A low-cost commercial remote sensing satellite attitude determination information fusion method

Abstract

The present invention discloses a low-cost commercial remote sensing satellite attitude determination information fusion method. On a commercial remote sensing satellite, two star sensors, two gyroscopes, two sun sensors, and one magnetometer are used to measure the satellite attitude. According to the data availability of each sensor, conditional judgment is performed, and quaternions and gyroscope angular velocity are used to perform attitude determination, and finally satellite attitude determination information is obtained. The information fusion method of the present invention is simple, does not rely on complex algorithms and expert experience, occupies less on-board resources, is easy to maintain, and reserves an interface for ground-based forced intervention in the on-board information fusion method, thereby improving the reliability and practicality of the fusion algorithm.

Description

Technical Field

[0001] The invention belongs to the field of satellite technology, and in particular relates to a remote sensing satellite attitude determination information fusion method. Background Art

[0002] In order to ensure the stable acquisition of spacecraft's spatial orientation information, satellites usually need to be equipped with a variety of attitude sensors to measure the relative angular position and angular velocity of the spacecraft's body coordinate system relative to a certain reference coordinate system, and determine the spacecraft's attitude through on-board determination and filtering algorithms.

[0003] Zhang Rui and others from Shanghai Microsatellite Engineering Center published "Small Satellite Attitude Determination System and Method" (publication number CN101556155A), which uses horizon, sun sensor and magnetometer as sensors, and switches multiple attitude determination algorithms on board to improve the reliability of the system. However, the attitude measurement capability of the sensor used in this scheme limits the accuracy of attitude determination and cannot meet the requirements for remote sensing satellite use.

[0004] In response to the high-precision attitude determination requirements of remote sensing satellites, Cheng Chunxiao and others from Aerospace Science and Industry Corporation of China proposed "A satellite multi-sensor fusion attitude determination method and system", which uses the idea of ​​federal filtering to improve the accuracy of satellite attitude determination. However, since the algorithm relies on the design of sub-filters and main filters, the calculation is complex and occupies a lot of on-board computing resources, which is not conducive to the actual promotion of satellite engineering.

[0005] To this end, Zhao Junsuo and others from the Institute of Software of the Chinese Academy of Sciences published the "Method for Integrated Adaptive Nanosatellite Attitude Determination" (publication number CN106249590A), which outputs the attitude determination sensor combination scheme under different conditions based on the fuzzy set controller, while ensuring the determination accuracy and reducing the on-board resource consumption. However, on the one hand, the formulation of fuzzy set rules relies heavily on prior expert knowledge; on the other hand, the method focuses on the combination of sensor selection, and pays little attention to the fusion of on-board measurement information.

[0006] With the rise of commercial satellites, especially remote sensing commercial satellites in recent years, on the one hand, it is necessary to provide high-precision attitude determination results, and on the other hand, it is necessary to develop low-cost on-board attitude determination solutions.

[0007] Early attitude determination schemes were simple, reliable and low-cost, but most of the attitude sensors used were of low accuracy and failed to fully utilize multi-sensor information for data fusion, making them unsuitable for remote sensing satellites with high attitude determination accuracy. To ensure high-precision attitude determination, remote sensing satellites usually use star sensors as the main attitude determination components, combined with data fusion or adaptive algorithms to ensure the reliability of attitude determination. However, traditional algorithms usually rely on complex filter combination ideas and prior expert experience, with high design and maintenance costs, and sensors are usually equipped with backup configurations, which do not fully meet the needs of low-cost commercial remote sensing satellites. Summary of the invention

[0008] In order to overcome the deficiencies of the prior art, the present invention provides a low-cost commercial remote sensing satellite attitude determination information fusion method. On a commercial remote sensing satellite, two star sensors, two gyroscopes, two sun sensors, and a magnetometer are used to measure the satellite attitude; according to the data availability of each sensor, conditional judgment is performed, and quaternions and gyroscope angular velocity are used to perform attitude determination, and finally satellite attitude determination information is obtained. The information fusion method of the present invention is simple, does not rely on complex algorithms and expert experience, occupies less on-board resources, is easy to maintain, and reserves an interface for ground-based forced intervention in the on-board information fusion method, thereby improving the reliability and practicality of the fusion algorithm.

[0009] The technical solution adopted by the present invention to solve the technical problem includes the following steps:

[0010] Step 1: On a commercial remote sensing satellite, two star sensors, two gyroscopes, two sun sensors, and one magnetometer are used to measure the satellite attitude.

[0011] Step 2: Determine whether at least one star sensor and at least one gyro data are available in the current cycle: if the conditions are met, jump to step 12; otherwise, jump to step 3;

[0012] Step 3: Determine whether the data of both star sensors are unavailable: if both are unavailable, jump to step 4; otherwise, jump to step 13;

[0013] Step 4: Determine whether the data from both the sun sensors and the magnetometer are available: If the conditions are met, jump to step 8; otherwise, jump to step 5;

[0014] Step 5: Determine whether at least one gyroscope data is available: if so, jump to step 6; otherwise, jump to step 7;

[0015] Step 7: In this cycle, there is only gyro measurement angular velocity data, so the gyro integral calculation is performed based on the quaternion data of the previous cycle to determine the quaternion in this cycle; and the obtained quaternion and angular velocity are used as the attitude output for this cycle, and the attitude determination of this cycle ends;

[0016] Step 7: The sensor data in this cycle cannot determine the attitude, the attitude determined in the previous cycle is output, and the attitude determination failure is reported, and the attitude determination in this cycle ends;

[0017] Step 8: Determine the quaternion using the dual vector data based on the sun sensor and magnetometer data, and jump to step 9;

[0018] Step 9: Determine whether at least one gyroscope data is available: if available, jump to step 10; otherwise, jump to step 13;

[0019] Step 10: Determine whether the modulus of the gyroscope data is greater than a set threshold: if it is greater than the set threshold, jump to step 11; otherwise, jump to step 12;

[0020] Step 11: The quaternion and gyro angular velocity data determined by the double vectors of this cycle are used as the attitude output determined in this cycle, and the attitude determination of this cycle is completed;

[0021] Step 12: Using the quaternion and gyro angular velocity determined in this cycle, the attitude determination method is used to further improve the attitude determination accuracy, and the attitude determination of this cycle is completed;

[0022] Step 13: Using the quaternion determined in this cycle, the angular velocity of this cycle is estimated using the attitude determination method, and the attitude determination accuracy is further improved. The attitude determination of this cycle is completed.

[0023] Furthermore, the step 12 is specifically as follows:

[0024] Step 12-1: Use quaternion and gyro angular velocity as input;

[0025] Step 12-2: Let the attitude quaternion be q = [q0 q1 q2 q3] T , where q0 is a scalar and the gyro drift is b = [b x b y b z ] T , then the state variable x=[q0 q1 q2 q3 b x b y b z ] T ;

[0026] Step 14-3: The system state equation is

[0027]

[0028] in,

[0029]

[0030] ω=[ω xω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope, O 3×1 It is a zero matrix with 3 rows and 1 column;

[0031] Step 12-4: The Jacobian matrix of the system state equation is:

[0032]

[0033] in,

[0034]

[0035] Step 12-5: The measurement equation of the attitude quaternion is:

[0036] z q =Hx+v

[0037] Among them, z q represents the attitude quaternion q, v represents the measurement noise,

[0038]

[0039] Step 12-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

[0040] Furthermore, the step 13 is specifically as follows:

[0041] Step 13-1: Using quaternion q = [q0 q1 q2 q3] T As input;

[0042] Step 13-2: State variable x = [q0 q1 q2 q3 ω x ω y ω z ] T ; ω=[ω x ω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope;

[0043] Step 13-3: The model state equation is:

[0044]

[0045] Among them, T e is the external torque, and the moment of inertia matrix is ​​I:

[0046]

[0047] Step 13-4: The Jacobian equation of the system state equation is:

[0048]

[0049]

[0050]

[0051]

[0052] Step 13-5: The measurement equation of the attitude quaternion is:

[0053] z q =Hx+v

[0054] Among them, z q represents the attitude quaternion q, v represents the measurement noise,

[0055]

[0056] Step 13-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

[0057] Furthermore, the EKF algorithm is specifically:

[0058] Given the model state equation:

[0059]

[0060] Among them, x is the state variable of the system, v is the state noise of the system model, and u is the model input variable;

[0061] Given the model measurement equation:

[0062] y=h(x,w,u)

[0063] Where w represents the system measurement noise;

[0064] Assuming that the noise is independent and uncorrelated, v~(0, Q), w~(0, R), the EKF calculation process for a single cycle is:

[0065] a. Make forecast corrections:

[0066]

[0067]

[0068]

[0069]

[0070] Among them, H k The Jacobian matrix representing the measurement equation for the current period, Indicates the one-step recursive value of the previous cycle state, K k Represents the Kalman gain matrix calculated in the current cycle, P k / k-1 represents the one-step recursive value of the state error covariance matrix of the previous cycle, R represents the variance of the measurement noise, represents the error covariance matrix estimated in this period, represents the estimated state variables for this period;

[0071] b. Perform one-step recursion:

[0072]

[0073] Φ k =I+F k δt+o(·)

[0074]

[0075] Among them, δ represents the sampling period, o(·) represents the state transfer matrix Φ k The high-order error term when expanded, Q represents the variance of the state noise; when solving Φ k Taylor expansion to 1st or 2nd order accuracy as required.

[0076] The beneficial effects of the present invention are as follows:

[0077] 1. The present invention utilizes the common configuration of current commercial remote sensing satellites, and fully exploits the on-board attitude determination capability by using the information fusion method without increasing the cost of the single attitude determination machine, thereby providing a guarantee for the attitude accuracy requirements of remote sensing missions. At the same time, the information fusion method of the present invention is simple, does not rely on complex algorithms and expert experience, occupies less on-board resources, and is easy to maintain.

[0078] 2. The star sensor, gyroscope, sky sensor and magnetometer data availability judgment of the present invention only reads the available flag given by the satellite platform, and decouples the data processing from the attitude determination algorithm. There is no need to pay extra attention to the actual status of the sensor and whether the ground forces intervention in the use of a single machine, which simplifies the design difficulty of the determination algorithm.

[0079] 3. The present invention is based on a modular design concept and can be easily replaced with any attitude information estimation and correction algorithm that uses the preliminarily determined quaternion and angular velocity information. It does not rely on complex filtering algorithms and advanced expert knowledge while improving the accuracy of attitude determination. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0081] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0082] Therefore, the purpose of the present invention is summarized in the following two points:

[0083] (1) Based on the low-cost requirements of commercial remote sensing satellites, the capabilities of onboard sensors should be exploited as much as possible and high-precision attitude determination capabilities should be provided;

[0084] (2) Based on the needs of rapid design and production, provide simple and reliable commercial remote sensing satellite attitude determination solutions as much as possible, while reducing the difficulty of maintenance as much as possible.

[0085] The present invention discloses a low-cost commercial remote sensing satellite attitude determination information fusion method and device. Based on the existing general on-board configuration of commercial remote sensing satellites, star sensors, gyroscopes, sun sensors, and magnetometers are selected as attitude measurement devices, and a simple and reliable satellite attitude determination information fusion method is proposed to provide high-precision attitude information for remote sensing satellites.

[0086] (1) The attitude measurement device of the present invention is selected based on the general configuration of current commercial remote sensing satellites, using only two star sensors, two gyroscopes, two sun sensors, and one magnetometer, without increasing the satellite production cost;

[0087] (2) Based on the idea of ​​information fusion, the present invention fully exploits the capabilities of each sensor to ensure continuous output of high-precision posture information;

[0088] (3) The present invention is simple and reliable, easy to maintain, does not rely on prior expert knowledge, and occupies little on-board resources.

[0089] A low-cost commercial remote sensing satellite attitude determination information fusion method comprises the following steps:

[0090] Step 1: On a commercial remote sensing satellite, two star sensors, two gyroscopes, two sun sensors, and one magnetometer are used to measure the satellite attitude.

[0091] Step 2: Determine whether at least one star sensor and at least one gyro data are available in the current cycle: if the conditions are met, jump to step 12; otherwise, jump to step 3;

[0092] Step 3: Determine whether the data of both star sensors are unavailable: if both are unavailable, jump to step 4; otherwise, jump to step 13;

[0093] Step 4: Determine whether the data from both the sun sensors and the magnetometer are available: If the conditions are met, jump to step 8; otherwise, jump to step 5;

[0094] Step 5: Determine whether at least one gyroscope data is available: if so, jump to step 6; otherwise, jump to step 7;

[0095] Step 6: When there is only gyro measurement angular velocity data, the gyro integral is calculated based on the quaternion data of the previous cycle to determine the quaternion of this cycle; and the obtained quaternion and angular velocity are used as the output of the attitude determination of this cycle, and the attitude determination of this cycle ends;

[0096] Step 7: When the sensor data cannot determine the posture, the posture determined in the previous cycle is output and the posture determination is reported to be invalid, and the posture determination in this cycle ends;

[0097] Step 8: Determine the quaternion using the dual vector data based on the sun sensor and magnetometer data, and jump to step 9;

[0098] Step 9: Determine whether at least one gyroscope data is available: if available, jump to step 10; otherwise, jump to step 13;

[0099] Step 10: Determine whether the modulus of the gyroscope data is greater than a set threshold: if it is greater than the set threshold, jump to step 11; otherwise, jump to step 12;

[0100] Step 11: The quaternion and gyro angular velocity data determined by the double vectors of this cycle are used as the attitude output determined in this cycle, and the attitude determination of this cycle is completed;

[0101] Step 12: Using the quaternion and gyro angular velocity determined in this cycle, the attitude determination method is used to further improve the attitude determination accuracy, and the attitude determination of this cycle is completed;

[0102] Step 13: Using the quaternion determined in this cycle, the angular velocity of this cycle is estimated using the attitude determination method, and the attitude determination accuracy is further improved. The attitude determination of this cycle is completed.

[0103] Furthermore, the step 12 is specifically as follows:

[0104] Step 12-1: Use quaternion and gyro angular velocity as input;

[0105] Step 12-2: Let the attitude quaternion be q = [q0 q1 q2 q3] T , where q0 is a scalar and the gyro drift is b = [b x b y b z ] T , then the state variable x=[q0 q1 q2 q3 b x b y b z ] T ;

[0106] Step 12-3: The system state equation is

[0107]

[0108] in,

[0109]

[0110] ω=[ω x ω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope;

[0111] Step 12-4: The Jacobian matrix of the system state equation is:

[0112]

[0113] in,

[0114]

[0115] Step 12-5: The measurement equation of the attitude quaternion is:

[0116] z q =Hx+v

[0117] Among them, z q represents the attitude quaternion q, v represents the measurement noise,

[0118]

[0119] Step 12-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

[0120] Furthermore, the step 13 is specifically as follows:

[0121] Step 13-1: Using quaternion q = [q0 q1 q2 q3] T As input;

[0122] Step 13-2: State variable x = [q0 q1 q2 q3 ω x ω y ω z ] T ; ω=[ω x ω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope;

[0123] Step 13-3: The model state equation is:

[0124]

[0125] Among them, T e is the external torque, and the moment of inertia matrix is ​​I:

[0126]

[0127] Step 13-4: The Jacobian equation of the system state equation is:

[0128]

[0129]

[0130]

[0131]

[0132] Step 13-5: The measurement equation of the attitude quaternion is:

[0133] z q =Hx+v

[0134] Among them, z q represents the attitude quaternion q, v represents the measurement noise,

[0135]

[0136] Step 13-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

[0137] Furthermore, the EKF algorithm is specifically:

[0138] Given the model state equation:

[0139]

[0140] Given the model measurement equation:

[0141] y=h(x,w,u)

[0142] Assuming that the noise is independent and uncorrelated, v~(0, Q), w~(0, R), the EKF calculation process for a single cycle is:

[0143] c. Make forecast revisions:

[0144]

[0145]

[0146]

[0147]

[0148] d. Perform one-step recursion:

[0149]

[0150] Φ k =I+F k δt+o(·)

[0151]

[0152] Among them, in solving Φ k Taylor expansion to 1st or 2nd order accuracy as required.

[0153] The two attitude determination methods in the aforementioned invention content use the common EKF algorithm in the embodiments, but it should be pointed out that the actual methods may include but are not limited to any algorithm that uses quaternions and angular velocity to estimate and correct attitude information.

Claims

1. A low-cost commercial remote sensing satellite attitude determination information fusion method, characterized in that: The steps include: Step 1: On a commercial remote sensing satellite, two star sensors, two gyroscopes, two sun sensors, and one magnetometer are used to measure the satellite attitude. Step 2: Determine whether at least one star sensor and at least one gyro data are available in the current cycle: If the conditions are met, jump to step 12; Otherwise, jump to step 3; Step 3: Determine whether the data of both star sensors are unavailable: If both are unavailable, jump to step 4; Otherwise, jump to step 13; Step 4: Determine whether the data from both the sun sensors and the magnetometer are available: If the conditions are met, jump to step 8; otherwise, jump to step 5; Step 5: Determine whether at least one gyroscope data is available: if so, jump to step 6; Otherwise, jump to step 7; Step 7: In this cycle, there is only gyro measurement angular velocity data, so the gyro integral calculation is performed based on the quaternion data of the previous cycle to determine the quaternion in this cycle; and the obtained quaternion and angular velocity are used as the attitude output for this cycle, and the attitude determination of this cycle ends; Step 7: The sensor data in this cycle cannot determine the attitude, the attitude determined in the previous cycle is output, and the attitude determination failure is reported, and the attitude determination in this cycle ends; Step 8: Determine the quaternion using the dual vector data based on the sun sensor and magnetometer data, and jump to step 9; Step 9: Determine whether at least one gyroscope data is available: if available, jump to step 10; Otherwise, jump to step 13; Step 10: Determine whether the modulus of the gyroscope data is greater than a set threshold: if it is greater than the set threshold, jump to step 11; otherwise, jump to step 12; Step 11: The quaternion and gyro angular velocity data determined by the double vectors of this cycle are used as the attitude output determined in this cycle, and the attitude determination of this cycle is completed; Step 12: Using the quaternion and gyro angular velocity determined in this cycle, the attitude determination method is used to further improve the attitude determination accuracy, and the attitude determination of this cycle is completed; Step 13: Using the quaternion determined in this cycle, the angular velocity of this cycle is estimated using the attitude determination method, and the attitude determination accuracy is further improved. The attitude determination of this cycle is completed.

2. A low-cost commercial remote sensing satellite attitude determination information fusion method according to claim 1, characterized in that: The step 12 is specifically as follows: Step 12-1: Use quaternion and gyro angular velocity as input; Step 12-2: Let the attitude quaternion be q = [q0 q1 q2 q3] T , where q0 is a scalar and the gyro drift is b = [b x b y b z ] T , then the state variable x=[q0 q1 q2 q3 b x b y b z ] T ; Step 14-3: The system state equation is in, ω=[ω x ω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope, O 3×1 It is a zero matrix with 3 rows and 1 column; Step 12-4: The Jacobian matrix of the system state equation is: in, Step 12-5: The measurement equation of the attitude quaternion is: z q =Hx+v Among them, z q represents the attitude quaternion q, v represents the measurement noise, Step 12-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

3. A low-cost commercial remote sensing satellite attitude determination information fusion method according to claim 1, characterized in that: The step 13 is specifically as follows: Step 13-1: Using quaternion q = [q0 q1 q2 q3] T As input; Step 13-2: State variable x = [q0 q1 q2 q3 ω x ω y ω z ] T ; ω=[ω x ω y ω z ] T is the angular velocity of the body relative to the inertial system measured by the gyroscope; Step 13-3: The model state equation is: Among them, T e is the external torque, and the moment of inertia matrix is ​​I: Step 13-4: The Jacobian equation of the system state equation is: Step 13-5: The measurement equation of the attitude quaternion is: z q =Hx+v Among them, z q represents the attitude quaternion q, v represents the measurement noise, Step 13-6: After obtaining the corrected quaternion and gyro zero bias based on the EKF algorithm, subtract the zero bias value from the gyro measurement value of this period to obtain the corrected determined angular velocity.

4. A low-cost commercial remote sensing satellite attitude determination information fusion method according to claim 2 or 3, characterized in that: The EKF algorithm is specifically: Given the model state equation: Among them, x is the state variable of the system, v is the state noise of the system model, and u is the model input variable; Given the model measurement equation: y=h(x,w,u) Where w represents the system measurement noise; Assuming that the noise is independent and uncorrelated, v~(0, Q), w~(0, R), the EKF calculation process for a single cycle is: a. Make forecast corrections: Among them, H k The Jacobian matrix representing the measurement equation for the current period, Indicates the one-step recursive value of the previous cycle state, K k Represents the Kalman gain matrix calculated in the current cycle, P k / k-1 represents the one-step recursive value of the state error covariance matrix of the previous cycle, R represents the variance of the measurement noise, represents the error covariance matrix estimated in this period, represents the estimated state variables for this period; b. Perform one-step recursion: Φ k =I+F k δt+o(·) Among them, δ represents the sampling period, o(·) represents the state transfer matrix Φ k The high-order error term when expanded, Q represents the variance of the state noise; when solving Φ k Taylor expansion to 1st or 2nd order accuracy as required.

Images