092] The probability density of the residual sample under the two assumptions is:

[0093]

[0094]

[0095] According to the probability density function, the likelihood ratio of the two hypotheses can be calculated as:

[0096]

[0097] Taking the logarithm of formula (20), we can get the statistic of SPRT detection method:

[0098]

[0099] Then, by simplification, the iterative relationship of the statistic can be obtained as follows:

[0100] λ k =λ k-1 +Δλ k (twenty two)

[0101] In the known P f False alarm rate and P m Under the premise of the alarm rate, the threshold of the statistic can be set by the SPRT method proposed by Wald as follows:

[0102]

[0103] According to the relationship between statistics and thresholds, the fault and normal status of the system are detected:

[0104]

[0105] The above is the traditional SPRT algorithm. In order to solve the problem that the slow-varying fault detection cannot detect the end of the fault, the fading SPRT algorithm introduces a fading factor in equations (17) and (22) to reduce the influence of historical residuals and statistics, thereby achieving the purpose of detecting the end time of the fault. The changes are as follows:

[0106]

[0107] 2.4. IGGⅢ anti-difference method:

[0108] Commonly used equivalent weight functions include Huber method, Danish method and IGG III method. IGG III is also called the three-stage method. This method corresponds to the weight retention area, weight reduction area and elimination area in the weight function, which realizes the full use of effective information, the limited use of suspicious information and the exclusion of harmful information. It is a processing method suitable for processing measurement data.

[0109] When a fault is detected, the fault measurement information can be isolated by using the robustness method. System robustness is to use the robustness factor to amplify the observation noise variance R k , adjust the gain matrix K k The value of , thereby reducing the weight of observation information and weakening the optimal estimate of the state at the kth moment The influence of. Formula (8) can be rewritten as follows:

[0110]

[0111] In formula (26), T k is the robustness factor. n is the measurement dimension.

[0112] This embodiment uses the IGG III method to construct the robustness factor, and its function expression is:

[0113]

[0114] In formula (27): e k is the standardized residual; k0 and k1 are adjustment coefficients, which are constant values. k0 is usually 1.0 to 2.5; k1 is 3.5 to 8.0. k The calculation method is as follows:

[0115]

[0116] In formula (28), ε k is the residual at the kth moment, is the residual distance, which is calculated as follows:

[0117]

[0118] In formula (29): is a set of residual vectors at the kth moment, and med(·) is the median.

[0119] 3. Robust algorithm based on fault cutting:

[0120] 3.1、Problem description:

[0121] In the actual process, slowly varying noise and continuous mutation noise are difficult to be calculated by the noise covariance matrix R. k It can be effectively processed by the method of allocating robust weights. It can be assumed that the i-th dimension measurement information at time k contains a set of residual vectors of slowly varying faults: Assume that the fault starts at time k0. contains a set of normal residual vectors before the fault starts and a set of abnormal residual vectors after the fault starts Then the standardized residual is calculated according to formula (29):

[0122]

[0123] Substituting equation (11) into equation (30), the relationship between the standardized residual and the slowly varying fault is as follows:

[0124]

[0125] As time k→∞ gradually increases, The statistical results of are no longer maintained within a certain range. The residual data of the i-th dimension at the k-th time is obtained by formula (28): and Relationship:

[0126]

[0127] In order to more clearly observe the impact of slow-varying faults on the parameters in IGGⅢ, a set of residual vectors The length is 100, which is a normal distribution value with a mean of 0 and a variance of 1. A slow-changing fault 0.01*(k-50) is inserted at the 50th second of the simulation. The constant coefficients of IGGⅢ are k0=1 and k1=3.5. Formulas (31) and (32) are calculated to obtain the changes in the residual standard deviation, standard deviation distance, and weight factor. Compare the two cases of no fault and with fault, as shown in Figure 1 As shown, as time goes by, the residual value ε k As the residual standard deviation of the IGGⅢ algorithm increases, each The standard deviation distance from the current moment increases slowly, and the weight factor gradually tends to 0. In the integrated navigation, as the fault time continues, the weight factor t calculated by IGGⅢ k tends to 0, and from formula (26) we can get that when the weight factor t k →0 corresponding to the measurement noise R k →∞. From formula (8), we can get the corresponding K k →0, calculated by formula (9) As the fault time increases, the system only performs one-step state prediction, which leads to the inertial divergence of fault dimension measurement and the failure of the IGGⅢ weighted robustness method.

[0128] In order to more clearly observe the impact of continuous mutation failure on various parameters in IGGⅢ, The sudden fault value 5 is inserted from 50s to 100s in the simulation and continues until the end of the simulation. Figure 2 As shown in the figure, when the sudden fault is added, the residual value ε kThe parameters in the IGGⅢ algorithm, the residual standard deviation, suddenly increased and then slowly decreased. The standard deviation distance from the current moment increases rapidly, and the weight factor jumps directly to 0 first, and then gradually approaches the normal factor after a period of time. k →0During this period The subsequent weight factor changes to the normal range. During the combined navigation anti-error process, the dimension where the fault occurs will show a tailing phenomenon of first diverging and then adjusting, and rapid isolation cannot be achieved.

[0129] 3.2. Double-check algorithm optimization:

[0130] In the measurement vector of integrated navigation, if there is a fault in one-dimensional measurement information, due to the correlation between the residuals of the observation vector, it will cause the residuals of several other observations to be affected, thus causing abnormal phenomena in the overall positioning result of the integrated navigation. The fading SPRT can accurately detect the location of the fault, and the residual chi-square test result is a numerical value, which cannot locate the specific location of the fault, so the residual chi-square test needs to be optimized.

[0131] This embodiment proposes to optimize the residual chi-square test by using a sequential idea. Assume that only the latitude, longitude and altitude three-dimensional position information of the combined navigation is observed. Assume ε i,k is the position measurement residual at the kth moment, where i = 1, 2, 3. The chi-square statistic λ corresponding to the residual of the i-th dimension at this time i,k for:

[0132]

[0133] In formula (33), P (i,i),k / k-1 P k,k-1 The diagonal elements of row i and column i, R (i,i),k For R k The diagonal elements in the i-th row and i-th column.

[0134] In the fading SPRT method, the covariance P of the measurement dimension corresponding to the residual statistical result exceeding the detection threshold in the process of calculating the statistic is calculated. (i,i)k,k-1 Replace with P at the previous moment (i,i)k-1 , to prevent the calculation of P (i,i)k,k-1 As the fault continues to be applied, it continues to grow, thus weakening the growth trend of the detection statistics

[18] At the same time, in order to ensure that the statistic is not negative and affect the growth rate of the statistical results, the statistic is taken

[0135] 3.3. Fault cutting steps:

[0136] Fault cutting is mainly divided into three steps: fault type inspection, processing factor update, and robustness and reset processing. Through the different processing methods of the previous moment and the current moment, the slow-changing faults that fail to reach the residual chi-square test are reset, and the continuous mutation faults and slow-changing faults that reach the residual chi-square test threshold are reset and robustness interval processing is implemented to achieve the effect of cutting continuous mutation faults and slow-changing faults. The main steps are as follows:

[0137] Step 1 Fault type inspection:

[0138] Calculate the current measurement residual ε k , and count the residuals at each moment to obtain the residual group vector Assume the residual chi-square test statistic is The corresponding residual chi-square test threshold is The vanishing SPRT test statistic The corresponding test threshold is Two tests are performed on the residual of the i-th dimension at the k-th moment in a sequential manner, resulting in three test situations:

[0139] 1) When and At this time, the inspection result is no fault.

[0140] 2) When and At this time, the test result is a slow-changing fault.

[0141] 3) When Must exist At this time, the test result is a sudden failure.

[0142] The test situation corresponding to the i-th latitude at the k-th time is defined as FT i,k ,To sum up, the inspection process of the double inspection method is as follows Figure 3 shown.

[0143] Step 2: Processing factor update:

[0144] Obtain different detection situations FT through Step 1 i,k Update the processing factor. Different processing factors correspond to different processing methods. NT i,k is the robustness factor, NR i,k is the reset processing factor. The processing factor value of 1 indicates that this processing is performed, and the processing factor value of 0 indicates that this processing is not performed. Different detection situations correspond to different processing methods and processes. When k = 0, the initial state of the processing factor corresponding to the fault type is shown in formula (34). When k is not 0, the processing factor update process for case 1 and case 2 is shown in Table 1, and the processing factor update process for case 3 is shown in Table 2. Figure 4 shown.

[0145]

[0146] Table 1 Update table of treatment factors for cases 1 and 2

[0147]

[0148]

[0149] Step 3 Anti-error and reset processing:

[0150] After Step 2, the corresponding relationship between the three processing factor combination modes and the processing methods is shown in formula (35).

[0151]

[0152] 1. Anti-difference processing:

[0153] Using IGGⅢ method to obtain the robustness weight factor t i,k , replace R in formula (33) (i,i),k Replace R (i,i),k / t i,k , to achieve anti-error processing at the current latitude.

[0154] 2. Reset processing:

[0155] First, the residual statistics of the previous moment The residual median of the previous k-1 moments is used to replace and reset, and the residual statistics of the previous moment are processed Prevent the influence of the fault residual in the previous step of anti-error processing on the calculation of the current residual median. At the same time, the residual vector ε that needs to be reset is i,k Reset with replacement using the median residual.

[0156] 4. Simulation and analysis:

[0157] 4.1. Simulation parameter settings:

[0158] This embodiment verifies the effectiveness of the proposed algorithm by simulating a flight training subject trajectory data. Assume that the initial position of the aircraft is N: 30°58′39″N, E: 104°16′54″E, the altitude is H: 0m, the initial speed is 0kt, and the initial heading is 0°; the initial position errors of the three directions of northeast and celestial are 1m, 1m, and 3m, the initial misalignment angle is 0.5°, and the initial speed error is 0.1m / s. The parameters of the inertial navigation are set as: gyro zero bias stability eb=15° / h; gyro random walk web=500° / sqrt(h); accelerometer zero bias stability db=1ug; accelerometer measurement random walk web=500ug / sqrt(Hz). The GNSS positioning position error is set as: east error 1m, north error 1m, celestial error 3m; the simulation trajectory includes the aircraft's takeoff, climb, acceleration, turning and other motion states such as Figure 5 shown.

[0159] This embodiment verifies the effectiveness of the proposed algorithm by applying continuous mutation faults and slow-changing faults to the latitude and longitude information measured by GNSS. Applying the faults shown in Table 2 to normal measurement information, the measurement error changes corresponding to each measurement dimension with and without faults are obtained as follows: Figure 6 As shown in the figure, the false alarm rate and missed alarm rate of the residual chi-square test and the SPRT method are both 5%, and the corresponding chi-square threshold is 12.84 and the SPRT threshold is 5.29.

[0160] Table 2 Measurement information fault settings

[0161]

[0162] 4.2. Improved detection algorithm verification analysis:

[0163] By applying corresponding faults to the GNSS position measurement information and using KF as the data fusion algorithm, the positioning error change of the integrated navigation without robustness processing is obtained as follows: Figure 7 It can be seen that during the fault application process, the navigation positioning result without robustness processing is significantly affected by the fault. Figure 8 In order to compare the results of the residual chi-square test with the traditional chi-square test under the influence of this fault, the results of the sequential residual chi-square test can clearly distinguish the dimensional position of the fault measurement information, and the statistics are highly consistent with the traditional residual chi-square statistical results. Fig. 9The changes in the detection results of the improved gradually-changing SPRT method, the gradually-changing SPRT method and the improved residual chi-square test when only slowly-changing faults are added in latitude. In the figure, after the slowly-changing faults are added at the 400th second, the critical moment for the improved gradually-changing SPRT method to reach the SPRT threshold is 404s, and the critical moment for the gradually-changing SPRT method to reach the SPRT threshold is 408s. The critical moment for the improved chi-square test to reach the chi-square threshold is 419s. This shows that the improved gradually-changing SPRT method has better slow-changing fault detection sensitivity than the gradually-changing SPRT method. At the same time, the critical moment of the SPRT method is less than the critical moment of the chi-square test. The SPRT method has a better detection effect on slowly-changing faults than the residual chi-square test. And combined with Fig. 9 As shown in the figure, the improved fading SPRT method can successfully detect the end time of the fault, which is consistent with the fault changes in the integrated navigation process.

[0164] 4.3. Analysis on the effectiveness of treatment factors:

[0165] The improved fading SPRT method and the improved chi-square test method are used as the detection algorithm of this embodiment. Fig.10 and Fig.11 This is the test result of applying a continuous mutation fault to the longitude and a slow-changing fault to the latitude in the fault cutting robustness process of this embodiment. Fig.10 It can be seen that the start time of the improved gradually fading SPRT method and the improved chi-square test method are 408s and 412s respectively, while the occurrence time of the slow-changing fault is 400s. Fig.12 The change of the processing factor in the mid-latitude is obtained. At the 408th to 451st second of the simulation, the NR and the anti-difference NT are reset to alternately respond to realize the cutting processing of the slowly varying fault and then perform anti-difference. Fig.13 The alternating changes in the weight factors of the corresponding time periods also verify the effectiveness of the cutting process. At the same time, the fault end time detected by the improved chi-square test method is 451s, and the fault end time detected by the improved fading SPRT method is 480s. The latitude measurement in the simulation of the 451s to 480s also contains certain slow-changing faults. Fig.12 Changes in the processing factors at mid-latitudes. At this time, only NR responds and continues to reset the measurements during this period. Fig.11 This is the detection result of the measurement information being subjected to continuous mutation noise from the 100th to 120th second of the simulation. It can be seen that the two detection algorithms detect the same fault start time and the fault end time distribution is 121s and 124s. In this time period, Fig.12 The processing factor at the middle longitude only responds to NR, which means that the algorithm is resetting the slowly varying fault at this time, proving that the fault has been cut and the change in the processing factor is consistent with the theoretical change result.

[0166] 4.4. Analysis of combined navigation results:

[0167] In this embodiment, the normal KF integrated navigation result without any fault is taken as method 1, the result of only IGGⅢ anti-error processing is taken as method 2, and the IGGⅢ anti-error result after fault cutting processing is taken as method 3. The integrated navigation positioning error result is as follows: Fig.14 As shown. It can be seen that method 2 only performs IGGⅢ anti-error processing, which can effectively isolate the sudden fault in the height error, and has a certain processing effect on the slowly varying fault and continuous sudden fault in the latitude and longitude, making the existing fault much smaller than the imposed fault, but it cannot achieve complete anti-error processing for the slowly varying fault and the continuous sudden fault. Method 3 is the combined navigation result of the fault cutting anti-error method proposed in this embodiment. The positioning result of this method is consistent with the positioning result of method 1 in the case of no fault, which proves that the processing algorithm of this embodiment is effective for the anti-error results of slowly varying faults and continuous sudden faults. In order to more clearly show the processing effect of the algorithm, this embodiment compares the RMSE values ​​of the three directions of latitude, longitude and altitude corresponding to several methods by calculating them, as shown in Table 3. It can be seen that the RMSE value of the algorithm of this embodiment is slightly smaller than the RMSE value of method 1. This is because the algorithm of this embodiment detects and resets the randomly occurring slowly varying noise in the navigation process during the combined navigation process, so that the positioning result of the algorithm of this embodiment is slightly better than that of method 1. Compared with method 1, the latitude and longitude high positioning accuracy increased by 5.03%, 12.90%, and 19.94%, respectively, and compared with method 2, the distribution increased by 45.01%, 13.53%, and 1.09%. The height dimension positioning accuracy of method 2 and method 3 is not much different because the height dimension only adds a single-moment mutation fault, and the IGGⅢ anti-error method can effectively remove a single mutation fault.

[0168] Table 3. RMSE of the algorithm error

[0169]

[0170] In order to solve the problem that the traditional robustness method has poor effect on the processing of continuous mutation faults and slowly changing faults, this embodiment proposes a robustness method for fault cutting. First, based on the traditional residual chi-square test method and the gradually disappearing sequential probability ratio test (SPRT), improvements and optimizations are made to improve the accuracy of fault location and the sensitivity of fault detection. According to the two improved detection methods, the fault type is identified, the processing factor is set according to the fault type, and the fault cutting processing is achieved by resetting the processing factor and the robustness processing factor in an alternating manner. In the reset processing, the median of the residual statistics is used to reset the fault residual at the current moment, and the IGGⅢ method is used in the robustness processing to perform robustness processing on the fault after cutting. The simulation results show that the algorithm can effectively handle the impact of continuous mutation faults and slowly changing faults on combined navigation, and has a significant improvement effect on the positioning accuracy compared to the traditional IGGⅢ robustness method, and to a certain extent improves the navigation positioning accuracy in the fault-free situation.

[0171] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A robust method based on fault cutting, characterized in that: Applied to the GNSS / SINS dual navigation sensor integrated navigation system, specifically including the following steps: Step 1: Improve and optimize the residual chi-square test method and the gradual disappearance sequential probability ratio test method; Step 2: Identify the fault type according to the two optimized and improved detection methods, set the processing factor according to the fault type, and implement fault cutting processing by resetting the processing factor and the robustness processing factor in an alternating manner.

2. The robustness method based on fault cutting according to claim 1, characterized in that: The step 1 is specifically as follows: The sequential idea is used to optimize the residual chi-square test. Assuming that only the latitude and longitude three-dimensional position information of the integrated navigation is observed, let ε i,k is the position measurement residual at the kth moment, where i = 1, 2, 3; the chi-square statistic λ corresponding to the residual of the i-th dimension at this time i,k for: Where P (i,i),k / k-1 P k,k-1 The diagonal elements of row i and column i, R (i,i),k For R k The diagonal elements of row i and column i; In the gradually disappearing sequential probability ratio detection method, the covariance P of the measurement dimension corresponding to the residual statistical results exceeding the detection threshold in the process of calculating the statistic is calculated. (i,i)k,k-1 Replace with P at the previous moment (i,i)k-1 , used to prevent the calculation of P (i,i)k,k-1 As the fault continues to be applied, it continues to grow, thereby weakening the growth trend of the detection statistics; at the same time, in order to ensure that the statistics are not negative and affect the growth rate of the statistical results, the statistics are taken 3. The robustness method based on fault cutting according to claim 2, characterized in that: In step 2, the fault type is identified according to the two optimized and improved detection methods as follows: Calculate the current measurement residual ε k , and count the residuals at each moment to obtain the residual group vector Assume the residual chi-square test statistic is The corresponding residual chi-square test threshold is The test statistic of the sequential probability ratio test The corresponding test threshold is Two tests are performed on the residual of the i-th dimension at the k-th moment in a sequential manner. The tests include: 1) When and At this time, the inspection result is no fault; 2) When and When , the test result is a slow-changing fault; 3) When Must exist At this time, the test result is a sudden failure; The detection situation corresponding to the i-th latitude at the k-th time is defined as FT i,k .

4. The robustness method based on fault cutting according to claim 3, characterized in that: In step 2, the processing factor is set according to the fault type as follows: By obtaining different detection situations FT i,k Update the processing factor. Different processing factors correspond to different processing methods. NT i,k is the robustness factor, NR i,k To reset the processing factor; a processing factor value of 1 indicates that this processing is performed, and a processing factor value of 0 indicates that this processing is not performed.

5. The robustness method based on fault cutting according to claim 4, characterized in that: In step 2, the reset processing factor and the robustness processing factor are alternately responded to in the following manner: The corresponding relationship between the three treatment factor combination modes and the treatment methods is as follows: Anti-error processing: Use the IGGⅢ method to obtain the anti-error weight factor t i,k , the chi-square statistic λ i,k R (i,i),k Replace R (i,i),k / t i,k , realize the anti-error processing of the current latitude; Reset processing: First, the residual statistics of the previous moment The residual median of the previous k-1 moments is used to replace and reset, and the residual statistics of the previous moment are processed Prevent the influence of the fault residual in the previous step of anti-error processing on the calculation of the residual median at the current moment; at the same time, the residual vector ε that needs to be reset is i,k Reset with replacement using the median residual.

Images