State Estimation of a Nonlinear CSTR Using a Novel Asynchronous Data Fusion Based on Adaptive Extended Kalman Filter

摘要

This paper presents the state estimation problem for nonlinear industrial systems using asynchronous measurements to simulate the circumstances of practical processes. There exist three distinct difficulties encountered in real-world applications, i.e. lack of perfect knowledge on model-reality mismatch and noise distribution matrices, diversified sampling rate and communication delays. For the first drawback, an adaptive fading extended Kalman filter (AFEKF) is utilized to simultaneously alleviate both model uncertainty and measurement noises. For the second problem, a distributed AFEKF is proposed to cover the issue of multi-rate measurement signals. To meet the last challenge, three methods are proposed which encompass the fusion of modified AFEKF and Alexander EKF(Alexander 1991) method. A comparative study was further conducted on a simulated nonlinear CSTR to demonstrate the extent of improvement achieved over the existing methodologies. Simulation outcomes indicate a significant superiority of the proposed approaches.