工业时滞过程的采样控制设计与批量优化

发布时间：2018-03-22 11:52

本文选题：采样控制系统　切入点：时滞　出处：《大连理工大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着数字控制系统的迅速发展,基于采样数据的计算机控制系统在实际工业中得到了越来越广泛的应用。本文基于实际工程中常用的采样系统离散域低阶时滞模型,分别对具有时滞响应的稳定型、积分型和不稳定型过程提出了离散域两自由度控制设计方法。采用近期文献提出的两自由度控制结构,解析地设计了设定点跟踪控制器和闭环系统抗扰控制器,可以分别对系统设定点跟踪响应和负载扰动响应进行独立调节和优化。根据H2最优控制性能指标设计设定点跟踪响应控制器,闭环系统抗扰控制器是通过提出期望的闭环系统互补灵敏度函数,反向推导而确定。该控制方案的突出优点是设定点跟踪响应的时域指标可以通过单调地调节控制器中唯一的调节参数定量整定。对于稳定型、积分型过程,根据期望闭环抗扰传递函数反向推导得出的抗扰控制器可以物理实现,然而对于不稳定过程,由于期望的抗扰控制器分子分母中存在隐含的不稳定零极点对消,使得闭环控制系统不能保证内部稳定性。为此本文采用有理Pade近似来逼近期望抗扰控制器形式,以便实际应用。关于所设计的控制系统稳定性,应用小增益定理分析得出在一些典型的模型不确定性情况下,闭环控制系统能够保持鲁棒稳定性的充要条件,并且利用图形化数值分析的方法给出抗扰控制器中单一的调节参数的整定范围。应用近期文献中的案例来验证说明本文中所提出的离散域两自由度控制设计方法的有效性和优越性。在上述两自由度控制设计方案的基础之上,提出了一种可以应用于带时滞批量生产过程的迭代学习控制方法。该控制方法的主要思想是利用历史批次中的信息来改善当前批次的控制性能,为此在上述两自由度控制结构中引入一个“存储器”,用来存储历史和当前批次中的过程输出、模型输出和控制信号,并且为控制当前生产过程周期提供前一批次中的过程输出、模型输出和控制信号,以便迭代学习控制和性能优化。在实际应用中,初始批次采用上述两自由度控制方案来保证控制系统的鲁棒稳定性,然后从下一批次开始采用迭代学习控制律,逐步实现完全跟踪期望设定点轨迹。此外,在所提出的迭代学习控制算法中引入一个滤波器来保证迭代误差的收敛性,并且给出误差收敛的充分条件。最后,通过应用一个仿真案例来说明该迭代学习控制方法的可行性和有效性。
[Abstract]:With the rapid development of digital control system, the computer control system based on sampling data has been more and more widely used in practical industry. Two degree of freedom control design methods in discrete domain are proposed for stable, integral and unstable processes with time-delay response. A set point tracking controller and a closed-loop disturbance rejection controller are designed analytically. The system set point tracking response and load disturbance response can be independently adjusted and optimized, and a set point tracking response controller is designed according to the H2 optimal control performance index. The closed-loop system disturbance rejection controller is based on the proposed complementary sensitivity function of the closed-loop system. The advantage of this control scheme is that the time-domain index of tracking response can be adjusted quantitatively by monotonously adjusting the unique adjusting parameters in the controller. The disturbance rejection controller derived from the inverse derivation of the expected closed-loop disturbance rejection transfer function can be physically realized. However, for the unstable process, because of the existence of implicit unstable zero pole cancellation in the molecular denominator of the desired disturbance rejection controller, The closed loop control system can not guarantee the internal stability. In this paper, the rational Pade approximation is used to approximate the desired disturbance rejection controller for practical application. By using the small gain theorem, the sufficient and necessary conditions for the closed-loop control system to maintain robust stability under some typical model uncertainties are obtained. The tuning range of a single adjustment parameter in the disturbance rejection controller is given by means of graphical numerical analysis, and the two degree of freedom control design method in discrete domain proposed in this paper is verified by a case study in recent literature. Based on the above two degrees of freedom control design, An iterative learning control method which can be applied to batch production process with time delay is proposed. The main idea of the control method is to improve the control performance of the current batch by using the information in the historical batch. For this reason, a "memory" is introduced into the above two-degree-of-freedom control structure, which is used to store the process output, model output and control signal in the history and current batch, and to provide the process output in the previous batch for the control of the current production process cycle. The model outputs and control signals for iterative learning control and performance optimization. In practical applications, the above two degree of freedom control scheme is used in the initial batch to ensure the robust stability of the control system. Then, the iterative learning control law is adopted from the next batch to realize the complete tracking of the desired set point trajectory step by step. In addition, a filter is introduced into the proposed iterative learning control algorithm to ensure the convergence of the iterative error. The sufficient conditions of error convergence are given. Finally, a simulation case is used to illustrate the feasibility and effectiveness of the iterative learning control method.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP273

【相似文献】