定值干扰下时滞系统固定结构控制器优化设计

发布时间：2018-02-10 23:19

本文关键词： 抗扰 PID控制非凸优化内模控制不确定系统数据驱动多变量系统分散控制部分解耦控制　出处：《北京化工大学》2016年博士论文　论文类型：学位论文

【摘要】：近年来,随着扰动概念的拓展,研究者越发重视对控制系统的抗扰性能进行研究。而在化工工业,控制系统的扰动抑制性能尤为重要。因此,研究抗扰控制是当前热点,也具有实际意义。本论文结合化工对象的特点,以抗扰为核心,针对单变量和多变量时滞系统,在存在定值扰动干扰下,研究化工工业广泛使用的固定结构控制器的优化设计,提出了满足抗扰需求的模型综合设计方法和基于数据驱动的性能优化方法。1.基于模型综合的PID和带扰动观测器的控制系统设计。内模控制是模型综合设计思想的典型代表(其它模型综合方法可以转化或者等价成内模原理)。本文深入研究了内模-PID的控制性能,提出了不同对象下满足扰动抑制性能的内模滤波器结构,并使用优化手段得到了最优的滤波器参数。类似地,论文还设计了带扰动观测器的控制系统。2.抗扰PID的直接优化。先进的PID设计方法将PID的整定转化成一个约束优化问题。一般地,将控制系统的输出性能指标(在时域上定义)作为目标函数,将系统的鲁棒性能指标(在频域上定义)作为约束条件。此类约束优化问题是一个非凸优化问题,是一类特殊的优化问题。为解此特殊的优化问题,本论文首先研究系统的稳定性,求出满足系统稳定性的参数空问,然后求出满足约束条件的参数空间,在两者的交集中寻找使得目标函数最优的参数。如此得到的最优解是全局最优解,同时提出的方法只需要对象的频率响应,不需要低阶近似模型,从而可用于高阶对象。论文将相关结果用到了不确定时滞对象的抗扰设计上3.基于数据驱动的带扰动观测器的控制系统设计。数据驱动控制器设计可以看成是直接基于性能指标的设计方法,此类方法通常是在模型未知的情况下直接优化某一性能指标。通过改变设定值,可以很容易得到以设定值跟踪为目标的性能指标值,从而优化控制系统,但是如果要优化以扰动抑制为目标的性能指标,则需要引入一个足够明显的扰动信号,这对系统通常是不利的,因而限制了其在抗扰方面的应用。鉴于此,本论文提出一种间接的数据驱动控制方法。具体而言,通过引入模型参数为中间变量,直接优化模型参数,而利用指定的函数关系确定控制器参数,达到间接优化控制器的目的。其中,该指定的函数关系根据扰动抑制性能提出,从而达到设计具有良好扰动抑制的基于数据驱动的控制系统的日的。4.多变量时滞对象的抗扰设计。最常见的多变量控制方案是分散控制,本论文首先考虑了在此类控制结构中,系统的扰动抑制问题。在分析外部扰动对系统输出影响的基础上,论文提出了一种基于非参数等价模型的分散控制器抗扰设计方法。除了分散式控制结构,另外一种典型的多变量系统控制方案是解耦控制,其目的在于消除回路之间的耦合。但是如果从抑制扰动的角度出发,可以发现在某些情况下,回路之间的耦合有利于扰动抑制。基于此,本论文提出了部分解耦的控制策略,其目的是消除不利于扰动抑制的耦合,而保留有利于扰动抑制的耦合,从而得到比完全解耦控制更优的扰动抑制性能。
[Abstract]:In recent years, with the development of the concept of disturbance, the researchers pay more attention to the research on anti disturbance performance of the control system. And in the chemical industry, the control system disturbance rejection performance is particularly important. Therefore, the anti disturbance control research is the current hot topic, also has great practical significance. This paper combined with the characteristics of chemical engineering, anti disturbance the core for single variable and multi variable time-delay systems in the presence of constant disturbances, optimal design of fixed structure controller is widely used in chemical industry research, put forward to meet the integrated design method of anti disturbance model of demand and performance optimization of.1. based on data driven control system design based on PID model and with disturbance observer the internal model control is a typical model of integrated design (other models can be transformed into a comprehensive method or equivalent internal model principle). This paper studies internal model -PID The control performance, put forward different objects to meet internal disturbance suppression performance of the filter structure, and use the optimization method get the optimal parameters of the filter. Similarly, the paper also designed the direct optimization of.2. control system with disturbance observer immunity. PID PID advanced design method of PID tuning into a the constrained optimization problem. In general, output performance index of control system (defined in the time domain) as the objective function, the robust performance index system (defined in the frequency domain) as constraints. The constrained optimization problem is a non convex optimization problem is a special class of optimization problems. For the optimization problem this special, this thesis studies the stability of the system, which satisfy the parameters of air system stability asked, and then calculated the parameter space which satisfies the constraints in both the intersection to find the target function of the And the optimal parameters. So the solution is the global optimal solution, the method proposed in this object need only the frequency response does not require low order approximate model, which can be used for high order objects. This paper will use the related results of uncertain anti disturbance design object with time delay is 3. based on disturbance observer data driven control system design. Data driven controller design can be regarded as a direct design method based on performance index, this method is usually the direct optimization of a performance index in the model is unknown. By changing the settings, can easily get to the setpoint tracking performance index of the target value, to optimize the control system, but if you want to in order to optimize the performance indexes of disturbance suppression goals, you need to introduce a sufficiently obvious disturbance signal, the system is usually negative, thus limiting its rejection in the party Application. In view of this, this thesis proposes a drive control method for indirect data. Specifically, through the introduction of model parameters as the intermediate variables, directly optimize the parameters of the model, and the function between the specified parameters of the controller, the controller to achieve indirect optimization. The function of the specified according to the disturbance rejection performance is proposed, which to achieve anti disturbance design design has good disturbance rejection control system based on data driven,.4. multi variable delay object. Multi variable control scheme is the most common decentralized control, this paper firstly considered in this control structure, the system disturbance rejection problem. Based on the effect of external disturbances on the system output analysis last, this paper puts forward a nonparametric equivalent model of decentralized controller anti-interference design method based on distributed control structure. In addition, another typical variable The control scheme is system decoupling control, its purpose is to eliminate the coupling between loops. But from the point of view of disturbance rejection, can be found in some cases, the coupling between the loops for disturbance rejection. Based on this, this paper puts forward control strategy of decoupling, its purpose is to eliminate the coupling disturbance rejection is not conducive to the reserved for coupling disturbance rejection, so as to obtain the complete decoupling control performance of disturbance rejection than better.

【学位授予单位】：北京化工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TP273

【参考文献】