Design of Microelectronics Microsystem Neural Learning Controller Based on Subspace State Recognition

Abstract

Aiming at the large delay and large inertia in cone storage and irrigation, this paper adopts machine learning and predictive control, proposes a closed-loop N4SID (numerical subspace state system identification) algorithm, which is adjusted by IMC based on Smith predictor PI controller and parameter tuning are given, and the steps designed in cascade control are given. The experimental results verify the effectiveness of the proposed method.Compared with closed-loop SVA and open-loop methods, the method proposed in this paper has better verification results and pole assignment. According to this model, the PI controller is adjusted using IMC based on Smith Smith predictor. Compared with the traditional ZN adjustment method, this method shows good results in a conical system. In addition, the PI parameters can be adjusted to obtain robust control using the H∞-based mixed sensitivity algorithm. The contribution of this paper is the use of closed subspace system identification to identify the MIMO state space. The developed model validates the IMC tuned PI controller based on the Smith predictor in the factory.