Estimation and control of industrial processes with particle filters

Abstract

We present a probabilistic approach to state estimation and control of industrial processes. In particular, we adopt a jump Markov linear Gaussian (JMLG) model to describe an industrial heat exchanger. The parameters of this model are identified with the expectation maximisation (EM) algorithm. After identification, particle filtering algorithms are adopted to diagnose, in real-time, the state of operation of the heat exchanger. The particle filtering estimates are then used to drive an automatic control system.