SD Based LADRC Decentralized Control For Multi-variable Systems

It is very common that there exist strong couplings in the thermal processes. Interaction among variables is a measure of the control difficulty. When plants have strong couplings, decoupling control is obviously a brilliant choice. However, for common decoupling methods, there always exist contradiction between the decoupler complexity and the control difficulty of the decoupled plant. Once the decoupling method is chosen, the decoupler and the plant are uniquely determined. The general simplified decoupling method(SD), which considers both the decoupler complexity and the control difficulty, can obtain general plant which is easy to control by choosing the decoupler structure. As a traditional method, it also depends heavily on the model precision. Moreover, the complete decoupling cannot be achieved due to the reason that precise model is hard to get in practice, which poses challenges into control of multivariable systems.As a new kind of control strategy which does not rely on model, active disturbance rejection control(ADRC) has strong adaptability and disturbance rejection performance, making it the perfect method for the incomplete decoupling problem of multivariable systems. By the extended state observer(ESO) of ADRC, couplings, model uncertainty and disturbance are conceived and estimated as a total disturbance, which is further compensated by the linear control law. Based on this idea, this paper proposes a control strategy which combines ADRC with SD. Firstly, SD decoupler is designed according to the model, after which ADRCs are designed for the diagonal elements of the general plant. As ADRC can estimate and compensate the total disturbance(incomplete decoupling, external or inner disturbance etc.), it is a proper way to deal with strong coupling, parameter perturbation and uncertainty.Typical thermal process —Circulating Fluidized Bed Boiler (CFBB),which has strong couplings and large time-delay, is discussed in the third part. Simulations of the SD+ADRC and some other methods show the advantages of the proposed method from two aspects(1. tracking 2. disturbance rejection).Then, part four studies the fuel processing system of the fuel cell. When gain scheduling control is utilized in the non-linear processing system, many controllers are needed, which increases the control cost. Part four applies the SD+ADRC into three linear models(with different levels of coupling) of fuel processing system at three operation conditions. Two strategies (3 decouplers+3 ADRCs and 1 decouplers+1 ADRCs) are proposed in order to prove the robustness of this strategy, which can provide new thinking for the control of the fuel processing system.Finally, the typical laboratory process—the quadruple-tank process(with couplings) is on the agenda. Simulations under parameter perturbations and valve perturbations respectively show the good control performance, disturbance rejection and robustness, which proves the feasibility of SD+ADRC.