Research On Robust Tuning And Decoupling Control Scheme For Linear Multivariable Systems With Time Delays

With the rapid development of a variety of advanced manufacturing technologies, more and more manufacturing processes are constructed as multivariable control systems with high dimensions. Therefore, how to design the superior decoupling control strategies and tuning method has become one of most urgently needed modern process control technologies at present. The model control methods for the multivariable processes are most based on the state space method. But the design procedure of this methid is considerably complex and the order of the result controller tends to be quite high. Thus, they are difficult to be realized in the industrial processes owing to the obstacles in the technology and economy. As to the controller tuning, the usual schemes are numberial method or the trial and error method. These methods result in the heavey caculation load and are not convenient for tuning on-line. Therefore, the issue, how to build the relationship between the robust performance and the time-domain performace, develop a robust tuning scheme, and further enhance the decoupling capability are stringently required to be solved.This dissertation initiates research from the robust tuning scheme for the single parameter controller. Based on the robust stability conditions and the nominal performance of the system, some analytical calculation methods are proposed to determine the controller parameters. Then, in terms of the proposed tuning scheme and frequency-domain control theory, some decupling control methods are studied for the multivariable processes with time delays. The main contribution lies in: 1) For single-parameter diagonal controllers and full matrix controllers of the multivariable control system, a robust tuning scheme is proposed. Firstly, to guarantee the robust stability of the system, a calculation method is developed to derive the tuning bound of the controller parameters in the presence of the multiplicative and additive uncertainties. Secondly, an estimation method is proposed to determine the parameter values within the tuning bound. Finally, to make the system output performance satisfy time-domain design specifications, the dominant pole adjustment method and amplitude ratio adjustment method are utilized to optimize the determined controller parameters. The proposed tuning scheme constructs a bridge between the parameter values and robust performance and time domain specifications. Moreover, since the parameter tuning issue is transformed to the dominant pole or the amplitude ratio tuning problem, the number of the adjustable variables is reduced to one. The interactions of the control system are quantitatively included in the calculation methods. Therefore, compared with the trial and error method, the proposed tuning scheme is more accurate and time-saving.2) For linear stable multivariable processes with multiple time delays and right-half-plant (RHP) zeros, the inverted decoupling scheme is developed. This scheme mainly focuses on two-input-two-output (TITO) processes. The decoupler is designed by inserting additive terms such that the inverted decoupling technique accommodates a wider field than the one introduced in the published literatures. Due to the stability issue, some multivariable processes still cannot be decoupled by the inverted decoupling structure. To solve this problem, another modified decoupling scheme with unity feedback structure is suggested for implementation. Furthermore, in the presence of multiplicative and additive uncertainty, the sufficiency and necessary condition is analyzed to guarantee the robust stability of the system.3) For linear stable multivariable processes with time delays, one-degree-of-freedom and two-degree-of-freedom control schemes are presented respectively. Firstly, the proposed decoupling method for the TITO processes is developed for linear stable multivariable processes based on the one-degrees-of-freedom structure. Then, develop this method to the two-degree-of-freedom structure. Thus, the setpoint tracking or disturbance rejection of each control loop can be tuned conveniently by adjusting its corresponding single parameter. Simulation results demonstrate the effectiveness of the proposed control schemes for the different disturbance signals. In the presence of multiplicative uncertainty, the sufficiency and necessary condition is analyzed to guarantee the robust stability of the system.4) For the stable multivariable non-square systems with multiple time delays, a robust decoupling control method is developed by using the modified internal model control structure. In order to improve the robustness of the control system, a filter matrix is added to the IMC structure. The parameters of filter are determined according to the time delay of each control loop. In the presence of parameter uncertainty, the low bounds of the control parameters are derived quantitatively to guarantee the robust stability of the system. Compared with the traditional internal model control structure, the proposed method provides a wide robust stability region. For the nominal system, the performance of the control loop is only determined by its controller parameter and not affected by others.5) For decoupled processes with inverse response, a new analytical design method is proposed. Firstly, the RHP zeros are approximated to the form of the dead time in terms of Pade approximation theory. Then, based on IMC theory, an analytical PID controller design procedure is developed for the transformed inverse response processes. The resulting controller includes a single adjustable control parameter. By using the dual-locus diagram method, the stability region of the control parameter can be derived. Meanwhile, the sufficient and necessary condition for holding robust stability of the proposed closed-loop system is provided. By monotonously tuning the single controller parameter, the good compromise between the nominal performance and robust stability of the closed loop system can be realized. The proposed method simplifies the controller design procedure and avoids the disadvantage of the experience method. In this dissertation, the tuning and decoupling control schemes are studied for the linear stable multivariable systems with time-delays. According to the robust stable condition and time-domain specifications, this dissertation proposes a robust tuning scheme, and develops decoupling methods by using different control structures. This will provide theoretical guidance and effective techniques for the technology reconstruction, improvement of production process automation and economy of production cost.