IDENTIFICAÇÃO ESTOCÁSTICA NO ESPAÇO DE ESTADO POR ANÁLISE DE CORRELAÇÕES CANÔNICAS: TEORIA, APLICAÇÃO E VALIDAÇÃO

Angie Joan Forero Forero¹; Jorge Andres Puerto Acosta¹; Celso P. Bottura¹

¹ UNICAMP

Baixar PDF doi:10.20906/CPS/CBA2018-0114

Resumo

In this paper, we treat the stochastic realization problem with exogenous inputs, which is to find a state vector $x$ and calculate the state space model in the innovative form. The state vector $x$ is given by the base vector of the predictor space calculated using the canonical correlation analysis. This method is applied to experimental identification of a vibrational rotary electromechanical system. Finally by the step forward prediction by the Kalman filter, is validated the identification due to the CCA method.

Palavras-chave: State space identification; Multivariate stochastic system; CCA method; Kalman filter