分数阶PIλ控制
- 格式:doc
- 大小:435.00 KB
- 文档页数:6
分数阶PIλDμ控制器设计及应用实验姜萍;王丽颖;马霄;孙凌燕【摘要】按照有限记忆数字实现法,利用RSLogix5000中结构化文本语言,设计了以CompactLogix控制器为核心的分数阶PIλDμ双容水箱液位控制实验平台,并完成了分数阶PIλDμ控制器在罗克韦尔PLC平台上的控制实验,解决了分数阶PIλDμ控制器的工程实现问题。
结果表明,分数阶PIλDμ控制器在惯性比较大的系统中具有很好的控制特性,结构化文本编程的方法符合IEC 61134国际标准,方便可行,适用性强,可以推广到实际生产过程中,具有一定的工程应用价值。
%The fractional order PIλDμ control experiment platform is used by the limited memory digital method to design a liquid level control of double liquid tank system of fractional order PIλDμ with compact Logix controller with the structured text language in RSLogix5000 ,the simulation and experiment of fractional order PIλDμ are carried on the Rockwell PLC platform ,it solves the engineering implementation issue of fractional order PIλ Dμ controller .The result shows that the fractional order PIλ Dμ controller has good control characteristics in big inertial system .Indeed the method of the structured text programming accords with the international standard IEC 61131 ,which is convenient and practical ,and has fully illustrated that the fractional order PIλDμ controller has a certain engineering application value and can be applied to the actual production .【期刊名称】《实验技术与管理》【年(卷),期】2015(000)008【总页数】4页(P87-90)【关键词】分数阶PIλDμ控制器;有限记忆法;结构化文本;RSLogix 5000【作者】姜萍;王丽颖;马霄;孙凌燕【作者单位】河北大学电子信息工程学院,河北保定 071002; 河北大学罗克韦尔自动化实验室,河北保定 071002;河北大学电子信息工程学院,河北保定 071002;河北大学电子信息工程学院,河北保定 071002;河北大学电子信息工程学院,河北保定 071002【正文语种】中文【中图分类】TP273分数阶微积分理论建立至今已经有300 多年的历史[1],但早期主要侧重于理论研究。
分数阶参数不确定系统的PIλ控制器梁涛年;陈建军【期刊名称】《控制理论与应用》【年(卷),期】2011(028)003【摘要】利用求解分数阶参数不确定系统稳定域的方法,设计了使分数阶参数不确定系统具有鲁棒性的分数阶PIλ控制器.首先采用Kharitonov理论,将分数阶参数不确定系统分解成若干个参数确定的子系统,然后用D分解方法分别求出在PIλ控制器的控制下,使各个子系统都取得较大稳定域的参数入值.再采用此λ值构建PIλ控制器并计算各个子系统的稳定域.各个子系统稳定域的交集即为参数不确定系统在PIλ控制器控制下的稳定域.同时证明了所构建的PIλ控制器能稳定整个参数不确定系统组.最后在稳定域内取控制器参数值,便构成了所设计的PIλ控制器.文中采用实例对此设计方法进行验证,并用所构建的PIλ控制器对参数不确定系统组的各个子系统进行阶跃响应分析,结果表明PIλ控制器对参数不确定系统具有较强的鲁棒性.%The paper presents a method for designing the robust fractional order PIλ controller by computing the stability region of the fractional order system with uncertain parameter.Firstly, the Kharitonov theorem is adopted to decompose the original fractional order system with uncertain parameters into several subsystems with parametercertainties.Secondly,the D-decomposition technique is applied to compute the stability region o f each subsystem to determine the parameter λ value which uniformly ensure a bigger stability region for all subsystem.Thirdly, with the parameter λ value, we design a fractional order PIλ controller foreach subsystem and computer its stability region.The intersection of the obtained stability regions is considered the stability region of the original system under the control of the designed PIλ controller.This paper proves that the designed PIλ controller stabilizes the original fractional order system wi th uncertain parameters.Finally, the fractional order PIλ controller is constructed based on the control parameters in the stability region.The proposed method is illustrated by an example.The step response of each subsystem is analyzed when using this PIλ controller.The result shows that fractional order PIλ controller has stronger robustness for the fractional order system with uncertain parameters.【总页数】7页(P400-406)【作者】梁涛年;陈建军【作者单位】西安电子科技大学,机电工程学院,陕西,西安,710071;西安电子科技大学,机电工程学院,陕西,西安,710071【正文语种】中文【中图分类】TP273【相关文献】1.分数阶PIλDμ控制器参数λ和μ分析 [J], 梁涛年;陈建军;尚保卫;王妍2.分数阶PIλDμ控制器数字实现与参数优化 [J], 郑翠;赵慧;蒋林;李苑3.基于敏感传递函数的分数阶PIλ控制器的参数整定 [J], 杨征颖;王德进;史万祺4.基于BP神经网络的分数阶PIαDβ控制器参数整定研究 [J], 那景童;徐驰;5.基于人工鱼群算法的分数阶PIλ控制器参数整定 [J], 张学典;王富彦;秦晓飞因版权原因,仅展示原文概要,查看原文内容请购买。
分数阶PIλDμ控制器的参数对系统性能的影响严慧于盛林李远禄(南京航空航天大学自动化学院,南京市210016)摘要:分数阶PIλDμ控制器比传统整数阶PID器多了两个可调参数,微分阶数μ与积分阶数λ,所以它的设计更加灵活,应用更加广泛。
分数阶PIλDμ控制器中的三个参数Kp,Ki和Kd与整数阶PID器的参数相同,它们在两个控制器中的作用也相同;分数阶PIλDμ控制器中的参数μ,λ的大小分别决定了控制器微分作用与积分作用的强弱。
本文研究了分数阶PIλDμ控制器每个参数变化对分数阶系统的影响,仿真结果表明,参数μ与λ分别主要影响系统的超调和影响系统的稳态精度。
关键字:分数阶PIλDμ控制器;参数变化;分数阶控制系统Influence of F ractional-order PIλDμController’s Parameters onSystem PerformanceYa,Yu Sheng-ling,Li Yuan-lu(Automation College in NanJing University of Aeronautics and Astronautics, NanJing 210016) Abstract: Fractional-order PIλDμcontroller has two more adjustable parameters, differential order μand integral order λ, so it can be designed flexibly can be applied widely. The three parameters, Kp,Ki and Kd of fractional-order PIλDμcontroller are same as the parameters of integer-order PID controller, and the three parameters do the same effect in both controllers. The value of another parameters, μand λof fractional-order PIλDμcontroller decide the effect of differential and integral. In this paper, the influence of the changes of fractional-order PIλDμcontroller’s parameters and fractional-order controlled system’s parameters on performances of fractional-order control system is researched. The simulation outcomes verify that, parametersμandλaffects the over regulation of the system and the steady precision respectively.Key words: Fractional-order PIλDμcontroller; Parameters changes; Fractional-order control systems1. 引言在现代的工业控制中,PID控制由于其控制结构简单、参数易于整定、鲁棒性强等优点,一直在占据着主导地位。
Buck电路的分数阶建模与PIλDμ控制作者:方数丞王晓刚来源:《机电信息》2020年第02期摘要:基于分数阶微积分理论与实际中电感与电容的外特性呈分数阶的事实,运用状态空间平均法建立了在电感电流连续情况下的分数阶Buck电路的数学模型和电路模型,提出了分数阶Buck电路纹波分析与连续条件,推导出占空比至输出电压的传递函数和输入电压至输出电压的传递函数。
此模型较整数阶模型更能精确反映实际电路工作状态。
基于Matlab/Simulink软件对模型进行了仿真,验证了该模型的正确性。
基于ITAE最优控制方法设计了分数阶PID控制器对该模型进行控制,并对补偿后的传递函数进行了仿真,验证了该控制器的有效性。
关键词:分数阶微积分;Buck变换器;建模;分数阶PID控制0 引言自从1695年Leibniz在给L′Hospital的书信中第一次提出关于将微分阶次从整数阶推广到非整数阶的含义的问题,再到由Leibniz所提出的问题开创了一门持续发展了300多年的关于分数阶微积分的学说。
直至1960年开始,分数阶微积分学逐步推广到科学与工程领域,大量学者做出了杰出贡献。
其中,意大利学者Caputo与Mainardi教授提出了基于分数阶导数建立的耗散问题[1];斯洛伐克学者Podlubny教授提出了分数阶比例-积分-微分控制器的模型[2];法国学者Oustaloup教授的研究组提出了分数阶鲁棒控制理论,并将其成功应用于汽车工业的悬挂控制。
近年来,分数阶微积分的应用越来越受到各工程学科的关注,电气工程领域也不例外[3-4],一方面,在传统电路中引入分数阶元件可以使电路设计变得更加自由和灵活[5-7];另一方面,某些电气元件的分数阶模型可能取代目前使用的常规模型。
张波教授在文献[8]中提出了一种buck-boost电路的分数阶建模方法。
文献[9]、文献[10]分别建立了电感电流连续模式和电感电流伪连续模式下boost变换器的分数阶模型。
基于相角裕度与鲁棒性解析法设计分数阶PIλ控制器高学利;吕广芝【摘要】介绍了一种基于期望相角裕度与提高系统鲁棒性的解析法设计分数阶PIλ控制器.首先,根据相角裕度的定义和提高系统的鲁棒性得到3个非线性方程组,并用图解法得到PIλ控制器的积分阶次λ;然后,由解析法得到控制器的比例增益kp和积分时间常数TI.比较传统Z-N法设计的常规PI控制器,仿真表明解析法得到分数阶PIλ控制器能得到期望的裕度并具有更好的鲁棒性.【期刊名称】《化工自动化及仪表》【年(卷),期】2014(041)003【总页数】4页(P243-245,343)【关键词】分数阶PIλ;控制器;相角裕度;解析【作者】高学利;吕广芝【作者单位】烟台万华化工设计院有限公司,山东烟台264006;烟台万华化工设计院有限公司,山东烟台264006【正文语种】中文【中图分类】TH86随着现代工业的发展和工业过程控制精度的不断提高,对工业过程数学模型的建立提出了更高的要求,传统的整数阶模型不能达到精度要求,因为实际的工业过程控制中有许多系统都是分数阶系统。
随着分数阶微积分理论的发展[1],人们建立分数阶微分方程来描述实际分数阶系统,这样可以使模型更加精确。
由于PID控制器具有结构简单、鲁棒性强及易于操作等特点,被广泛地应用于工业过程控制中。
对分数阶被控系统,用传统整数阶PID控制器来控制往往达不到理想的控制效果。
当系统参数变化时,整个闭环系统控制效果变差甚至会不稳定,所以,针对分数阶被控系统,科学家们提出了分数阶控制器。
由于分数阶PID控制器具有变量多、自由度高的特点,这给研究人员提供了更大的想象空间,并进行着不同方向的研究[2,3]。
目前,分数阶控制理论已在工业过程控制中,特别是在冶金、化工、电力、轻工及机械等行业得到了广泛应用。
笔者将介绍一种基于相角裕度与鲁棒性解析法设计的分数阶PIλ控制器。
通过仿真对比常规PI控制器,表明设计的分数阶PIλ控制器具有更好的鲁棒性。
PID 控制器是工业上应用最广泛的控制器之一,它在控制整数阶被控对象时能取得很好的控制效果;然而,对于一些复杂的实际系统,用分数阶微积分建模比整数阶模型更为精确,为了得到更好的控制效果,将控制器的阶次扩展到分数阶得到PI λD μ控制器模型。
本文对包括PI λD μ控制器积分阶次λ、微分阶次μ在内的5个参数,提出了一种基于遗传算法整定分数阶PID 控制器参数的方法,仿真结果表明,对于分数阶系统,采用PI λD μ控制器会取得比常规PID 控制器更好的控制效果,并验证了本方法的有效性。
PI λD μ控制器比常规PID 控制器多了两个可调参数积分阶次λ和微分阶次μ,控制器参数的整定范围变大,控制器能够更灵活的控制受控对象,但是控制器参数的增多也使得参数的整定变得困难,控制器参数的好坏将直接影响着控制效果。
我们给出了一种基于遗传算法直接整定PI λD μ控制器5个参数的方法,并对分数阶控制器和整数阶控制器对同一被控对象的控制效果进行了比较,最后给出了一个实际系统的分数阶模型,通过仿真,对比了本文方法和其他参数整定方法,给出相应结论。
分数阶系统是用分数阶数学模型能更好描述的一类系统。
为了区别整数阶模型,分别用fc G 和ic G 表示PI λD μ控制器和常规PID 控制器,Gf 和Gi 表示分数阶被控对象和整数阶被控对象。
分数阶控制器传递函数,)(s G fc 的表达式如下:μλs K s K K s G d i P fc ++=-)(其中,积分阶次λ、微分阶次μ都大于0,对比于常规的PID 控制器s K s K K s G d i p ic ++=-1)(可以看出,PI λD μ控制器多了两个可调参数,当积分阶次λ、微分阶次μ都取1时,PI λD μ控制器即为常规PID 控制器,可见常规PID 控制器是PI λD μ控制器的特殊形式。
根据式(6)可以得到分数阶控制系统单位反馈结构图如图1所示 分数阶积分K is -λ+-E(s)Y(s)Gf(s)比例Kp 分数阶微分K d s μR(s)Gfc(s)图1 单位负反馈分数阶闭环控制系统结构图从图1中可以得到,分数阶闭环系统的传递函数)()(1)()()()()(s G s G s G s G s R s Y s G fc f fc f s +== 分数阶系统的时域分析考虑一类简单的分数阶微分方程)()()()()(121121t u t y D a t y D a t y D a t y D a n n t n t n t t =++++--αααα其中,u(t)为某已知函数,假设输出信号y(t)及其各阶导数的初始均为0,则可以由Laplace 变换写出系统传递函数模型n n sa s a s a s a s G n n αααα++++=--1211211)(本文采用Grunwald-Letnikov 分数阶微积分定义,可以得到y(t)的每个阶次的微分如下:⎥⎥⎥⎦⎤⎢⎢⎢⎣⎡-+=-≈∑∑-=--=-)()()()(][1)(][0)(jh t y t y h jh t y h t y D h a t j j h a t j jt a i i i i i αααααωω 将上式带入方程中(8)可以写出分数阶微分方程的数值解为⎥⎥⎥⎦⎤⎢⎢⎢⎣⎡--=∑∑∑-===)()(1)(][1)(11jh t y h a t u ha t y h a t j j n i i n i i i i i αααω 应用上述算法就可以求得任意输入的分数阶系统的数值解,编写了一个step ()函数来求解一般微分方程的单位阶跃响应曲线。
Abstract —This article provides a graphical parameter tuningmethod of λPI controllers for fractional-order time-delay systems. First, the complete stabilizing region of λPI controllers in proportional- integral plane, for a fixed λ, is determined in terms of a graphical stability criterion applicable to time-delay systems. Then, the stabilizing region is maximized analytically with respect to parameter λ to expect the most various behaviors of the closed-loop systems. Finally, by defining appropriate functions relative to the requirements of gain and phase margins, the curves in the maximized stabilizing region satisfying the pre-specified gain and phase margins are drawn, which releases a flexible parameter tuning procedure. Numerical examples are given to illustrate the design steps.Keywords —Fractional order systems, λPI controllers, Time-delay systems, Graphical stability criterionI.INTRODUCTIONHE PID controller is the most commonly used control tool in industrial processes and is a very important topic in both academic research and control engineering applications [1-3]. The reason of this wide range of interest is its relatively simple structure that can be easily understood and implemented and its control ability in many practical processes. The conventional research efforts towards PID controllers was focuses mainly on the tuning rules of PID parameters, e.g., the well known Ziegler- Nicholes rules [4] for the processes with the S-shape reaction curve. In recent years, considerable attention has been paid to the so-called fractional-order systems whose models (the plants and/or the controllers) are described by fractional-order differential equations, i.e., equations involving noninteger-order derivatives [5]. This is due to the fact that many real control systems are well characterized by such equations. The frequency response and the transient response of the noninteger integral and its application to control systems were first introduced in [6]. Frequency analysis was also applied to fractional-order PID-like controllers, i.e., in the TID (tilt-integral-derivative) scheme proposed in [7], where the proportional compensating unit of a classical PID algorithm was replaced by an element referred to as a “tilt ” compensator. The synthesis approach to the CRONE control introduced in [8-9] pursues the “fractal robustness ” on the basis of a desired frequency template [10]. The conventional PID controller was extended to the fractional-order form of μλD PI both in the time domain [11] and in the frequency domain [12,5]. For time-delay systems, the design methods forλPI andμλD PI controllers have attracted more interests in recentThis work is supported by the National Nature Science Foundation of PR China under Grant 60874028.research [13-16]. In [13], taking advantage of the additional freedom λ provided by λPI controller, one more specification than in the case of conventional PI controller has achieved via a nonlinear function minimization subject to some given nonlinear constraints, improving the performance of the system and making it more robust to plant uncertainties (gain and time-constant changes). The robustness of λPI controller offered by fractional-order λ was further studied and confirmed in [14] by solving a nonlinear function minimization problem also. The results of D-decomposition method applied to the parameter space design of fixed structure controllers for the integer-order systems were generalized to the case of fractional-order μλD PI controllers [15-16]. The stability regions were derived analytically in different parameter space selected from five parameters of μλD PI controllers.In this article, a graphical stability criterion applicable to complex functions with delay factor is employed to investigate the stabilizing regions in the parameter space, i.e., in ),(i p k k -plane, of λPI controller for a fixed value of λ. By an analytical optimization procedure, the optimal value of λ corresponding to the maximum area of stabilizing region is computed. In this maximum stabilizing region, the specifications of phase and gain margins of fractional-order systems with time-delay is considered, which gives a quite flexible tuning method for λPI controllers.II. M AIN R ESULTSA. Stability CriterionConsider SISO unity feedback system shown in Fig.1, where )(s G is the transfer function of the plants with time-delayse s D s N s G θ-=)()()( (1) where D(s)andN(s)are integer order or fractional-orderpolynomials of complex variable s , 0>θis the time-delay, C(s)is the λPI controllerFig. 1.Unity feedback systemλsk k s C ip +=)( (2)A Graphical Tuning of λPI Controllers for Time-Delay SystemsDe-Jin Wang, and Jiang-Hui ZhangSchool of Electronic Information and Automation, Tianjin University of Science and Technology, Tianjin, 300222, PR China email: wdejin56@Twhere p k and i k are the proportional-gain and integral-gain, respectively,20<<λdenotes the fractional-order of the integrator. Taking 1=λ, the classical PI controller is recovered. The closed-loop fractional-order characteristic polynomial (quasi-polynomial, see [5]) is given by si p es N k s k s D s s θλλδ-++=)()()()(Multiplying both sides of the above equation by se θyields )()()()(*s N k s k e s D s s i p s++=λθλδ (3) The objective of this subsection is to determine thestabilizing region, based on a graphical stability criterion, in ),(i p k k -plane for a fixed )2,0(∈λ.To this end, substituting jw s = into (3), i.e. along the imaginary axis in s-plane, we get)(])([)()()(*jw N k jw k e jw D jw jw i p w j ++=λθλδ (4) Noting that2sin2cosλπλπλj j += (5) )()()(w jD w D jw D i r += (6) )()()(w jN w N jw N i r += (7) and partitioning )(*jw δ into its real and imaginarycomponents yields )()()(*w j w jw i r δδδ+=where)()2cos()()()2cos()()()(2222ωλπβωωωλπαθωωωωωδλλr i i r p i r r N k N N k D D +++++++= (8) )()2sin()()()2sin()()()(2222ωλπβωωωλπαθωωωωωδλλi i i r p i r i N k N N k D D +++++++= (9) where αdenotes the phase of the complex function (6), β the phase of (7), respectively.From (8) and (9), it is clear that both r δand i δdepend on parameters ),,,(ωλi p k k . Suppose that we have found, in one way or another, a point ),,,(0ωλi p k k on the imaginary axis such that⎪⎩⎪⎨⎧====0),,,(0),,,(0000ωλδδωλδδi p i i i p r r k k k k (10)i.e. there is a root on the imaginary axis in s-plane. Then, according to the Implicit Function Theorem, if the Jacobi Matrix),,,(00ωλδδδδi p k k i i p i i r pr k k k kJ ⎥⎥⎥⎥⎦⎤⎢⎢⎢⎢⎣⎡∂∂∂∂∂∂∂∂= (11)is nonsingular, the equations (10) has a unique local solution curve ))(),((ωωi p k k . Moreover, we have the following proposition [17].Proposition 1 The critical roots are in the right-half plane in the parameter space to the left of the curve ))(),((ωωi p k k when we follow this curve in the direction of increasing ω, whenever det 0<J and to the right when det 0>J . Here J is the Jacobi Matrix defined in (11).Solving equations (10) for p k and i k in terms of λ andω yields2sin)()()2sin()()(2222λπωωλπβαθωωωi r i r p N N D D k ++-++-= (12)2sin)()()sin()()(2222λπωωβαθωωωωλi r i r i N N D D k +-++= (13) On the other hand, it follows, from (8), (9) and (11), that20,0,0))()((2sindet 22<<>∀<+-=λωωωλπωλi r N N J (14)As the coefficients of the characteristic quasi-polynomial (4)and θare real, we have that if ωj is a root of (4), then, so toothe complex conjugate of it. Therefore, it is sufficient toconsider ),0[+∞∈ω, which corresponds to two cases:0=ωand ),0(+∞∈ω.When ),0(+∞∈ω, (12) and (13) determines a part of thestabilizing boundary line in ),(i p k k -plane, and fromProposition 1, we can identify which side of the line belongsto the stabilizing region as shown in the following examples. On the other hand, if the numerator )(s N of the transfer function )(s G in (1) contains a constant term, when 0=ω,0)(*=ωδ leads to0=i k (15) which gives another part of the stabilizing boundary line. Example 1 Integer- order plant controlled by fractional- order controller. Consider the following integer first-order plant with time-delay s e s s G 4.01105)(-+=For this plant, we determine the stabilizing parameters of thefractional-orderλPI controller (2). For fixed ,2.0=λ1=λ and 6.1=λ, respectively, from (12) and (13), the stabilizing boundary curves in ),(i p k k -plane can be drawnfor sufficiently large ω as shown in Fig.2. In the frequency range ),(10ωω, 00=ω, according to the sign of detJgiven by (14) and Proposition 1, the right side of each curve, as ω increases, belongs to stabilizing regions. Thus, the initial pieces of these curves in the frequency range ),(10ωω, together with the straight line (15), constitute the stabilizing regions. From Fig.2, it is clear that different values of λcorrespond to different shapes and areas of the stabilizing regions.kp k iFig. 2 Stabilizing boundary curves for different λExample 2 Fractional-order plant controlled by fractional- order controllers. Consider the following fractional-order plant with time-delays e a s s G 1.0221)(10)(-+=where a is a real number. Fixing 2.0=λ, we discuss theshapes of the stabilizing regions for different values of a . Similar to the procedure of Example 1, the stabilizing region in ),(i p k k -plane are plotted in Fig.3 for 5.0=a , 0=a , and 1-=a , respectively. It is observed that as a decreases, the stabilizing region shrinks. B. Optimization of λIt is expected that the bigger stabilizing region canλ=0.2kpk iFig. 3 Stabilizing region for different values of aprovides more various behaviors of control system, and achieves better response performances. From the above two examples, it is observed that the stabilizing regions have “sector” shapes which enables us to find the maximum stabilizing region by selecting an appropriate value of λ as discussed in the following.From the discussion of the above subsection, the parameter λof λPI controller affects the shape and area of the stabilizing region.We first note that when 0=i k , the stabilizing boundary curve intersects with p k axis, and the correspondingintersection frequency is1ω see Fig. 2. By letting0)(=ωi k in (13), we obtain that 1ωis the solution of theequation0)sin(=-+βαθωwhich is independent of λ.This fact leads us to utilize the Leibnitz Sector Formula to calculate the area of the stabilizing region,ωωωωωλωωd k k k k P i p i p ⎰'-'=10)]()()()([21)( (16)and the optimum value of λ corresponding to the biggeststabilizing region is determined by the solution of the equation0)(=λλd dP (17) Example 3 Recall Example 1. In this case, we have 00=ωand =1ω 3.99. By using the approximate integration, the relation curve between )(λP and λ is plotted in Fig. 4, from which it is seen that we have two optimum values of λ, i.e., 0→λ and 6.1=λapproximately. Choosing 2.0=λ and 6.1=λ, the stabilizing regions are shown in Fig. 2, which are bigger than the conventional case, 1=λ.λP (λ)Fig.4 Relation curve between )(λP and λIII.PERFORMANCE DESIGNOnce the stabilizing regions are known, we can further consider the synthesis of λPI controllers with given gain margin (GM) and phase margin (PM), i.e., the objective is to make the open-loop transfer function,)()(s G s C , meet the followingmj p p p i p p h e j D j N j k j k p 1)()()()(-=⋅+-θωλλωωωω (18) for the case of GM, andm gj j g g g i g p e e j D j N j k j k φθωλλωωωω-=⋅+-)()()()( (19) for the case of PM, respectively, where m h and m φ are the desired GM and PM, respectively, p ω and g ωare the phase and gain crossover frequencies of the open-loop system,respectively. For satisfactory response performance, the gain margin should be greater than 2, and the phase margin should be between30to60. In the sequel, let us only consider the case of phase margin since it is closely related to overshoot. (19) Leads us to define the following complex function of ω0)()()()()(=+⋅+=-m j j i p e e j D j N j k j k F φθωλλφωωωωω (20) from which, it follows that0)(])([)()()(=+++ωωωωλφθωλj N k j k e j D j i p j m (21)By comparing (4) with (21), it is exhibited that thedifference between (4) and (21) is the factor m φ in (21). Hence, for a given PM=m φ, the stabilizing analysis strategy in the previous section can be extended to the case of (21). First, from (21), along the same line as solving the solution (12) and (13), we have2sin)()()2sin()()(2222λπωωλπβαφθωωωi r m i r p N N D D k ++-+++-= (22)2sin)()()sin()()(2222λπωωβαφθωωωωλi r m i r iN N D D k +-+++= (23) which determines the achievable PM curve in the stabilizing region.Then, based on the above discussion, the following algorithm is developed on the achievement of the given PM for λPI controllers. Algorithm 1 For achieving the given PM of λPI controllers Step 1 Select an optimum value of λ in the range 20<<λ, using the formula (16) as shown in Fig.4. Step 2 For this λ, plot the curve ))(),((ωωi p k k using (12) and (13) for big enough ω, and identify the stabilizing region according to the sign of 0det <J given by (14) based on Proposition 1.Step 3 Given a desired m φ, draw the achievable PM curve given by (22) and (23) for big enough ω.Step 4 Choose a gain crossover frequency g ω along this curve in the stabilizing region according to the desired band width of closed-loop system, and compute the corresponding point ))(),((g i g p k k ωω.In the case of achieving the given GM of λPI controller, the design procedure is similar to Algorithm 1.If one wants to achieve the desired PM and GM simultaneously, the two curves for achieving the PM and GM can be drawn on the same graph, and the intersection of the two curves is the solution. Note that this intersection of the two curves may not exist in stabilizing region for bigger GM and/or PM.Example 4 Recall Example 1 again. First, selecting 2.0=λ, which corresponds to a relatively bigger stabilizing region (see Fig.2), and setting the desired PM to45, we draw the corresponding achievable PM curve in the stabilizing region using Algorithm 1, as shown in Fig.5 (a) (the inner solid line) . Next, for comparisons, choosing 1=λ, corresponding to the conventional PI controller, and PM=45, the achievable PM curve is plotted in the same figure (the inner dashed line). Clearly, the two PM curves have an intersection point A. The coordinates of point A reads 82.2=p k and 14.1=i k approximately, and the corresponding gain crossover frequencies are 89.1=g ω for2.0=λ and 46.1=g ω for 1=λ, respectively. Underthis set of parameters of λPI controller, the step responses are shown in Fig.5 (b) for the two cases. Although the given PM are the same (45), the overshoot for 2.0=λis lessthan that for 1=λ, and the response for 2.0=λ has a faster rise time than that for 1=λ, because the former has a bigger gain crossover frequency. Finally, it is observed, from Fig. 4, that the stabilizing region has another local maximum point at 6.1=λ. We compare the case of 2.0=λ with the case of 6.1=λ. Setting PM=45, the two PM curves have an intersection point B as shown in Fig. 6 (a) and the parameters at point B read 72.2=p k and 22.1=i k ,88.1=g ω for 2.0=λ, 86,0=g ω for 6.1=λ. Thestep responses are plotted in Fig. 6 (b), which shows that although 6.1=λ is a local maximum point of stabilizing region, the step response is worse than those for 2.0=λ and 1=λ.To summarize, a smaller value of λ can give a bigger stabilizing region and a better system performance.kpk i(a) Stabilizing regions and PM curvesStep ResponseTime (sec)A m p l i t u d e(b) Step responsesFig. 5 Comparisons between 2.0=λ and 1=λ.IV.C ONCLUSIONSIn this article, a graphical tuning method of λPI controllers for time-delay systems has been proposed. The maximum stabilizing region with respect to λ has been obtained analytically. In this region, the specifications ofachievable gain and phase margins have been discussed and the comparisons for different value of λ have been made bothin parameter space and in time domain.kpk i(a) Stabilizing region and PM curvesStep ResponseTime (sec)A m p l i t u d e(b) Step responsesFig. 6 Comparisons between 2.0=λ and 6.1=λ.R EFERENCES[1] K. Astrom, and T. Hagglund, PID controllers: Theory, Design, and Tuning. Research Triangle Park, NC: Instrument Society of American, 1995.[2] G.J. Silva, A. Datta, and S.P. Bhattacharyya, PID Controllers for Time- Delay Systems. Berlin, Germany: Birkhauser, 2004.[3] K. Astrom, and T.Hagglund, Advanced PID Control. Research T. Park, NC:ISA, 2005[4] J. G. Ziegler, and N. B. Nichols, “ Optimum setting for automatic controller,” Trans. ASME, Vol.64, pp. 759-768, 1942.[5] I. Podlubny, Fractional Differential Equations. London. UK, Academic Press, 1999[6] S. Manable, “The non- integer integral and its application to control systems ,” Electrotechnical Journal of Japan, Vol. 6, No. 3/4, pp. 83-87, 1961[7] B. J. Lurie, Three- parameter tunable tilt- integral- derivative (TID) controller. Us patent Us5371670, 1994[8] A. Oustaloup, B. Mathieu, and P. Lanusse, “The CRONE control of resonant plants: application to a flexible transmission, ” Euro. J. Control, Vol. 1, , pp. 113-121, Feb.., 1995[9] A. Oustaloup, X. Moreau, and M. Nouiliant, “ The CRONE suspension ,” Control Eng. Prac., Vol. 4, pp. 1101-1108, Aug., 1996 [10] A. Oustaloup, J. Sabatier, and P. Lanusse, “From fractal robustness toCRONE control ,” Fract. Calc. Appl. Anal., Vol. 2, pp. 1-30, Jan., 1999 [11] I. Petras, “The fractional-order controllers: methods for their synthesis and application ,” J. Electr. Eng., Vol. 50, No. 9-10, pp. 284-288, 1999 [12] I. Podlubny, “Fractional-order systems and μλD PI controllers,” IEEE Trans. Automat. Control, Vol. 44, pp. 208-214, Jan., 1999[13] C. A. Monje, A .J. Calderon, B .M. Vinagre, Y. Chen, and V. Feliu, “On fractional λPI controllers: Some tuning rules for robustness to plant uncertainties ,” Nonlinear Dynamics, Vol. 30, pp.369-381, 2004 [14] G. Maione, and P. Lino, “New tuning rules for fractional λPI controllers ,” Nonlinear Dynamics, Vol. 49, pp. 251-257, 2007[15] S. E. Hamamci, “Stabilization using fractional- order PI and PID controllers ,” Nonlinear Dynamics, Vol. 51, pp. 329-343, 2008[16] S. E. Hamamci, “An algorithm for stabilization of fractional-order time delay systems using fractional- order PID controllers ,” IEEE Trans. Automat. Control, Vol. 52, pp. 1964-1969, Oct., 2007[17] O. Diekmann, S. A. van Gils, S. M. verduyn Lunel, and H.-O. Walther, Delay Equations: Functional-, Complex- and Nonlinear Analysis. Applied Mathematical Sciences. Springer-Verlag, 1995.时滞系统λPI 控制器参数整定的图解法王德进 张江辉天津科技大学电子信息与自动化学院 天津 300222摘要:本文给出一种分数阶时滞系统λPI 控制器参数整定的图解法。