内容发布更新时间 : 2024/5/24 3:24:39星期一 下面是文章的全部内容请认真阅读。

题目：现代控制理论作业

Research on the Application of Model Free Adaptive (MFA) Control in Gas Turbine

Abstract -In this paper, the current application .status of gasturbine control technology i.s

1

introduced, it presents the modelfree control theory and method in perspective of definition ofmodel free control and model free adaptive control algorithmsmainly existed and its advantages. This paper introduces .someunsolved problems by using traditional control approaches. Itgives reasons why the model free is .suitabler the control ofgas turbine. Afterwards, it describes how the model freeadaptive control technology is applied to gas turbine and itsrelated techniques in detail in this paper. In this part, it couldprovide theoretical户undation户r applying model free controlin actual gas turbine control .system. Finally, conclusions andissues needed further research are putrward at the end ofthis paper.Keywords一Gas Turbine, Model Free Adaptive (MFA),MFAC, and Neural Network

I. INTRODUCTTON

Gas turbine is a typical large-scale dynamic systemwith multiple dimensions. It is a nonlinear, time-varyingsystem with uncertain parameters, and contains a numberof un-modeled dynamic parts. Most of its parametershave the complex characteristics of delay and saturation

[1]. It's extremely difficult to control gas turbine withperfect performance, while energy efficiency andenvironment protection are becoming the basicrequirements for energy technology. It is much harder toachieve optimal and coordinated control for gas turbine

using single conventional control approach. So advancedcontrol theory is needed in gas turbinecontrol system.

As part of modern control theory, Model FreeControl is an advanced control strategy. It has ainnovative theory compared withother model basedcontrol approaches. Model free adaptive controltechnique has excellent performance in the aspect ofadaptiveness, robustness and non-linearness [2], itstypical characteristic is model-free, which is suitable forthe system of nonlinear, strong coupling, stronginterference, and time-varying like gas turbine. At thesame time, some complex systems controlled bycorresponding model free controllers are very conciseand effective [2]. Obviously, with the development ofrelated technologies, mode free control applied in gasturbine and similar complex industrial processes controlwill be drawn greater attention. Thus, oriented gasturbine control technology, some related issues derivedfrom using MFA control theory are discussed andstudied in this article

II. CTJRRFNT STATTJS OF C;AS TTJRRTNF CONTROT

The gas turbine is a complex thermal power systemthat is difficult to establish its complete and accuratemathematical model, its corresponding analytical modelis strong nonlinear. At present, single conventionalcontrol approach is always appliedtogas turbine. Suchas, linear ratio method and low value select switch arcused to control the flow of fuel [3]. The ratio algorithmwith inertia is used to control the load [4]. Proportional-integral based on pre-selection model is used to controlthe temperature of key components, as well as somesimple proportional-integral control algorithms arc stillused in the acceleration control [6]. There arc otheradvanced algorithms, such as adaptive PID [5],predictive control, open-loop optimal control [7],observer controller based on models and so on, whicharc applied to

2

controthe temperature and pressure ofatmosphere inlet of compressor, pressurfluctuationsbefore fuel control valve, and speed control. Certainlythey can achieve certain effect with these controlalgorithms stated above. However, as gas turbine isrunning, the performance of components,

other

auxiliarymachines

and

accessories

or

systems

would

have

beenchanged,andwith existing nonlinear, time-varying withlarge time delay and random interferences. So singleapplication of these conventional control methods havebeen difficult toensure that ever-changing performancerequirements of gas turbine, and even the convergence and stability of control system arc always unsure.Particularly in the variable conditions of states, withthese fixed parameters control methods arc not easy toachieve satisfying results. Reversely, it may makecorresponding control technologies andstrategics muchmore complex. In other words, it's becoming harder toachieve optimal andcoordinated control in gas turbinewith the objective of energy saving, high efficiency and environmental protection with them.

It can be seen, that applying classic controlalgorithms or those algorithms based models and derivedfrom modern control theory has been impossible to meetrequirements in gas turbine or other complex industrialprocess control systems. It means the algorithms with

high parameters and structure adaptive arc neededurgently to apply in gas turbine to achieve adaptivecontrol in its life cycle, optimal control of the wholesystem, and maintain gas turbine operating with highefficiency. III.

MFTHOFJ ANFJ TFC'HNOI

Model-Free Adaptive (MFA) control [8], as its nameimplies, is an adaptive control technique without settingup model of the process, it uses I/O data of the controlledsystem to design controller, even without anyinformation of the system. In other words, MFA is acontrol theory and method, which do not need anyinformation of system model within the corresponding controller.

It has been paid extensive attention since the theoryhad been put forward. In theory, Professor Han Zhi-gangofficially presented in [9] in 1994. Initial research workfocused on basic form of model free control law issues[10]. It proved the stability of MFA control system, and

provided a reasonable valid theorem, and researched theestimation method for characteristicparameter in pan-model. As well as HOU Zhong-sheng, who proposedmodel free control theory and applications in his PhDthesis [1 l ] in 1993一1994. Reference [12] put forward afunctional combination of ways based on MFAC,functional combination is so called that the control lawsis not based on mathematical models of controlled object,but focus on thesystem combinatorial optimization andhe rate of functional requirements rather than thecontrolled object. Dr. George Cheng in [2] providedMFA control technology based on neural network. Theirresults had analyzed the concept from theoretical andapproach to realization indetail, which laid a solidfoundation for the following development andapplication of the theory.

3

In the area of methods and technologies, it includesmethod stated in PhD thesis of Professor HOU Zhong-sheng. The idea of this model introduces a new pseudo-gradient vector and pseudo-order, and uses a series ofdynamic linear time-varying model (tight format, partialformat, wide format linear model) to replace the discrete-time nonlinear systems in the vicinity of the controlledsystem, this model uses the I/O data to estimate thepseudo-gradient vector of the process, for the MFAcontrol [13]. Another idea is proposed by Dr. GeorgeCheng, this idea realizes adaptive control for processthrough introducing into neural network. This methodhas been stated and described in [14], it has beendemonstrated that this algorithm can be applied to anysingle input or multi-input open-loop system which isstable controllable and continuous or continuous reactionsystem without complicated manual tuning, quantitativeknowledge of the process, or any identifiers of thecontrolled system, as well as the learning process.

Model free control combines the advantages oclassical and modern control theory; also it brokethrough the shackles of those theories, and opened up awhole new way methods always need a deep understanding of processand its environment. Generally, differential anddifference equations arc used to describe the dynamiccharacteristics of the process. However, in gas turbineand other complex industrial process control system,which arc too complex or difficult to understand theinherent laws, therefore, it is hard to obtain quantitative knowledge of process. As part of modern control theory,model free control has lots of research results since itdeveloped. Adaptive, robust, easy-to-use, as well asmodel free and low cost controller can be used to solvethe complex problems in industrial control. Model freecontrol has many merits, such as it has not processidentifier and identified mechanisms in MFA control system, no need to know the accurate quantitativeknowledge of process and complexly adjust theparameters of controller; also it has stability analysis andcriteria for closed-loop system to ensure the stability ofthe control system [2].

I V. HY MODFI, FRFF ('ONTROI, IS SiJITABI,FFOR (;AS TiJRBINF ('ONTROI, SYSTFMA

As in part II ,even conventional modern controltheory and methods have been applied in gas turbine, butthose algorithms always based on mathematical model ofthe system and even need an ideal applicableenvironment. However, the actual working conditions of

gas turbine arc changing at any time, the outputsrequirements arc vari-ational with different operatingstates. So these existed algorithms have been hard toachieve the adaptive control with parameters and systemstructure in gas turbine. Fortunately, it is such thecomplexity of these issues as much more advancedmodern control theory can be applied in this field withfull merits.

Clearly, according to the current practical controltechnologies applied in gas turbine, we needs to improvethe efficiency by using reliable and low-cost methods

with good control quality and convenient adjustableparameters, which can alternate to the traditionalmethods based model. This kind of theory and controltechnology could achieve

4

the parameters and structureadaptive control in gas turbine. From a theoretical pointof view, in order to avoiding to face the robustness anddynamic problems produced by applying controlmethods based on models in gas turbine, new control

theory technology to shake off the modern controlmethods relied on models arc needed to explore. Andfrom a historical point of control theory, control theory isdeveloped from simple regulator and PID control thatneedn't mathematical model, to classical control theorybased on transfer function model, and modern controltheory based on state-space, up to now intelligent controlto get rid of mathematical model of object. Thus, modelfree control applied in gas turbine and other industrialprocess not only the gas turbine control technologydevelopment requires, it's also suitable for the forwarddevelopment of control theory. B. Model Free Control Solves Key Issues in Gss Turbine

The actual operating process of gas turbine iscomplex it is hard to establish accurate mathematicalmodel. From the view of current un-resolved problems in gas turbine, some model free control methods would be applied put forward in this article. As follows: (1) Large dynamic characteristics within the variable

conditions in the operating process of gas turbineIn the variable states of start up, speed up, slow down,top and smooth operating process of gas turbinevariable parameters changesdrastically in working statesEven in the same state, almost all parameters havesignificant difference as the conditions changed. Thewhole operating process presents great dynamic andnonlinear characteristics. Models of this kind object arcimpossible to set up, especially for those stronglynonlinear systems. Even a mathematical modeestablished, it would be with high cost and much timeand based on static condition that must be simplified andlinearized. Actually it has un-modeled dynamics androbustness features, so control theory based onmathematical model arc not suitable for gas turbine.Model free control method came forth and developedwith the objective of solving the bottleneck problems of modern control theory, and its model free feature hasprovided the method to realize effective control incomplex process. Thus, it has approach to obtain

distributed optimal control in gas turbine while operatingin variable states. Through designing such as timevarying MFAC, can effectively settle the adaptive control and time varying control issues.

(2) Strong coupled and corresponding MFA controlmethods among the components and parameters in gasturbine Figure 1 illustrates the coupled relationship withinthe core components and parameters in macro level. Inhe micro, the parameters among the gas turbine still

existed serious

coupled relationships. Such as inlet guidevane of compressor, combustion, speed of turbine and soon, these objects is related with other system parametersstrongly and complexly. Under normal circumstances,the multi-input multi-output (MIMO) processesformedby coupled relationship arc

5