Fault Diagnosis Methods For Spacecraft Attitude Determination System

The attitude determination system provides the current attitude information of the spacecraft,which is the premise of attitude control.As the main attitude sensors of the spacecraft,star sensors and gyroscopes have a high probability of failure in orbit.If the failure cannot be diagnosed and isolated in time,the wrong attitude information will be introduced into the closed-loop control of the system,which may cause the payload could not work normally and even caused damage to the spacecraft.In engineering applications,bounded by the performance of on-board computers,the computational complexity and operation time of algorithms are strictly restricted.Therefore,it is of great significance to study fault diagnosis methods with high diagnostic accuracy and easy implementation on orbit.In this context,fault diagnosis methods for spacecraft attitude determination system are studied in this paper from the perspective of engineering application.The main contributions are as follows:The modelling and analysis of spacecraft attitude determination system are studied.It is difficult for star sensors to extract various error terms since the real attitude cannot be obtained by the spacecraft on orbit.To solve this problem,an error analysis method for star sensor on-orbit measurement values is proposed.Then,based on the real measurement data,the rationality of the proposed error classification and analysis method is verified.Finally,the star sensor model and gyroscope model including common fault types are established.The fault diagnosis method for attitude sensors based on dedicated filter is studied.It is known that too many resources will be consumed if fault diagnosis methods based on hardware redundancy only are used to realize fault isolation.To solve this problem,a fault diagnosis method combining hardware redundancy and analytical redundancy is studied,and a fault diagnosis method combining dedicated observer and extended Kalman filter is proposed for attitude sensors.This method regards the vector part of deviated quaternion as the residual.Then,the detection and isolation of attitude sensor faults are achieved based on the 3-? threshold evaluation principle.This method can quickly and accurately identify the intermittent faults,slow-changing faults of gyroscopes and intermittent faults,constant value faults of star sensors.The fault diagnosis method for attitude sensors based on simplified filter is studied.Considering the performance constraint of on-board computers and the excecution time constraint of on-board software,the fault diagnosis method based on dedicated filter are simplified from the following three aspects: Kalman filter algorithm,gain matrix solution,fault detection and isolation.First,the mathematical model of integrated attitude determination using star sensor and gyroscope is optimized from additive Kalman filter algorithm to a multiplicative one where state predictions do not need to be reset.As a result,the computational burden is reduced effectively while the filter performance still meets the requirements.Then,the measurement noise variance matrix in Kalman filter algorithm is regarded as independent scalars and the gain calculation is sequentially performed.In this way,the matrix inversion is avoided,and the amount of calculation is reduced by 1/3 while the convergence time and the stable error are completely consistent.Finally,a fault diagnosis and isolation method implemented by only one filter is proposed based on the philosophy of simplified filter,whose computational complexity is only about 1/4 of the one based on dedicated filter while ensuring the same fault diagnosis performance.The fault diagnosis method for attitude determination system based on comprehensive application model and data-driven is studied.In order to solve the problem that the fault diagnosis based on Kalman filter cannot locate and separate faults in face of multiple faults,a fault diagnosis method combining model and empirical mode decomposition is proposed,realizing the combination of model-based and data-based diagnostic methods.At the same time,in order to improve the feasibility of empirical mode decomposition in engineering applications,the conventional algorithm has been simplified accordingly under the premise that the performance meets the requirements.The fault simulation and diagnosis method verification based on a semi-physical simulation platform are studied.In order to verify the engineering practical value of the fault diagnosis methods proposed,a semi-physical simulation test for spacecraft control system based on the real on-board computer hardware is carried out.The main work includes: according to the engineering technological requirements,a fault simulation and injection module is generated for spacecraft attitude determination system where faults in star sensor or gyroscope may occur and may occur in the same time;then,on-board software codes based on Ada language are written.The test results show that the fault diagnosis method for spacecraft attitude determination system designed in this paper can be run on on-board computers,and the fault diagnosis result is accurate,having the possibility of engineering implementation.