模型预测控制系统时滞特性分析与输入输出故障处理
文献类型：学位论文
| 作者 | 杜德伟
|
| 学位类别 | 硕士 |
| 答辩日期 | 2016-05-25 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 导师 | 苑明哲 ; 邹涛 |
| 关键词 | DMC控制 时滞失配 扰动抑制 输入-输出故障 变结构控制 |
| 其他题名 | Time-delay Characteristic Analysis and Input-OutputFaults Treatment of Model Predictive Control System |
| 学位专业 | 计算机应用技术 |
| 中文摘要 | 随着模型预测控制越来越多地被应用到工程实践中去，人们对预测控制的研究也更加深入和全面。时滞对象的控制问题和控制系统故障处理方法一直以来都是研究的热点，本文对预测控制在时滞对象系统中的特性进行了分析，并深入研究了预测控制系统对输入-输出故障的处理方法。考虑到时滞问题在化工过程中的普遍存在以及传统控制方法在处理大时滞对象控制问题时复杂的设计和较难实现等问题，本文对预测控制在时滞对象控制中的特性进行了分析，并与史密斯预估算法相比较，分别从模型时滞失配、系统受不可测扰动影响等情形，总结出预测控制在解决大时滞对象控制问题中的优势，主要体现在当模型时滞失配时预测控制较强的鲁棒性和模型时滞失配及系统受不可测扰动影响时控制器参数的整定对控制效果的改进。在实际生产过程中，保证系统安全、稳定、高效地运行是非常关键的，对系统可能发生的故障进行检测、定位尤为重要。系统故障诊断技术的研究经历了从基于模型的方法到基于数据的方法这样一个发展过程，已经形成了完备的理论体系和丰富的技术积累，本文通过分析预测控制结构以及故障时控制器的特性，研究预测控制自身对输入-输出故障的处理能力，结论可以作为预测控制系统故障诊断技术的辅助方法。主要从预测控制算法的内模结构出发，分析了预测误差对输入-输出故障的“响应”，结合故障时预测控制算法目标函数对控制输入增量的优化结果，实现控制器自身对输入-输出故障的检测、分类以及定位。工业过程中许多关键的设备是具有冗余配置的，但也有许多环节是以设置旁路的方式进行生产运行。本文在明确输入-输出故障源的前提下，通过分析预测控制系统稳态输入与稳态输出的关系、方系统及非方系统控制求解问题，提出一种面向输入-输出故障的变结构预测控制方法，将故障通道的输入变量或输出变量从控制器结构中剔除，同时相应地改变控制器结构，针对输入和输出故障分别给出变结构下的控制求解方法，实现在不具有冗余配置的控制环节中的一种次优控制方式。 |
| 英文摘要 | As the model predictive control is increasingly applied to the engineering practice, the research on predictive control is more in-depth and comprehensive. TDO (Time Delay Object) control problem and the system fault processing method, has long been research hotspot. In this paper, we analyzed the predictive control performance in the object system with time-delay, and further studied the predictive control system with input-output fault processing method. The time delay problem is common in the chemical process. However, it is complex to design and difficult to implement for conventional control methods to deal with large TDO control. In this paper, we analyzed the predictive control in the delay object control characteristics, and compared them with SEM (Smith-Estimation method). We researched on the model mismatch problem and delay system problem, which influenced by unpredictable disturbances, respectively. We also summed up the advantages of predictive control to solve the problem of large TDO control. The advantages are mainly reflected in controller parameter setting and the improvement of control effect in predictive control, when the model in time-delay mismatch context TDO control has stronger robustness, besides the control system is affected by the unpredictable disturbance at the same time. In the actual production process, it is the most critical to ensure the operation system runs safely, stably and efficiently. It is also important to detect and locate the possible failure of the system. The research on system fault diagnosis technology has experienced a development process from the method based on model to the method based on data. It has formed a complete theoretical system and technology accumulation. In this paper, we researched on the predictive control of input-output itself fault handling ability, through the analysis of predictive control structure and the characteristics of controller in the fault state. We found that it could be used as an auxiliary method of fault diagnosis technology of the predictive control system. Based on internal model predictive control algorithm structure, we analyzed the prediction error of input-output fault "response". Combining with the optimization result of the control input increment optimized by the prediction control algorithm objective function in the fault state, we implemented the input-output fault detection, classification and location of the controller. In the process of industrial, many of the key equipment have a redundant configuration. However, many segments of the industrial process ensure the normal work by setting side roads. In this paper, on the premise of clearing and defining the input-output fault source, we proposed a variable structure predictive control method oriented by input-output fault, through the analysis of steady-state input predictive control system and the relationship between the steady output and the solution of the square and non-square control system. Accordingly, we removed the fault channels of input and output variables from the controller structure. We also changed the controller structure accordingly at the same time. Aiming at the malfunctions of the input and output, a control algorithm is given under the variable structure control. Therefore, we realized a suboptimal control method without redundant configuration in the controlling units. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 页码 | 68页 |
| 源URL | [http://ir.sia.cn/handle/173321/19658]  |
| 专题 | 沈阳自动化研究所_信息服务与智能控制技术研究室 |
| 推荐引用方式 GB/T 7714 | 杜德伟. 模型预测控制系统时滞特性分析与输入输出故障处理[D]. 中国科学院沈阳自动化研究所. 2016. |
入库方式： OAI收割
来源：沈阳自动化研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。