A combined closed loop optimal design of experiments and online identification control approach

The main contribution of this paper is to propose a new approach, based on any specified continuous nonlinear model, for the combined optimal closed loop control of a process and online identification of a given model parameter. It deals with the optimal design of experiments: the idea is to twofold: first, based on a sensitivity model based predictive control, find the value of the input to apply during the experiment that allows optimizing a criteria based on the sensitivity of the process measure with respect to the unknown model parameter. Secondly, and in the meantime, based on the input/output measures collected, a process model and an observer, estimate online the model parameter. Moreover, constraints dealing with input, state and output limitations are accounted for. The main advantage of this approach is that both optimal control task and identification task are solved together online in order to get the online estimation of the unknown model parameter. This approach is applied here on a simple case in chemical engineering.