Abstract：

Identification of stochastic nonlinear dynamical systems has long been recognized as a difficult problem. There are known maximum likelihood solutions to identification of stochastic linear systems with additive Gaussian noise. However, if the noise is non-Gaussian and if the dynamics are nonlinear, a straightforward maximum likelihood solution is not possible except in special cases. In this talk, we present an approximate maximum likelihood solution for stochastic nonlinear dynamical systems. As part of this approach, we exploit the ability of Markov Chain Monte Carlo methods to approximate non-Gaussian density functions and the ability of expectation maximization to handle hidden state variables. If time permits, an extension of this work to input design will also be presented.

Dr. Bhushan Gopaluni graduated with a Bachelor’s degree in Chemical engineering from Indian Institute of Technology, Madras, India. He then obtained a Ph.D. in Chemical engineering from the University of Alberta, Edmonton, Canada.  He is currently an assistant professor in the department of chemical and biological engineering at the University of British Columbia, Vancouver, Canada. He specializes in process modeling and control. In particular, he is interested in novel approaches to system identification, experiment design, fault detection and diagnosis, and control.