Two-layer structure predictive control of target tracking mode for dual time-scale system

In the dual time-scale system, the time for the fast and slow model to reach steady-state values is inconsistent. Based on the "point" model strategy of integral variables, this paper compares and analyzes the slow model variables with the integral variables and designs a two-layer structure predictive controller for the dual time-scale system under the model truncation. The steady-state optimization layer uses the step response characteristics of slow model variables to derive the multi-step controlled variable prediction values after the truncation point. Based on the improved displacement matrix, the error correction coefficient is modified by introducing an error compensation factor in the dynamic control layer to reduce the accumulated error of model truncation. Under the truncated dual time-scale system model, the improved strategy proposed in the paper can not only maintain the amount of calculation of the predictive control but also solve the problem that the steady-state target calculation is not accurate enough due to model mismatch. Finally, the elfectiveness of the improvement strategy is verified by simulation comparison.