Parameter Set Estimation of Time Varying Systems
AIR FORCE RESEARCH LAB BOLLING AFB DC
Pagination or Media Count:
Parameter set estimation PSE, a class of system identification schemes which aim at characterizing the uncertainty in the identification experiment, will play a vital role in robust identification for control. An important step in current research along these lines is development of PSE algorithms for systems which are time varying in nature this is particularly true if the identified model set is to be used in an adaptive setting. In this dissertation, the Optimum Volume Ellipsoid OVE algorithm for parameter set estimation of time-invariant systems is extended to time-varying systems. Building on this development of the OVE algorithm for Time-Varying systems OVETV, two algorithms are also presented for reducing the computational complexity of the optimal time update equations. These algorithms, scalar addition and scalar multiplication, reduce the computational complexity of the time update equations by constraining the new ellipsoid to be parameterized by the previous ellipsoid and a single new parameter. Following this, extensions to the OVE and OVETV algorithms are presented. The algorithms are extended to handle multiple-input, single-output MISO systems. It is shown how knowledge of dependencies in the parameter variations can be exploited to reduce the number of computations in the resulting algorithm. We show how the scope of OVE-ISP, an OVE-based input synthesis procedure, can be extended to handle systems with known transportation lag. Lastly, it is shown how the OVE and OVETV algorithms can be utilized for fault detection and isolation FDI. Two methods are suggested for detecting faults in dynamical systems. The first method relies on a consistency check which is integral to the OVE and OVETV algorithms, while the second method utilizes an ellipsoid intersection test to detect a fault.
- Operations Research
- Computer Programming and Software