Modeling and tracker for unknown nonlinear stochastic delay systems with positive input constraints

An approach for universal modeling and tracker design for input-constrained unknown nonlinear input time-delay stochastic sampled-data systems is newly proposed in this paper. First, the improved observer/Kalman filter identification (OKID) method, which uses the current output measurement to estimate the current state, is newly proposed in this paper and it is shown that it outperforms the traditional OKID method. In addition, it is shown that the newly proposed current output-based Kalman filter is a well-performed output estimator in the extreme case, in which it is not a filter anymore, becoming a universal way of formulating an artificial system model of a real physical process without disturbing its normal operation. Consequently, the proposed artificial system model has the following advantages: (i) It is capable of quantifying the stochastic and deterministic characteristics of the dynamical system of interest; (ii) It is capable of carrying out the analyses of various control-design methodologies to achieve the performance specifications in the pre-study phase; and (iii) It is capable of estimating missing and/or abnormal output measurements during the testing and/or practical operating phases. Furthermore, an alternative re-designed current output-based observer is newly proposed in this paper, in order to develop a modified observer-based model predictive control (MPC) with input constraints to improve the performance of the unknown nonlinear time-delay stochastic system. When the proposed artificial system model is used together with the proposed constrained MPC, a long-time prediction of future input–output sets in a closed-loop setting can be carried out. Finally, the operation of a temperature controlled real nonlinear input time-delay blast furnace process is presented as a case study in this paper, to show the effectiveness of the proposed mechanism.