MPC performance assessment and nonlinear state estimation

Luo Ji
Joining the group in 2009, Luo Ji's research is currently focus on the controller performance evaluation discipline and the state estimation of nonlinear system.

MPC performance evaluation (MPA) is an important issue in the theoretical implementation and the industrial application. Due to the existence of white noise (may contain process and measurment noises), the unmodeled disturbance, noise model mismatch, or even the process model mismatch, the output performance and input cost may be quite different even if the same control stratedy is used. To address this issue, we have proposed a stage cost function and examined its value by lab simulations. This could be applied as the evaluation critierion between different controller results, even for some which are actually not MPC controllers. The cases when unmodeled disturbance or model mismatch exist are also studied.

The evaluation and implementation of nonlinear state estimators is also critial in real applications since (1) most industial systems are a lot states which can not be measured directly (2) the controller needs some predictions where these states are to make the right decision. From the previous work, Moving Horizon Estimators (MHE) could be a good candidate method to depict the state space from those nonlinear processes. My work contains (1) theorectically design and verify a stable, convergent, well-defined nonlinear MHE on realistic industrial plants, (2) practical implementation of nonlinear MHE algorithm on real chemical plants and (3) more advanced and simplified estimation algorithms for special operating conditions.