A comprehensive decoupling control strategy for a gas flow facility based on active disturbance rejection generalized predictive control

Abstract

A comprehensive decoupling control (CDC) method, based on the active disturbance rejection generalized predictive control (ADRC-GPC) technology, is proposed in this paper for square multi-input multi-output (MIMO) systems. The central idea of this approach is that the total disturbance, which includes the partial coupling information among loops, the internal uncertainties, and the external disturbance, can be estimated by output through the extended state observer (ESO) and compensated in the feedback loop by feedback control law. Such estimating and cancelling can readily decouple the primal coupling system into multiple single-input single output (SISO) subsystems with the form of cascade integrators. Then the generalized predictive controller is designed for each subsystem. The proposed design is effectively used in the gas flow facility. Mathematical simulations and experiments show that compared with the proportional integral derivative (PID) method, the ADRC-GPC controller can achieve better dynamic performance, decoupling ability, disturbance rejection capability, and performance robustness.

Control por rechazo activo de perturbaciones: guía de diseño y aplicación

“…Tal es el caso del LADRC basado en error que se propone en (Madonski et al, 2019), el cual logra expresar todas las estructuras del lazo en unaúnica función de transferencia que relaciona la señal de control con el error de seguimiento. Otras contribuciones como (Wu et al, 2018) y (Cheng et al, 2019) han tomado el LADRC como base para la propuesta de nuevos algoritmos de control que combinan esta formulación con técnicas avanzadas como el control predictivo generalizado (GPC, del inglés Generalized Predictive Control) reduciendo así la dependencia de esteúltimo del modelo matemático de la planta.…”

Active design of dynamic GP models for model predictive control using expected improvement

“…Model predictive control (MPC) has attracted a lot of attention in the community of automatic control and is becoming increasingly popular in other industries as well. [1][2][3] The main reason is that the increase in computing power available for control has resulted in the development of control techniques, like dynamic matrix control (DMC) [4] and generalized predictive control (GPC), [5] and other methods. These methods heavily rely on models for computing not only the current effect of the implemented control but also the future effect.…”