Petrova R. Synthesis of high-speed adaptive observers for linear objects

The thesis deals development and research of efficient methods for synthesis of adaptive observers for linear dynamic systems which adequately reproduce properties of the objects under lack of complete information about their properties and operating conditions. The problem of building of an observer of an implicit type that provides asymptotically accurate estimation is considered. An observer with reduced number of estimated parameters is synthesized. Special issues of observation matrix calculation algorithm implementation are discussed. A method for building of multi-step estimation procedures is proposed that use nonlinear transformation of input variables and estimation errors. A simulation of the synthesized adaptive observers was performed that fully confirmed theoretical conclusions. Using the proposed approaches, dynamic models were built that describe technological processes of absorption-desorption part of soda ash production and adequately reflect properties of the real object.