工业预测控制的实用非线性处理技术研究
文献类型：学位论文
| 作者 | 孙浩杰
|
| 学位类别 | 硕士 |
| 答辩日期 | 2017-05-24 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 沈阳 |
| 导师 | 邹涛 |
| 关键词 | 模型预测控制 非线性系统 分程控制系统 Hammerstein-Wiener模型 双层结构预测控制 |
| 其他题名 | Research on Practical Nonlinear Processing Technology of Industrial Model Predictive Control |
| 学位专业 | 控制工程 |
| 中文摘要 | 模型预测控制算法自产生以来在工业中应用越来越广泛并不断显示出其优越性。工业过程中几乎所有的被控对象都不同程度地具有非线性特征，非线性存在是造成预测控制优化计算量大，预测结果不准确甚至导致控制策略失效的重要原因之一。本文针对线性模型预测控制技术应用于一类非线性被控对象时的处理方法进行研究，主要工作内容及取得成果如下： 1. 针对包含分程控制回路的控制系统具有非线性特征，提出一种适用于多变量模型预测控制的分程控制实现方法。在分程控制的各线性区间内根据各分程阀的不同动作组合建立预测模型，通过在线比较控制器输出信号值与相应定义的中间阈值选择模型，模型发生切换时保留切换前模型中的输出预测值作为当前模型下的初始输出预测值。解决了控制过程存在分程控制需求时，多变量预测控制器的设计问题，提高了线性预测控制器的适用范围。 2. 针对一类Hammerstein-Wiener模型描述的非线性控制系统，提出一种基于逆模型补偿的预测控制策略。在控制优化计算中，利用Wiener非线性环节的逆模型分别对系统输出设定值和采样值进行变换，使得标称模型下控制器输出与闭环系统中线性环节的输入相一致；在控制实施过程中，将控制器输出量通过Hammerstein静态非线性环节模型逆变换后施加到实际被控对象上。通过非线性变换补偿将非线性过程的控制转化为线性系统控制，避免了对非线性模型进行优化计算量大及预测不准确的问题，取得了预期控制效果。 3. 将所提包含分程回路多变量控制系统以及Hammerstein-Wiener非线性系统的线性模型预测控制策略分别拓展到双层结构模型预测控制理论框架下展开论述。对于包含分程回路的控制系统根据包含分程回路的情况在各线性区间内建立稳态模型；对于Hammerstein-Wiener非线性系统则建立其动态线性环节的稳态模型，对其动态线性环节进行优化求解。稳态目标计算层确定优化问题后，需首先对问题的可行性进行判定，不可行时进行软约束调整。并明确稳态优化层功能是对上层优化器的目标实现跟踪还是在可行域内进行自优化求解。计算得到稳态最优解后传至动态控制层进行跟踪控制。本节研究内容对所提策略在实际工业过程中的应用具有重要参考价值和指导意义。 |
| 英文摘要 | The model predictive control algorithm has been widely used in industrial applications and shown its superiority constantly. Almost all of the controlled object in the industrial process have different degrees of non-linear. Nonlinear existence is one of the important reasons for the large amount of calculation, inaccurate prediction results and even failure of control strategy. In this thesis, focus on the application of linear model predictive control technology in industrial process, study on the processing method of a class of nonlinear controlled object. The main contents and achievements are as follows: 1.Based on the control system contains split control loop always has nonlinear characteristics, a method of split control implementation is proposed for multi-variable model predictive control. In each linear interval of split control loop, the predictive models are established respectively according to different combination of the split valves. The control models were chosen by comparing the controller signal output values with the predefined middle threshold. After the control models were switched, the predicted output values of the old control model were regarded as the original output predicted values of the new model. The design problems of multi-variable model predictive controller are solved when the process is needed to be split control, and enhanced the applicability of linear MPC. 2. A strategy of linear model predictive control is proposed for a class of nonlinear block structure systems which described by Hammerstein-Wiener model. The element of dynamic linear model in the system is used to be prediction model of controller, using the inverse model of static nonlinear input link to compensate nonlinear characteristic of input link. In nominal system, the signal from the linear controller can be recovered exactly as the signal entering the linear element of the Hammerstein-Wiener system. Output samples signals are transformed by inverse model of static output nonlinear link, then be used to feedback correction. In the algorithm, rolling optimization needs only linear optimization and avoid inaccurate predictive for nonlinear model. Further introduces model building strategy of the system and the method of obtain inverse model. 3. Discuss the proposed linear MPC strategy in the framework of two layer model predictive control theory. For the control system including the split loop, the steady-state model is established in the linear interval according to the condition of the divided circuit; For the Hammerstein-Wiener nonlinear system, the steady-state model of its dynamic linear link is established as the optimization model of steady-state target calculation layer. It is necessary to determine the feasibility and the soft constraint adjustment after the optimization problem is established, make clear that the steady-state optimization layer function is to achieve the goal of the upper optimizer or to solve the optimization problem in the feasible domain. The steady-state optimal solution is calculated and then passed to the dynamic control layer for tracking control. The content of this study has important reference value and guiding significance for the application of the proposed method in the actual industrial process. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 源URL | [http://ir.sia.cn/handle/173321/20522]  |
| 专题 | 沈阳自动化研究所_数字工厂研究室 |
| 推荐引用方式 GB/T 7714 | 孙浩杰. 工业预测控制的实用非线性处理技术研究[D]. 沈阳. 中国科学院沈阳自动化研究所. 2017. |
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。