GPS module / star sensor attitude accuracy evaluation method based on hybrid chi-square test

Abstract

The present invention discloses a method for evaluating the attitude accuracy of a GPS module / star sensor based on a hybrid chi-square detection, which specifically includes the following steps: (1) obtaining the attitude information of the GPS module and the star sensor; (2) respectively evaluating the accuracy of the attitude information of the GPS module and the star sensor by using the innovation chi-square detection, obtaining the corresponding measurement innovation test statistic, and judging whether the output data is abnormal; (3) evaluating the accuracy of the attitude combination information of the GPS module and the star sensor by using the state chi-square detection, obtaining the combined state test statistic, and judging whether the combined output data is abnormal; (4) when there is an abnormal situation in the output data in step (1) or step (2), constructing an equivalent weight factor and an equivalent covariance matrix; (5) constructing a filter gain equation and a state estimation equation, and using the Kalman filtering method to solve and obtain the final attitude result. The present invention realizes error compensation in the case of failure of the GPS module or the star sensor, and improves the robust performance of the filtering.

Description

Technical Field

[0001] The invention relates to the technical field of satellite / celestial navigation, and in particular to a GPS module / star sensor attitude accuracy assessment method based on hybrid chi-square detection. Background Art

[0002] With the rapid development of navigation, positioning and attitude determination and related fields, many different types of navigation technologies such as satellite navigation and astronomical navigation have been applied to various fields such as cities, oceans, and aviation. At present, the satellite navigation technology system based on GPS modules and Beidou has formed a dynamic and comprehensive global three-dimensional navigation, positioning and attitude determination service system, which can quickly obtain the precise position of land, sea and air targets. Position determination and attitude determination are usually two independent processes, and their correlation is often ignored in the combined navigation process. Due to the limitations of a single navigation system, two or more navigation technologies can be combined to provide navigation information, and the combination constitutes a complete integrated navigation system with more guaranteed accuracy and stability.

[0003] Under the existing technical background, the combined navigation system combining satellite navigation and inertial navigation system is the most common and widely used combined navigation scheme. However, there are two problems with this type of method. First, the measurement error of the inertial measurement device will accumulate over time until it seriously affects the system accuracy. Although it can be corrected by relevant technical means, the corrected speed cannot accurately match the flight speed of the target. Therefore, as the flight distance becomes longer or the positioning service time becomes longer, the accuracy of the instantaneous attitude calculation becomes lower, so it will be severely restricted in the process of medium and long-distance navigation. Secondly, the inertial measurement device is a high-loss device with quite high manufacturing and maintenance costs. The higher the measurement accuracy requirements for large-scale operations, the higher the cost and the lower the cost performance. Therefore, it is of great significance to study new models, methods and technologies for multi-level and multi-platform positioning and attitude determination to improve the real-time measurement and control capabilities of aerial targets and reduce manufacturing and maintenance costs. Star sensor is a high-precision attitude measurement instrument that has been increasingly recognized and applied in the design and operation of various aerospace vehicles. It not only has significant advantages such as small size, high precision and low power consumption, but also its measurement errors will not accumulate over time, which is helpful for improving the positioning and attitude accuracy in long-distance navigation scenarios. Therefore, it can make up for and improve the errors of other navigation systems in this regard. Summary of the invention

[0004] Purpose of the invention: In view of the above shortcomings, the present invention discloses a GPS module / star sensor attitude accuracy evaluation method based on hybrid chi-square detection, which can evaluate the attitude accuracy of the GPS module and the star sensor, thereby realizing error compensation in the case of relative failure of either the GPS module or the star sensor, and improving the anti-error performance of the filtering.

[0005] Technical solution: To solve the above problems, the present invention provides a method for evaluating the attitude accuracy of a GPS module / star sensor based on hybrid chi-square detection, comprising the following steps:

[0006] (1) Obtain attitude information from the GPS module and star sensor installed in the aircraft;

[0007] (2) Using new information χ 2 The detection evaluates the accuracy of the attitude information of the GPS module and the star sensor respectively, obtains the corresponding measurement innovation test statistics respectively, and judges whether there is any abnormality in the output data of the GPS module and the star sensor through the obtained measurement innovation test statistics;

[0008] (3) Using state χ 2 Detect and evaluate the accuracy of the combined attitude information of the GPS module and the star sensor, obtain the combined state test statistics, and judge whether there is any abnormality in the combined output data of the GPS module and the star sensor through the obtained combined state test statistics;

[0009] (4) When the output data in step (2) or step (3) is abnormal, an equivalent weight factor is constructed; specifically:

[0010]

[0011] ω=diag[ω 1 ,ω 2 ,…,ω n ]

[0012] In the formula, ω is the equivalent weight factor, ω i is the equivalent weight factor at each moment, i=1,2……n, n is a positive integer; T i,j (k) is the test statistic of the j-th chi-square test at time i, j = 1, 2, 3; when j = 1, the corresponding new information χ 2 Detect and obtain the GPS module measurement new information test statistic T i,1 (k); When j = 2, the corresponding new information χ 2 Detection obtains the star sensor measurement innovation test statistic T i,2 (k); when j = 3, the corresponding state is χ 2 Detection obtains the combined state test statistic T i,3 (k); T D is the boundary value in the boundary condition; k is the state;

[0013] Then, the equivalent covariance matrix is ​​obtained according to the equivalent weight factor, which is:

[0014]

[0015] In the formula, is the equivalent covariance matrix; V 1 (k) is the covariance matrix of the GPS module observation noise;

[0016] (5) Based on the equivalent covariance matrix, the filter gain equation and state estimation equation are constructed, and the Kalman filter method is used to solve the final attitude result.

[0017] Furthermore, step (2) specifically includes the following steps:

[0018] (2.1) Using the new information χ 2 The accuracy of the attitude information of the GPS module and star sensor is evaluated respectively, and the formula for obtaining the measurement information is:

[0019]

[0020] In the formula, v 1 (k) is the GPS module measurement information; v 2 (k) is the new information measured by the star sensor; 1 (k) is the GPS module observation vector matrix; z 2 (k) is the star sensor observation vector matrix; Prediction value of observation vector for GPS module; is the predicted value of the star sensor observation vector; H 1 is the GPS module observation matrix; H 2 is the star sensor observation matrix; k is the state; k+1 refers to the state after the k state, and k-1 refers to the state before the k state;

[0021] (2.2) According to the measurement innovation, the measurement innovation test statistic is obtained, and the formula is:

[0022] T(k)=v(k) T C(k) -1 v(k)

[0023] Where C(k) is the covariance matrix of the measurement information; substitute the corresponding v of the GPS module at different time i into 1 (k), C 1 (k) Obtain GPS module measurement innovation test statistic T i,1 (k); Substitute the corresponding v of the star sensor at different times i 2 (k), C 2 (k) Obtain the star sensor measurement innovation test statistic T i,2 (k); in which:

[0024] C 1 (k) = H 1P 1 (k|k-1)H 1 T +V 1 (k)

[0025] C 2 (k) = H 2 P 2 (k|k-1)H 2 T +V 2 (k)

[0026] Where P 1 (k|k-1) is the GPS module state covariance matrix, P 2 (k|k-1) is the star sensor state covariance matrix; V 1 (k) is the covariance matrix of the GPS module observation noise; V 2 (k) is the covariance matrix of the star sensor observation noise;

[0027] (2.3) Setting boundary conditions for abnormal observations;

[0028] T D ~x α 2 (t)

[0029] In the formula, α is the significance level; t is the degree of freedom;

[0030] By comparing the two measurement innovation test statistics with the boundary value, it is determined whether the output data is abnormal. If the measurement innovation statistic is less than or equal to the boundary value, it is determined that the output data is normal, otherwise it is determined that there is an abnormality.

[0031] Furthermore, step (3) specifically includes the following steps:

[0032] (3.1) The state quantity is set based on the attitude information of the star sensor. The formula is:

[0033]

[0034] In the formula, x is the state vector matrix; Φ represents the attitude angle of the star sensor; Θ represents the angular velocity of the star sensor; Ψ represents the angular acceleration of the star sensor; the state equation is established as:

[0035] x(k+1)=Fx(k)+w(k)

[0036] Where, F is the state transition matrix; w is the system noise;

[0037] (3.2) The observation vector is set based on the attitude information of the GPS module. The formula is:

[0038]

[0039] Where z is the observation vector matrix; represents the attitude angle of the GPS module; θ represents the angular velocity of the GPS module; ψ represents the angular acceleration of the GPS module; the observation equation is established, the formula is:

[0040] z(k)=H 1 x(k)+V 1 (k)

[0041] (3.3) Using state X 2 Detect and estimate the accuracy of the attitude information of the GPS module and star sensor to obtain the first state estimate The second state estimator And make a difference, and finally get the combined state test statistic T 3 (k) specifically:

[0042]

[0043] T 3 (k) = β(k)R(k) -1 β(k)

[0044] Where K′ is the original filter gain; R(k)=E{β(k)β T (k)}; T 3 (k) is the combined state test statistic; T i,3 (k) is T at different time i 3 (k) value;

[0045] (3.4) By comparing the combined state test statistic with the boundary value, it is determined whether the output data is abnormal. If the combined state test statistic is less than or equal to the boundary value, it is determined that the output data is normal, otherwise it is determined that there is an abnormality.

[0046] Furthermore, the filter gain equation constructed in step (5) is:

[0047]

[0048] Where K is the filter gain; P 1 (k|k-1) is the GPS module state covariance matrix; H 1 It is the GPS module observation matrix;

[0049] The state estimation equation is:

[0050]

[0051] In the formula, represents the state estimator; Represents the observation vector predicted value.

[0052] Furthermore, the step (1) of obtaining the attitude information of the GPS module and the star sensor is specifically as follows: the attitude information of the GPS module is calculated based on the direct method using the unrelated baseline vectors of two current epochs; and the attitude information of the carrier relative to the inertial system is obtained by comparing the star map obtained by the star sensor with the ephemeris and solving the comparison.

[0053] Furthermore, T D Set the value to 0.05.

[0054] In addition, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program. A computer-readable storage medium stores a computer program thereon, wherein the computer program implements the steps of the above method when executed by the processor.

[0055] Beneficial effects: Compared with the prior art, the method of the present invention has the following significant advantages: 1. By combining the new information x 2 Detection method and state χ 2 A hybrid χ 2 The posture accuracy evaluation algorithm of the detection can not only target the sudden increase of error caused by the state mutation, but also use the new information χ 2 The detection method can accurately identify its noise characteristics, thereby performing troubleshooting or anti-error processing; it can also identify slowly increasing errors by using the state χ 2 The detection method detects the state obtained by the filter and the state obtained by the state recursor without measurement update, so as to determine whether there is an abnormal situation in the system; the complementary advantages of the two detection methods are combined to achieve the improvement of the attitude accuracy evaluation and anti-error performance of the GPS module and star sensor; 2. By combining the hybrid χ 2 Detection, design in the GPS module or star sensor any sensor under the relative failure situation, construct equivalent observation covariance matrix, and the constructed equivalent observation covariance matrix is ​​added to the Kalman filter gain, realize error compensation under the relative failure situation, improve the anti-error performance of filtering, embody its superiority compared with the standard Kalman filter method under the failure situation; 3, the present invention integrates GPS module and star sensor, GPS module system has higher stability and positioning accuracy, also has good prospects in attitude measurement, and star sensor has the advantages of high attitude determination accuracy and no cumulative error, and the combination of GPS module and star sensor can complement each other in attitude determination. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Shown is a flow chart of the evaluation method of the present invention;

[0057] Figure 2 The figure shows the aerial trajectory of the aircraft set in the embodiment of the present invention;

[0058] Figure 3 The figure shows the attitude angle error diagram outputted corresponding to the evaluation method described in the present invention. DETAILED DESCRIPTION

[0059] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0060] Set the trajectory of the aircraft carrying the GPS module and star sensor in the air as follows Figure 2 As shown. Figure 1 As shown, the invention discloses a method based on mixed χ 2 The method for evaluating the attitude accuracy of the detected GPS module / star sensor includes the following specific steps:

[0061] Step 1: Obtain the attitude information of the GPS module and star sensor;

[0062] Specifically, the attitude angle output of the GPS module is calculated based on the direct method using two unrelated baseline vectors of the current epoch to obtain attitude information; the star map obtained by the star sensor is compared with the ephemeris to obtain the attitude information of the carrier relative to the inertial system.

[0063] Step 2: Use new information x 2 The detection evaluates the accuracy of the posture information; specifically includes the following steps:

[0064] (1) Using the new information χ 2 The detection evaluates the accuracy of the positioning and attitude information of the star sensor and GPS module. The attitude measurement noise mean square error of the attitude angle output of the GPS module is set to 0.3°, and the attitude measurement noise mean square error of the star sensor output is set to 20″. The measurement information after state estimation in the filtering can be expressed as:

[0065]

[0066] In the formula, v 1 (k) is the GPS module measurement information; v 2 (k) is the new information measured by the star sensor; 1 (k) is the GPS module observation vector matrix; z 2 (k) is the star sensor observation vector matrix; is the predicted value of the observation vector; is the predicted value of the star sensor observation vector; H 1 is the GPS module observation matrix; H 2is the star sensor observation matrix; according to the actual situation, the measurement information is generally in the form of a white noise sequence, which basically obeys a normal distribution with a mean of zero.

[0067] (2) Construct the measurement innovation test statistic based on the measurement innovation. The formula is:

[0068] T(k)=v(k) T C(k) -1 v(k)

[0069] Where C(k) is the covariance matrix of the measurement innovation; substitute the corresponding v of the GPS module and star sensor at different time i respectively 1 (k), v 2 (k), C 1 (k), C 2 (k), obtain the final GPS module measurement innovation test statistic T i,1 (k), star sensor measurement innovation test statistic T i,2 (k); in which:

[0070] C 1 (k) = H 1 P 1 (k|k-1)H 1 T +V 1 (k)

[0071] C 2 (k) = H 2 P 2 (k|k-1)H 2 T +V 2 (k)

[0072] Where P 1 (k|k-1) is the GPS module state covariance matrix, P 2 (k|k-1) is the star sensor state covariance matrix; V 1 (k) is the covariance matrix of the GPS module observation noise; V 2 (k) is the covariance matrix of the star sensor observation noise;

[0073] By detecting the measurement innovation statistics, it can be reflected whether the observation values ​​output by the GPS module and the star sensor are abnormal. When an abnormal situation occurs, the mean value of the measurement innovation is no longer zero. The following assumptions can be listed:

[0074]

[0075] Among them, H 0 Indicates that there is no abnormality in the system, H 1Indicates that the GPS module or star sensor output is abnormal.

[0076] Since T is subject to χ with t degrees of freedom 2 distribution, this assumption can be stated as follows:

[0077]

[0078] (3) Set the boundary conditions for determining whether an abnormal situation occurs. The formula is:

[0079] T D ~x α 2 (t)

[0080] Where, T D represents the boundary value; α is the significance level, T D The value range is usually set to 0.01 to 0.15, and is set to 0.05 when performing posture accuracy assessment.

[0081] If T(k)≤T D , then it is considered that the output observation value has no abnormality; where T(k) takes T 1 (k), T 2 (k); otherwise, a failure is considered to exist; otherwise, an abnormality is considered to exist.

[0082] Step 3: Use state χ 2 Detect and evaluate the accuracy of the combined attitude information of the GPS module and star sensor;

[0083] (1) Setting the state vector is the attitude determination result of the star sensor, where Φ represents the attitude angle of the star sensor; Θ represents the angular velocity of the star sensor; Ψ represents the angular acceleration of the star sensor; the established state equation is:

[0084] x(k+1)=Fx(k)+w(k)

[0085] Where x is the state vector matrix; F is the state transition matrix; w is the system noise;

[0086] (2) Setting the observation vector is the attitude determination result output by the GPS module, where: represents the attitude angle of the GPS module; θ represents the angular velocity of the GPS module; ψ represents the angular acceleration of the GPS module; the observation equation is established, the formula is:

[0087] z(k)=H 1 x(k)+V 1 (k)

[0088] Where z is the combined state observation vector matrix;

[0089] (3) Using state χ 2 Detection through The two state estimates are subtracted for subsequent detection, where is the observed value z(k) obtained after the Kalman filter operation, and It is obtained by using prior information and then recursively utilizing the state recurser. Related to measurement information, it can be affected by system anomalies, and On the contrary, the two can be expressed as:

[0090]

[0091] Where K′ is the original filter gain; for The two estimated errors e obtained by recursion 1 (k), e 2 (k) can be expressed as:

[0092]

[0093] The difference between two state estimates can be expressed as:

[0094]

[0095] β(k) is a Gaussian random vector with a mean of zero, and its variance is expressed as: R(k) = E{β(k)β T (k)};

[0096] This yields the combined state test statistic:

[0097] T 3 (k) = β(k)R(k) -1 β(k)

[0098] T i,3 (k) is T at different time i 3 (k) value;

[0099] (4) By comparing the combined state test statistic with the boundary value, it is determined whether the observation value output by the GPS module and the star sensor combination is abnormal. If T 3 (k)≤T D , then it is determined that the output observation value has no anomaly, otherwise it is determined that there is an anomaly.

[0100] Step 4: Based on the above two 2The detection results of the detection method, when the observation value is abnormal, by constructing an equivalent weight factor in the Kalman filter process and adding the constructed equivalent observation covariance matrix to the filter gain, the attitude accuracy of the GPS module and the star sensor is evaluated. Specifically:

[0101] (1) Construct the equivalent weight factor ω;

[0102]

[0103] ω=diag[ω 1 ,ω 2 ,…,ω n ]

[0104] In the formula, ω is the equivalent weight factor, ω i is the equivalent weight factor at each moment, i=1,2……n, n is a positive integer; T i,j (k) is the test statistic of the j-th chi-square test at time i, j = 1, 2, 3; when j = 1, the corresponding new information χ 2 Detect and obtain the GPS module measurement new information test statistic T i,1 (k); When j = 2, the corresponding new information χ 2 Detection obtains the star sensor measurement innovation test statistic T i,2 (k); when j = 3, the corresponding state is χ 2 Detection obtains the combined state test statistic T i,3 (k); T D is the boundary value in the boundary condition; k is the state;

[0105] (2) The equivalent covariance matrix of the corresponding observations is constructed according to the equivalent weight factors, which is:

[0106]

[0107] In the formula, is the equivalent covariance matrix;

[0108] By adding the equivalent covariance matrix to the Kalman filter gain, the mutual robust fusion estimation of the GPS module and the star sensor can be realized. When the detection value is abnormal, the weight of the abnormal component can be reduced by the gain K through the equivalent weight factor and the equivalent covariance matrix, thereby enhancing the robustness of the GPS module / star sensor fusion estimation Kalman filter.

[0109] Step 5: Construct the filter gain equation and state estimation equation according to the equivalent covariance matrix; specifically:

[0110] The filter gain equation is:

[0111]

[0112] Where K is the filter gain; P 1 (k|k-1) is the GPS module state covariance matrix; H 1 is the GPS module observation matrix; k+1 refers to the next state after the k state, and k-1 refers to the previous state before the k state;

[0113] The state estimation equation is:

[0114]

[0115] In the formula, represents the final state estimator; Represents the predicted value of the final observation vector.

[0116] The weights of the state and observation can be adjusted dynamically according to the gain K. From the filter gain equation, it can be seen that the K value can be obtained by giving the initial value of the covariance matrix P 0 It is continuously iterated and updated. From the state estimation equation, it can be seen that the smaller K is, the greater the weight of the state estimation will be, and the larger K is, the greater the weight of the observation will be.

[0117] The subsequent combined navigation attitude determination process is carried out through Kalman filtering, and the attitude output of the carrier is finally obtained by solving the problem, such as Figure 3 As shown, the attitude output of the carrier is obtained by the above method, including attitude angle, angular velocity, and angular acceleration.

[0118] In addition, the present invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program. A computer-readable storage medium stores a computer program thereon, wherein the computer program implements the steps of the above method when executed by the processor.

Claims

1. A method for evaluating the attitude accuracy of GPS modules / star sensors based on hybrid chi-square detection. It is characterized in that The following steps are involved: (1) Obtain attitude information from the GPS module and star sensor installed in the aircraft; (2) Using new information χ 2 The detection evaluates the accuracy of the attitude information of the GPS module and the star sensor respectively, obtains the corresponding measurement innovation test statistics respectively, and judges whether there is any abnormality in the output data of the GPS module and the star sensor through the obtained measurement innovation test statistics; The specific steps include: (2.1) Using the new information χ 2 The accuracy of the attitude information of the GPS module and star sensor is evaluated respectively, and the formula for obtaining the measurement information is: In the formula, v 1 (k) is the GPS module measurement information; v 2 (k) is the new information measured by the star sensor; 1 (k) is the GPS module observation vector matrix; z 2 (k) is the star sensor observation vector matrix; Prediction value of observation vector for GPS module; is the predicted value of the star sensor observation vector; H 1 is the GPS module observation matrix; H 2 is the star sensor observation matrix; k is the state; k+1 refers to the state after the k state, and k-1 refers to the state before the k state; (2.2) According to the measurement innovation, the measurement innovation test statistic is obtained, and the formula is: T(k)=v(k) T C(k) -1 v(k) Where C(k) is the covariance matrix of the measurement information; substitute the corresponding v of the GPS module at different time i into 1 (k), C 1 (k) Obtain GPS module measurement innovation test statistic T i,1 (k); Substitute the corresponding v of the star sensor at different times i 2 (k), C 2 (k) Obtain the star sensor measurement innovation test statistic T i,2 (k); in which: C 1 (k)=H 1 P 1 (k|k-1)H 1 T +V 1 (k) C 2 (k)=H 2 P 2 (k|k-1)H 2 T +V 2 (k) Where P 1 (k|k-1) is the GPS module state covariance matrix, P 2 (k|k-1) is the star sensor state covariance matrix; V 1 (k) is the covariance matrix of the GPS module observation noise; V 2 (k) is the covariance matrix of the star sensor observation noise; (2.3) Setting boundary conditions for abnormal observations; T D ~χ α 2 (t) In the formula, α is the significance level; t is the degree of freedom; By comparing the two measurement innovation test statistics with the boundary value, it is determined whether the output data is abnormal. If the measurement innovation statistic is less than or equal to the boundary value, it is determined that the output data is normal, otherwise it is determined that there is an abnormality; (3) Using state χ 2 The detection performs accuracy evaluation on the attitude combination information of the GPS module and the star sensor, obtains the combination state test statistics, and judges whether there is an abnormality in the combined output data of the GPS module and the star sensor through the obtained combination state test statistics; specifically, the following steps are included: (3.1) The state quantity is set based on the attitude information of the star sensor. The formula is: In the formula, x is the state vector matrix; Φ represents the attitude angle of the star sensor; Θ represents the angular velocity of the star sensor; Ψ represents the angular acceleration of the star sensor; the state equation is established as: x(k+1)=Fx(k)+w(k) Where, F is the state transition matrix; w is the system noise; (3.2) The observation vector is set based on the attitude information of the GPS module. The formula is: Where z is the observation vector matrix; represents the attitude angle of the GPS module; θ represents the angular velocity of the GPS module; ψ represents the angular acceleration of the GPS module; the observation equation is established, the formula is: z(k)=H 1 x(k)+V 1 (k) (3.3) Using the state χ 2 Detect and estimate the accuracy of the attitude information of the GPS module and star sensor to obtain the first state estimate The second state estimator And make a difference, and finally get the combined state test statistic T 3 (k) specifically: T 3 (k)=β(k)R(k) -1 β(k) Where K′ is the original filter gain; R(k)=E{β(k)β T (k)}; T 3 (k) is the combined state test statistic; T i,3 (k) is T at different time i 3 (k) value; (3.4) By comparing the combined state test statistic with the boundary value, it is determined whether the output data is abnormal. If the combined state test statistic is less than or equal to the boundary value, it is determined that the output data is normal, otherwise it is determined that there is an abnormality; (4) When the output data in step (2) or step (3) is abnormal, an equivalent weight factor is constructed; specifically: ω=diag[ω 1 ,oh 2 ,…,oh n ] In the formula, ω is the equivalent weight factor, ω i is the equivalent weight factor at each moment, i = 1, 2...n, n is a positive integer; T i,j (k) is the test statistic of the j-th chi-square test at time i, j = 1, 2, 3; when j = 1, the corresponding new information χ 2 Detect and obtain the GPS module measurement new information test statistic T i,1 (k); When j = 2, the corresponding new information χ 2 Detection obtains the star sensor measurement innovation test statistic T i,2 (k); when j = 3, the corresponding state is χ 2 Detection obtains the combined state test statistic T i,3 (k); T D is the boundary value in the boundary condition; k is the state; Then, the equivalent covariance matrix is ​​obtained according to the equivalent weight factor, which is: In the formula, is the equivalent covariance matrix; V 1 (k) is the covariance matrix of the GPS module observation noise; (5) Based on the equivalent covariance matrix, the filter gain equation and state estimation equation are constructed, and the Kalman filter method is used to solve the final attitude result.

2. According to the method for evaluating the attitude accuracy of a GPS module / star sensor based on hybrid chi-square detection in claim 1, It is characterized in that The filter gain equation constructed in step (5) is: Where K is the filter gain; P 1 (k|k-1) is the GPS module state covariance matrix; H 1 It is the GPS module observation matrix; The state estimation equation is: In the formula, represents the state estimator; Represents the observation vector predicted value.

3. According to the method for evaluating the attitude accuracy of a GPS module / star sensor based on hybrid chi-square detection in claim 1, It is characterized in that The specific steps of obtaining the attitude information of the GPS module and the star sensor in step (1) are: Based on the direct method, the attitude information of the GPS module is calculated using the unrelated baseline vectors of two current epochs; the star map obtained by the star sensor is compared with the ephemeris and the attitude information of the carrier relative to the inertial system is obtained by solving the solution.

4. According to the method for evaluating the attitude accuracy of a GPS module / star sensor based on hybrid chi-square detection in claim 1, It is characterized in that T D Set the value to 0.

05.

5. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the computer program, the steps of the method according to any one of claims 1 to 4 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images