【作者】 黄从智；

【导师】 白焰；

【作者基本信息】 华北电力大学（北京） ， 控制理论与控制工程， 2010， 博士

【摘要】 串级控制系统中,控制回路是通过实时网络闭合的,称之为网络化串级控制系统。网络化串级控制系统在实际工业过程控制中广泛存在,但由于系统结构的复杂性和网络性质的多样性,使得传统的串级控制理论和网络控制理论都无能为力,因而系统深入地研究网络化串级控制系统的分析与综合问题具有极其重要的理论价值和现实意义。本文基于工业过程控制实际,提出了网络化串级控制系统的概念,并将其归纳为四种典型结构,解决了其系统建模、稳定性分析、控制器设计、性能评估与优化、实验平台的设计与实现等问题,建立了一套完整、系统的关于网络化串级控制系统建模、分析与控制的理论体系。主要内容如下：首先,提出了网络化串级控制系统的四种典型结构形式,并采用提出的节点设备连接阵和网络传输阵以及系统配置图、方框图等三种方法,分别描述了这些不同结构的网络化串级控制系统。指出了网络化串级控制系统中存在的基本问题,如网络诱导时延、数据包丢失、时钟同步和量化误差等。其次,针对一类有不同性质网络诱导时延的网络化串级控制系统,在控制器分别都采用位置式PID和增量式PID的情况下,在离散时间域采用增广状态向量法建立了闭环系统的状态空间模型。针对通用结构形式的网络化串级控制系统,同时考虑网络诱导时延和丢包,提出了广义对象和广义控制器的概念,在离散时间域以状态空间表达式形式建立了统一的系统闭环模型。再次,针对两类有常数网络诱导时延的网络化串级控制系统,基于过程模型和期望闭环系统响应,采用一阶Pade近似纯迟延环节,以解析的形式提出了主控制器和副控制器PID整定参数的表达式。基于Lyapunov稳定性理论和LMI方法,针对一类有不确定网络诱导时延的网络化串级控制系统,在系统无扰动和有扰动时,分别设计了鲁棒H∞状态反馈和输出反馈控制律。基于μ综合理论,针对两类有不确定网络诱导时延的网络化串级控制系统,将不确定时延建模为乘性摄动,建立了μ综合控制框架,在副控制器为已整定好的传统PID时采用D-K迭代方法分别设计了μ综合主控制器。然后,以控制回路中无网络时网络化串级控制系统的稳态和动态性能指标为基准,以加权和形式提出了网络化串级控制系统的一个综合控制性能指标。主控制器采用模糊自整定PID控制器,副控制器仍采用传统的PID控制器,对两类网络化串级控制系统的性能进行了评估与优化。最后,设计并实现了一套基于基金会现场总线(FF)的网络化串级控制系统实验平台。设计了模糊PI在线自动优化程序,基于模糊自整定推理规则在线自动调整主控制器的PI参数,实现了储水罐水温的网络化串级控制系统。调试结果表明模糊自整定PID控制方法在网络化串级控制系统中实现的可行性和有效性。更多还原

【Abstract】 Cascade control systems, wherein the control loops are closed via real-time networks, are called Networked Cascade Control Systems (NCCSs for short). The NCCS has been widely employed in practical industrial process control, however, due to the complexity of its configurations and the variation of network characteristics, the theories of traditional cascade control system and networked control system are no longer valid. And therefore, it’s of great theoretical and practical meanings to investigate the analysis and synthesis of NCCS systematically and thoroughly.Based on practical industrial process control, the concept and typical configurations of NCCS are proposed. To develop an intact and systematic theory system for the analysis and synthesis of NCCS, the issues of modeling, stability analysis, controller designing, performance evaluation and optimization, as well as the design and implement of a pilot plant for an NCCS have been resolved. The main contents of this thesis are as follows:First of all, four typical configurations of an NCCS are proposed. The concepts of node-device connection matrix and network transmission matrix are proposed. With the proposed matrices, system configuration diagrams, and block diagrams, the four different configurations of NCCSs are described. The fundamental issues in an NCCS are pointed out, including network-induced delays, data packet dropout, clock synchronization, quantization error, and so on.Secondly, the closed-loop system models for a class of NCCSs with different characteristics of network-induced delays, in which the controllers are both positional PID or incremental PID, are developed in discrete time domain by state space method. The concepts of general process and general controller are proposed, and they are used to develop a uniform closed-loop system model for the NCCS with uniform configurations in discrete time domain by state space method. Both the network-induced delays and data packet dropout are considered.Thirdly, based on process models and desired closed-loop system responses, the analytical expressions of the PID tuning parameters in both of the primary and secondary controllers are proposed by approximating pure delay with first-order Pade expression for two typical configurations of NCCSs with constant network-induced delays. Based on Lyapunov stability theory and LMI approach, the robust H∞state feedback and output feedback control laws for a typical kind of NCCSs with and without disturbances are designed, in which the network-induced delays are uncertain but bounded. Based onμ-synthesis theory, the uncertain but bounded delay is modeled as a multiplied perturbation. Theμ-synthesis frameworks for two typical classes of NCCSs with uncertain but bounded network-induced delays are developed, and then the D-K iteration method is used to designμ-synthesis primary controllers for the system while the secondary controller is a PID controller and has been tuned beforehand.Then, based on the performance indices of stable and dynamical states of an NCCS without networks in the control loops, a comprehensive control performance index for an NCCS is proposed in the form of weighted summation. The performances for two kinds of NCCSs are evaluated and optimized, in which the fuzzy auto-tuning PID control method is employed in the primary controller while the secondary controller still adopts traditional PID control method.Finally, an NCCS pilot plant based on FF(Foundation Fieldbus) is designed and implemented. A fuzzy PI online automatic optimization program is designed, by which the PI tuning parameters of the primary controller can be automatically adjusted based on fuzzy auto-tuning inference rules. An NCCS for the tank water temperature is implemented and commissioned. The feasibility and effectiveness of fuzzy auto-tuning PID control method in an NCCS are validated by the practical commission results.更多还原