Soft Sensor for COx Content in Tail Gas of PX Oxidation Side Reactions based on Particle Filters and EM Algorithm

Abstract

Based on expectation-maximization algorithm, parameter estimation was proposed for data-driven nonlinear models in this work. On this basis, particle filters were used to approximately calculate integrals, deriving EM algorithm based on particle filter. And the effectiveness of using the proposed algorithm for the soft sensor of CO_x content in tail gas of PX oxidation side reactions was verified through simulation results.