Information Fusion Estimation Theory And Its Applications To Spacecraft Control

This dissertation mainly studies information fusion estimation theory and its application to spacecraft control. The major researching contents of the dissertation are summarized as follows:1. A theory frame for information fusion estimation is established. A unified linear fusion model for information fusion estimation is proposed,and it can describe varied information including measuring information,apriority information,forecasting data and estimation information,and it lays a foundation for the theory frame of information fusion estimation. The concept of information magnitude is presented,and indicate that the more information magnitude,the more accurate. The principles of optimal linear fusion are put forward,i.e.,(1) the infonnation magnitude of optimal fusion estimation equals the sum of all the information magnitude of supplied information,and (2) the product of optimal fusion estimation and its information magnitude equals the sum of all products of supplied information and their information magnitudes. Based on the principle of optimal linear fusion,and a conclusion is drawn that "the more measurements are used,the higher accuracy is obtained". This result can play an important role in the accuracy design of multi-sensor system. The infonnation orthogonalization theorem and the information magnitude decomposition theorem are proposed and proved. The former provides a uniform method to solve the infonnation correlation problem,and the latter to solve information sharing problem of general federal filter.2. Based on the information fusion estimation theory,various information fusion structures and their algorithms are summarized,including centralized,decentralized,fusion mode and all-information mode. Various Kalman filters in the case of correlative infonnation arc summarized,including those with the correlation of measuring noise and system noise,those with colored noise and those in the case of one filter stimulating another. Mathematical simulation results,as shown in figures 4.3 through 4.5,testify the validity of solving the problem of one filter stimulating another by using the method of prolonging fusion period.3. Based on the information fusion estimation theory,a significant conclusion is drawn that the allocation of information cannot change global estimating accuracy,only local estimating accuracy. Therefore,information magnitude sharing coefficient is unnecessary to be adaptivcly determined,which rectifies a false standpoint in the literatures [ 108.129,131,140.141].4. Because die common fault detection method for sensors,residual x2 test,is insensitive to gradual-change-fault detection,a kind of residual x 2 test based on moving subnormal state ispresented,and its validity is testified by mathematical simulation results,as shown in figures 5.2 through 5.5. With regard to the lag of soft fault detection,a kind of system reconstruction tactic based on hierarchic alliance structure of federal filters is proposed. Mathematical simulation results,as shown in figure 6.10,show the validity of isolating fault by using the tactic.5. Aiming to a specific satellite,a kind of scheme of high accuracy fault-tolerant satellite's attitude determination system is designed,and testified by mathematical simulation. The calculated simulation results,as shown in figures. 6.5 through 6.8 and table 6.1 show that the more attitude sensors are used,the more accurate the estimation can be achieved.6. Initial orbit determination for satellites based on angle measurements is investigated. When one circle of angle observing data is available,a new kind of method,diagram approach,is presented When only short-arc angle measurements is available,the Genetic Algorithm (GA) is used for orbit determination. And an Algorism Float-Coded Genetic Algorithm (AFCGA) is designed,which have high convergence speed,high optimization accuracy and simple realization.Below are the author's main creationary works in this dissertation:1. The theory frame for information fusion estimation has been establishe...