Fang Jiancheng

Study on Innovation Adaptive EKF for In-Flight Alignment of Airborne POS

The Position and Orientation measurement System (POS) is a dedicated Strapdown Inertial Navigation System (SINS)/Global Positioning System (GPS) integrated system for airborne remote sensing. In-flight alignment (IFA) is an effective way to improve the accuracy and speed of initial alignment for an airborne POS. During IFA, the GPS provides the position and velocity references for the SINS, so the alignment accuracy will be degraded by unstable GPS measurements. To improve the alignment accuracy under unstable GPS measurement, an adaptive filtering algorithm of an extended Kalman filter (EKF) combined with innovation-based adaptive estimation is proposed, which introduces the calculated innovation covariance into the computation of the filter gain matrix directly. Then, this innovation adaptive EKF algorithm is used for the IFA of the POS with a large initial heading error. Moreover, it is optimized by blocked matrix multiplication to reduce the computational burden and improve the real-time performance. To validate the proposed algorithm, the car-mounted IFA experiment is carried out for the prototype of the airborne POS (TX-D10) under a turning maneuver, taking Applanixs POS/AV510 as a reference and changing the GPS measurement artificially. The experiment results demonstrate that the proposed algorithm can reach a better alignment accuracy than the EKF under unknown GPS measurement noises.

A Novel 3-DOF Axial Hybrid Magnetic Bearing

In this paper, we propose a novel structure of permanent-magnet-biased axial hybrid magnetic bearing. It has four segments of poles to control three degrees of freedom (3-DOF). Based on the inner and outer air gaps in conventional axial magnetic bearings, a novel air gap, called the subsidiary air gap, is constructed between the permanent magnet and the stator poles. This air gap separates the bias flux paths from the control flux paths. As a result, lower power loss of the axial magnetic bearing can be achieved due to lower magnetic reluctance of the control flux paths. Furthermore, by means of the equivalent magnetic circuit method and the 3-D finite-element method (FEM), we analyze and model the 3-DOF axial hybrid magnetic bearing. Experimental results show that the presented axial magnetic bearing has good control performance and little coupling among X, Y, and Z directions. However, the rotational power loss will be large at high speed because of the alternating flux density in the thrust plate produced by four segments of stator poles. Therefore, we propose a novel stator, named the parallel-slot stator, and novel thrust plate to reduce the rotational power loss effectively, which is assembled by DT4 and nanocrystalline materials. Meanwhile, we have designed and assembled an axial hybrid magnetic bearing prototype with the novel stator and thrust plate, which is applied in the five-degrees-of-freedom reaction flywheel system with angular momentum of 15 Nms at 5000 r/min. It is validated by the experimental results.

A New Structure for Permanent-Magnet-Biased Axial Hybrid Magnetic Bearings

We propose a new structure for a permanent-magnet-biased axial hybrid magnetic bearing. Starting with the inner air gap and the outer air gap of conventional axial magnetic bearings, we first construct a novel air gap between the permanent magnet and the outer housing, which we call the second air gap, separating the bias flux paths from the control flux paths. As a result, the control flux paths will have lower reluctance, and the power loss of the axial magnetic bearing will be lower. Next, we modeled this axial hybrid magnetic bearing and analyzed it using the equivalent magnetic circuit method, 2-D finite element method (FEM), and 3-D FEM. We have designed and assembled an axial hybrid magnetic bearing prototype for a reaction flywheel system with angular momentum of 15 Nmiddotmmiddots at a speed of 5000 r/min. The theoretical analysis and the prototype experiments show such advantages as simple structure, good force current and force displacement, and high operating reliability.

Integrated Model and Compensation of Thermal Errors of Silicon Microelectromechanical Gyroscope

Based on the thermal interferential moment, the dynamic thermal error induced by accelerations of a microelectromechanical system (MEMS) gyroscope is analyzed. The electromechanical-thermal error is discussed. The integrated thermal error compensation method considering the electromechanical-thermal error and the dynamic thermal error induced by accelerations is proposed. The experimental results show that the bias temperature sensitivity is reduced by more than one order of magnitude compared with the raw bias temperature sensitivity of the gyroscope. The integrated compensation method is reasonable and effective in the temperature error compensation of the MEMS gyroscope and outperforms the classical compensation method in performance.Based on the thermal interferential moment, the dynamic thermal error induced by accelerations of a microelectromechanical system (MEMS) gyroscope is analyzed. The electromechanical-thermal error is discussed. The integrated thermal error compensation method considering the electromechanical-thermal error and the dynamic thermal error induced by accelerations is proposed. The experimental results show that the bias temperature sensitivity is reduced by more than one order of magnitude compared with the raw bias temperature sensitivity of the gyroscope. The integrated compensation method is reasonable and effective in the temperature error compensation of the MEMS gyroscope and outperforms the classical compensation method in performance.

Hybrid simulation system study of SINS/CNS integrated navigation

In flight tests, to demonstrate the performance of integrated navigation systems, which are strapdown inertial navigation systems/celestial navigation systems (SINS/CNS), will involve a lot of effort and a heavy financial budget. So, it is important to design a functional self-contained hardware in a loop simulation system on the ground for solving the verification of SINS/CNS integrated navigation systems. Aiming at solving the main program, a high precision, versatile and better real-time performance hybrid simulation system of SINS/CNS integrated navigation is presented. The system adopts the design of hardware-functional modularization and software-flow integration, and makes use of a trajectory generator to produce simulated nominal trajectory data which is a uniform reference to signal process, and employs unique time synchronization algorithms to collect actual errors data of inertial sub-system and celestial sub-system. By combining with nominal trajectory data, the data smoothing, all-sky autonomous star map identification and integrated navigation algorithm based on unscented Kalman filtering (UKF), this system can accomplish a lot of system performance testing. It has the features of flexibility and extensibility and can be used to effectively reduce the experimental expense and shorten the development time of integrated navigation system, which is significant for studying algorithm performance, system speciality, and practical application of integrated navigation system.

Quaternion-Optimization-Based In-Flight Alignment Approach for Airborne POS

Position and Orientation System (POS) is a key technology that provides motion compensation information for imaging sensor in airborne remote sensing. In-flight alignment (IFA) is an important alignment method that can improve reaction speed and accuracy of measurement for airborne POS. The traditional IFA methods based on the Kalman filter and the adaptive extended Kalman filter (AEKF) are severely affected by flight maneuver; severe maneuver may result in alignment accuracy degeneration or failure. On the basis of optimization-based alignment method for static base alignment, a novel IFA method based on quaternion optimization is devised in this paper, aiming to boost up the robustness and the accuracy of IFA, which extends the range of the application of optimization-based alignment method from static base alignment to IFA. The proposed algorithm has the advantage that it can avoid the affection from severe maneuver. The real-time implementation is given. Flight experiment demonstrates that, compared with the IFA method based on AEKF, the proposed algorithm is better in converging speed and robusticity with the same alignment accuracy.Position and Orientation System (POS) is a key technology that provides motion compensation information for imaging sensor in airborne remote sensing. In-flight alignment (IFA) is an important alignment method that can improve reaction speed and accuracy of measurement for airborne POS. The traditional IFA methods based on the Kalman filter and the adaptive extended Kalman filter (AEKF) are severely affected by flight maneuver; severe maneuver may result in alignment accuracy degeneration or failure. On the basis of optimization-based alignment method for static base alignment, a novel IFA method based on quaternion optimization is devised in this paper, aiming to boost up the robustness and the accuracy of IFA, which extends the range of the application of optimization-based alignment method from static base alignment to IFA. The proposed algorithm has the advantage that it can avoid the affection from severe maneuver. The real-time implementation is given. Flight experiment demonstrates that, compared with the IFA method based on AEKF, the proposed algorithm is better in converging speed and robusticity with the same alignment accuracy.

Installation Direction Analysis of Star Sensors by Hybrid Condition Number

With the development of space programs, autonomous navigation of spacecraft is required in case of emergencies. Celestial navigation is a kind of fully autonomous navigation method, which has widely been used in many space missions. Because of the measurement noise in the celestial navigation system, the geometric relations between the spacecraft and various celestial bodies have a great impact on navigation accuracy. The positions of the sun, Earth, and other planets are fixed at a certain time, but stars are distributed all over the sky. The position of the measured star is mainly determined by the pointing direction of the star sensor. To explicitly demonstrate the effect of different pointing directions, an improved observability analysis method based on the hybrid condition number of the observability matrix for a nonlinear system is presented. The impact of the pointing direction of the star sensor on the navigation accuracy is analyzed using this analysis method, with the star-Earth angle used as a measurement for a near-Earth satellite. Simulations show that the value of the hybrid condition number is in good agreement with the position estimation error. This analysis method and the corresponding analysis results are useful in the design, construction, and analysis of celestial navigation systems for spacecraft.

Multisensor data synthesis using federated of unscented Kalman filtering

In this paper, a decentralized unscented Kalman filter (UKF) in federated configuration is developed for multisensor navigation data fusion. The UKF is a nonlinear, distribution approximation method that uses a finite number of points to propagate the states probability distribution through the systems nonlinear dynamics. In multisensor data fusion application, the configuration features of the federated unscented Kalman filter (FUKF) are investigated. To elaborate the concept of this filter structure, a case study of the strapdown inertial navigation system (SINS) integrated with astronavigation system (ANS) and global positioning system (GPS) is presented. Simulation results demonstrate the validity of the proposed filter configuration

A New Star Identification Algorithm based on Improved Hausdorff Distance for Star Sensors

High accuracy attitude for spacecraft can be determined by star sensors. The key technology is star pattern recognition. Based on the Hausdorff distance (HD) algorithm, a robust identification algorithm has been used for matching point-sets, which does not require absolute point correspondence. HD identification is not suitable for large attitude angle changes around the boresight, which usually results in a low recognition rate and low speed of identification. A new image similarity measure combined with an improved HD algorithm is proposed for recognizing stellar maps. An improved HD based on scalar distances is implemented to guarantee an acceptable success rate of recognition for large attitude changes around the boresight, which is noise resistant. Another improved HD based on vector distances is constructed to guarantee the recognition speed using a star dimensional configuration. Appropriate gray and matching thresholds are selected to improve recognition speed. Results of the semiphysical simulation show that the proposed algorithm is better than the HD identification method in terms of noise resistance, recognition rate, and speed of operation.High accuracy attitude for spacecraft can be determined by star sensors. The key technology is star pattern recognition. Based on the Hausdorff distance (HD) algorithm, a robust identification algorithm has been used for matching point-sets, which does not require absolute point correspondence. HD identification is not suitable for large attitude angle changes around the boresight, which usually results in a low recognition rate and low speed of identification. A new image similarity measure combined with an improved HD algorithm is proposed for recognizing stellar maps. An improved HD based on scalar distances is implemented to guarantee an acceptable success rate of recognition for large attitude changes around the boresight, which is noise resistant. Another improved HD based on vector distances is constructed to guarantee the recognition speed using a star dimensional configuration. Appropriate gray and matching thresholds are selected to improve recognition speed. Results of the semiphysical simulation show that the proposed algorithm is better than the HD identification method in terms of noise resistance, recognition rate, and speed of operation.

Correction technique for velocity and position error of inertial navigation system by celestial observations

The residual sensor errors and misalignment of an inertial navigation system cause deviation from the true navigation solution. The navigation errors grow as function of time, sensor errors and vehicle dynamics. A celestial aided inertial navigation system (INS), carried on a space vehicle, has to rely on the independent INS output until the vehicle gets out of the dense atmospheric layer. The misalignment and gyro drift then can be corrected employing techniques for in-flight alignment by celestial observations. The accelerometer bias can be detected and compensated after the burn out point. Subsequent contribution to navigation error due to misalignment and sensor errors thus can be restricted. The accumulated velocity and position errors, however, cannot be directly determined from celestial observations and the knowledge of accelerometer bias. A technique for velocity and position error computation is presented. Simulations for strapdown inertial navigation system (SINS), mechanized in space stabilized reference frame, are carried out. Tremendous improvement in navigation accuracy is observed