Recursive Subspace Identification Algorithms With Application In Gyro Parameter Identification

System identification together with control theory and state estimationconstitute the three pillars of modern control theory. With modern control theorycontinuously being mature and improved, system identification theory also gainsrapid development. So far, this theory has been widely used in control engineeringand many other fields. As a new identification method which is based on thestate-space model, subspace identification method has attracted extensive attentionfrom scholars at home and abroad from the moment it has been proposed, because ofits advantages such as the prominent capability to handle the multiple-inputmultiple-output system, abbreviated as MIMO system, and so on. As manyengineering problems require real-timely estimating the system parameters online, sothe recursive subspace identification method has become a focus for researching inthe subspace identification field.In this paper, based on the offline subspace identification algorithms ofMulti-Variables Output-Error State-Space, abbreviated as MOESP, a recursivesubspace identification algorithm, which is based on QR decomposition andpropagator, has been researched and applied in identifying the parameters which arepart of the dynamic model of gyro in the seeker coordinator.Firstly, a recursive subspace identification method has been researched, which isbased on the MOESP offline subspace identification algorithms, and has given thebasic description of the problem solved by the recursive subspace identificationalgorithm. The thought of the recursive subspace identification algorithm which isbased on QR decomposition and propagator has been outlined, and the process ofhow the algorithm been implemented is listed.Secondly, the recursive subspace identification algorithm is verified by a seriesof numerical simulations. The simulations are mainly for linear systems withdifferent orders whose parameters are time-invariant, time-variant and suddenlychanged, in the case of noise absent, as well as noises existed, includingvariances-differed measurement noise and process noise. Through the above therecursive subspace identification algorithm is verified.Finally, the implementation of applying the recursive subspace identificationalgorithm on identifying the parameters of gyro has been achieved. Through therecursive subspace identification algorithm, the parameters which are part of thedynamic model of gyro in the seeker coordinator have been identified. As the recursive subspace identification algorithm can obtain an equivalent form of theoriginal system, but it can’t define the parameters of gyro directly. Therefore thediscretization of the continuous dynamic model of gyro is needed to get its discretelinear state space model, using the property that the systems through equivalenttransformation have the same eigenvalues and characteristic polynomial to identifythe parameters. At last the applicability of the algorithm in engineering field isverified by simulation.