Anda di halaman 1dari 6

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

Contents lists available at SciVerse ScienceDirect

Education for Chemical Engineers


journal homepage: www.elsevier.com/locate/ece

Process control: Domains, disciplines and cognitive difculties


Saidas M. Ranade a, , Hector Salazar b , Luis Andres Rodriguez c
a b c

General Physics Corp., Houston, TX, USA Dyprotec, S.A., Bogot, Colombia Ecopetrol, S.A., Barrancabermeja, Colombia

a b s t r a c t
The underlying stratum of domains, representations and disciplines makes the study of process control challenging. In this article the authors classify and provide specic remedies for addressing the underlying common difculties faced by novices in the eld of process control. The authors also share content that might be of direct value to practicing engineers. Engineers can use the identifying similarities and differences strategy for improving understanding of labels and identiers such as excitation and DC gain, but the focus needs to be on discerning subtle rather than gross differences. Good practical examples serve as helpful learning aids for complex concepts such as anti reset-windup and non-minimum phase zero but are not easy to nd. For big universal ideas such as causality and stability, the authors found a narrative style, linking content to learners prior knowledge and use of metaphors three techniques missing from most standard textbooks to be useful for overcoming the cognitive barriers. 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. Keywords: Learning process control; DC gain; Anti reset-windup; Non-minimum phase zero; Causality; Stability

Contents
1. 2. Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Tricky terminology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. DC gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Challenging concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Anti reset-windup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Non-minimum phase and non-minimum phase zero . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Deeper diversions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Causality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions and recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Appendix A. Supporting data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 2 2 2 3 3 3 3 3 4 5 5 5 6

3.

4.

5.

Corresponding author. E-mail address: smranade@swbell.net (S.M. Ranade). Received 1 July 2011; Received in revised form 30 October 2011; Accepted 31 October 2011 1749-7728/$ see front matter 2011 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ece.2011.10.003

e2

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

1.

Introduction
Question: What comes to your mind when you hear the word phase? Answers: Chemical engineer: Solid, liquid, vapor Electrical engineer: Phase lead or lag in AC currents Mechanical engineer: Planning, execution, auditing

Process control is a topic of interest and utility for engineers, scientists, economists and even philosophers. As a standalone discipline, just like chemical reactor engineering or electro dynamics, one expects subject specic jargon to exist such as PID controller, set point and cascade control. However since the subject seems to have been developed independently by each discipline as well as evolved separately as a discipline in recent years, in addition to its own language, it has terminology that has come from other disciplines. In process control there are different ways to identify a system. Chemical engineers traditionally are taught to write and test rst principles based models in the time domain. Electrical engineers are trained to use frequency analysis as a way to understand the dynamic behavior of a system. Laplace transform is a mathematical device used to make a round trip from the time to the frequency-dependent complex variable (s) domain and back. State space is another way of representing a set of differential equations. It can be used for representing multiple input and multiple output systems, is ideal for performing matrix operations and can provide more insight on a systems dynamic behavior than the inputoutput based models. State space also seems to be a preferred method for representing mathematical models among the scientists across the pond. There are also different perspectives of a system and its dynamics based on whether one works with hardware or software. The underlying stratum of domains, disciplines, perspectives and preferences allows multiple ways of linking concepts which makes the study of process control interesting and challenging. The authors had an opportunity to develop from ground up and deliver ve sessions of a course on fundamentals of regulatory process control to over hundred plant operators and process engineers. During the course of the entire training event lasting almost twelve months, the authors performed an empirical analysis of the muddiest topics and experimented with ways to overcome the learning hurdles. In July 2011, the authors also conducted an online survey of 43 engineers to verify the validity of the claims made in this article. The survey results are summarized in Appendix A. In this article the authors share key ndings and lesson learned from their investigation.

2.2.

Excitation

Two or more labels or words may represent exactly the same idea. An example of the same idea being called by different names depending on the discipline is the one that describes a change made to an input signal to study the dynamic response of a system. The label and the corresponding discipline are listed in Table 1. Other examples similar to the ones shown in Table 1 are the words autonomous/time-invariant and the words actuator/nal control element/control valve.

2.3.

DC gain

2.

Tricky terminology

In todays matrix organizations it is common for chemical engineers to work side by side with other types of engineers and scientists in cross-functional teams. A common vocabulary clearly helps improve the team communication. Working on process control related projects poses a communication challenge because of tricky terminology. Here we present examples that illustrate the three different types of labeling problems.

Another situation arises when one term from one discipline begins to be used in other disciplines but another term describing a similar idea already exists in that discipline. We asked several chemical engineers if they could explain the meaning of the term DC gain? The responses we got ranged from No idea to I guess it has something to do with direct current. Additional data further validating this claim is presented in Appendix A. DC gain is the steady state response of a system to a unit step input (DiStefano et al., 1995, p. 132). Some popular textbooks (Macia and Thaler, 2005, p. 222; Franklin et al., 1994, p. 94) use the phrase DC gain but do not state what DC in DC gain stands for. DC gain is the value of a transfer function of a system as the s variable in the Laplace domain approaches zero or as the input frequency goes to zero. Since DC current has zero frequency or is constant when plotted against time, the result of that calculation is called the DC gain. The term originated from electrical engineering and is seldom used by chemical engineering students. It appears to be a carryover from early frequency response analysis work done using DC motors. Its popularity in newer textbooks may be attributed partially because it is used as an internal variable name in Control System Toolbox; a product developed by Mathworks, Inc. The equivalent idea in the time domain is called the steady state gain which is the ratio of the output value of a system response to a unit step change in the input as time goes to innity. It is the gain of a system when time t or when the input excitation is not changing with time. The terms DC gain and steady state gain are not only at the intersection of two disciplines but also at the intersection of two domains: Laplace variable (s) and time.

2.1.

Phase
Table 1 Same idea, different labels. Preferred term
Input change Forcing function Excitation Command signal change

A word or a phrase may mean two completely different things to different people. As shown below, the term phase can have totally different meanings in different disciplines. For supporting data please see Appendix A. Other examples of words that have different meanings in different science disciplines are span, range and critical points.

Discipline
Chemical engineers Mathematicians Electrical engineers Robotics

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

e3

3.

Challenging concepts

The next category of difculty is related to the comprehension and application levels (Krathwohl et al., 1973) of cognitive processing. As is true in many disciplines some concepts are intrinsically difcult to grasp (see supporting data in Appendix A). A common characteristic of such concepts is that they embody many underlying facts and concepts. As illustrated next with two examples, one needs to comprehend all the embedded ideas to understand such concepts.

3.1.

Anti reset-windup

Consider the phrase anti reset windup. The word anti is unambiguous except it is not clear what it is referring to. When we asked students to explain what the term reset refers to many students were not clear about that concept. Reset mode is another name for Integral (I) part in a Proportional-Integral (PI) controller. Why another name? Just like proportional band (PB) is a parameter used in the eld to adjust the proportional gain, the term reset controller came from people working with the applications of PI controllers in the eld. A proportional-only mode leaves a steady-state offset. The integral action is added to the proportional mode to reset the offset to zero. It resets the output value to the set point. In the absence of the I mode, one would accomplish the same objective in a P-only controller by resetting the offset to zero by manually adjusting the bias. During situations such as activation of an override mode or non-surge part of a surge controller operation, or a long startup of a process, the I mode continues to reset the bias and keeps moving the output signal way past the valve saturation (Liptak and Liptk, 2005, p. 118). In case of a control system that keeps sending output signal past the valve saturation, it has to rst unwind from the windup before it can begin to have any effect on the process being controlled. A method used to prevent occurrence of the windup is called anti reset windup. One way to prevent the override problem is for example to stop the integration action when the output saturates. The word anti then applies to the windup caused by the reset mode. So it is not anti-reset windup but anti reset-windup. Smith (2009) presents an easy to comprehend example showing the effect of reset windup on the performance of a hot water.

To understand these terms one must start with a topic from high school Algebra called rational functions. Both poles and zeros are properties of rational functions. Knowing poles and zeros one can reconstruct the inputoutput differential equation for the LTI system. But poles have been emphasized more than zeroes because they have a direct impact on the stability of a system. A polezero graph shows the poles and zeros on a complex plane whose axes are the real and imaginary parts of the Laplace domain variable s. Most textbooks state that an inverse response of a system appears as a non-minimum phase in the polezero graph. However, what is left out is that the vice versa is not always true. A non-minimum phase does not always imply an inverse response. Non-minimum phase can also refer to an unstable process response or a dead time response. The addition of zero to non-minimum phase makes the term even more interesting. An inverse response representation on the polezero graph has at least one non-minimum phase zero. But, a non-minimum phase zero may also appear when a dead time response is approximated by a rational function.

4.

Deeper diversions

Some of the terms and topics in process control are universal and are of interest to researchers in the elds of social sciences, economics, cognitive sciences, history and philosophy. Here, cognitive difculties manifest in two types of symptoms. As illustrated below with the notion of causality, the symptom, validated via survey results summarized in Appendix A, is not being able to articulate precisely the effect of the big idea of causality on the system model. The example of stability illustrates the symptom, validated by the data in Appendix A, of how practitioners are able to remember the various denitions of stability but unable to explain the underlying pattern of classifying the types of stabilities.

4.1.

Causality

3.2. zero

Non-minimum phase and non-minimum phase

Even if you have no background in process control, it is clear that the term non-minimum phase zero is information (or Shannons entropy, Gleick, 2011, p. 229) rich. The two phrases are like the phrase space time curvature or Maxwells demons used by quantum physicists. Understanding these terms is important because they are useful in predicting the stability and controllability of dynamic systems. Some degree of confusion is likely to stem from the use of the word phase. Phase denotes the particular point in the cycle of a waveform, measured as an angle in degrees with reference to 0 . The phase in non-minimum phase really should be called phase shift or phase change because it is the difference in the phases of the output and the input for a frequency response for a given frequency. The drawback of using phase-shift instead of phase is that the expression will become longer than what it already is.

Since causality is a topic about which Aristotle, Einstein, Russell and Hume had something to say, it must at least be interesting if not also important. Many philosophers and scientists are interested in the concept of causality in a more general way because their primary focus of study is natural systems. In process control the focus is on man-made systems and the causality of interest is the temporal type. A system is causal if given the past and the present inputs one can fully dene the systems present output. For controllers to be implemented in the physical world the underlying models, algorithms and machines must be causal. If causality is such a universal idea, why is the principle of causality is not emphasized in common process control textbooks (for example Riggs, 2001; Bequette, 2003; Seborg et al., 2004) used in chemical engineering courses? This may be because feedback controllers use time as the independent variable and are always causal. Most chemical engineering textbooks also use a rst-principle (mechanistic) driven approach for deriving system models and for LTI use the state-space formulation. Although not stated explicitly in many textbooks, in the state-space form the idea of causality is encapsulated in the denition of a state of the system at time t = t0 as incorporating all effects of all previous inputs prior to time t and the assumption of initial condition of rest

e4

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

(zero initial conditions) or an initially relaxed condition for a system represented by a linear ODE also makes the system causal (Hsu, 1995, p. 61). In modeling of process dynamics in chemical engineering the applicability of the linear systems theory is assured by letting the output be a departure from the resting state (Norman, 1981) or equilibrium state. A system is said to be relaxed at time t = t0 if the energy stored at time t0 is zero (Gajic, 2003; Gopal, 2008). For a system at rest all the initial conditions are set to zero. Unlike chemical engineering process control books, books on signal processing and linear systems theory discuss the topic of causality rst because non-causal systems are of interest for example in signal processing, digital image processing and optics where the independent variable may not be time. Second, even when time is an independent variable, non-causal systems are useful in providing benchmarks of performance an ideal band-pass lter is one such example (Kailath, 1980, p. 24). In many textbooks the presentation on LTI systems begins by stating the differences between linear and non-linear systems, time-invariant and time varying systems and causal and non-causal system. Then the authors show the precise link between the assumption of linearity and time-invariance and the form of the differential equations representing the system but the impact of causality is illustrated with some trivial examples like y(n) = x(n + 1). It is important for students and practitioners of process control to know precisely how the condition of causality affects the mathematical representation of the system model. Causality affects the different forms of modeling dynamic systems such as convolution, differential equations and transfer functions in different ways. For LTI systems the convolution/impulse response is another way to relate the output at a time t with the systems response to an impulse function and another input signal. The causality condition affects the limits of integration of the input/output convolution model based on an impulse response. The precise impact of causality on impulse response/convolution model is illustrated by Kailath (1980, p. 24). The derivation shows an ideal way to clearly show how different assumptions and conditions affect a system model. The assumption of time invariance means we can select a lower bound as zero. The choice of t rather than innity as the upper bound comes from the assumption that we are dealing with a causal system. Newtons second law of motion written as a basic equation has an equal sign and does not distinguish between the input and the output. The standard way of writing differential equations does not by itself include the causal condition. It is in the auxiliary conditions such as conditions of initial rest that causality becomes a part of the model. If one used the initial state to solve a differential equation one is implying that knowing the past one can predict what the system will do each moment. For a non-causal system one would need to introduce different boundary conditions (Krener, 1980). The causal condition also affects the choice of the Laplace transform (LT) used in designing and verifying model properties. A unilateral LT can be used with the assumption that the input signal and the system are causal. Unilateral transform is not useful in analyzing non-causal systems and non-causal signals. In the transfer function model of system dynamics, the convention that the denominator is the input and the numerator is the output is expressed as a requirement that the degree of the denominator of the rational transfer function must be

equal to or greater than that of the numerator (Martin, 2000, p. 114). The mathematical language that the rational function is proper or strictly proper assures the casual link between the input and the output. Just stating a rule enables one to nd it but to internalize the rule a simple illustrative example can be quite benecial. De Silva (2005, p. 148) presents one such example showing the infeasibility of building a system with a unit step input producing an unlimited response. Other examples of topics like causality are nonlinearity and the history of development of the Laplace transform.

4.2.

Stability

Often the topic of causality in textbooks is followed by or preceded by a discussion on stability of LTI systems. The two are totally independent features of a system. Intuitively, we want systems we design to be insensitive to slight internal and external disturbances. We also want the output signal to track the transformed input signal and not be innite at one or several operating points (Shmaliy, 2007, p. 97). One likely source of difculty in understanding the concept of stability is that there is not one single kind of stability. Hence, we must supply the additional phrase in the sense of. The tendency in the literature is to quickly jump to the mathematical denitions of types of stability and focus on the procedures for how one might be able to check if a system or a feedback controller is stable or not. The weak link is the lack of focus on the concept of stability and the lack of a classication framework. Most practicing chemical engineers remember the picture, illustrated in Fig. 1, of the ball shown at various points on a terrain consisting of hill, a valley and a plateau rst introduced in their undergraduate physics courses. Many chemical engineers also recollect the stability discussions related to the steady state performance of an isothermal CSTR with an exothermic reaction. However, the discussion on stability in many textbooks begins with the denition of stability in the sense of Lyapunov and the associated mathematical denitions missing out on the opportunity to relate the new ideas to students previous knowledge about stability. Although we talk about stability of a system, what really we should be referring to is the stability of equilibrium points or steady state operation points. The idea of static equilibrium is that all forces are in perfect balance. The idea of steady state is that the macro-state properties of the system are not changing with time. In simple terms there are two kinds of stabilities. Internal stability related to system states often called the zero input stabilities and external stability related to the system output also called the zero state stabilities. The depiction of the ball at various points in a terrain is a useful device to dene the stable, unstable, meta- or semi-stable and neutrally stable equilibrium states. An example of such a system is illustrated by Brogan (1991, p. 343). It is also a useful illustration to comprehend the ideas of locally vs. globally stable states which are related to the distance of the starting point of the ball position relative to an equilibrium point. Local implies validity only as long as the starting point is in the vicinity of the equilibrium point. Stability in the sense of Lyapunov is a very mild requirement on equilibrium points and is akin to the idea of locally stable equilibrium points. In particular it does not require trajectories starting close to the origin tend to the origin asymptotically (Murray et al., 1994, p. 180). The two other adjectives used to characterize internal stability are asymptotic and uniform which are related to

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

e5

Fig. 1 Equilibrium points and stability concepts. Derived from Brogan (1991, p. 343). the time aspects of the system reaction to small perturbations. Asymptotic implies eventually. Uniform implies for all starting points (t0 ). The adjective exponential refers to the rate of approach to the equilibrium point. External stability may also be characterized in different ways. The most common method is called BIBO which is bounded-input, bounded-output stability. BIBO stability is used to characterize the behavior of relaxed systems (with zero initial conditions) (Brogan, 1991, p. 346). The limitation of external stability is that it does not guarantee internal stability. Hence for LTI systems the rigorous state-space approach is preferred. However building a good state space model requires extra investment. Also, important to know is that a feedback can make a stable system unstable. Other topics like stability are chaos and limit cycles. how to why. Unfortunately, many engineering books still jump too quickly into terse mathematical statements and do not cover the why in depth. Underlying assumptions and their meaning needs to be explored in more depth. As illustrated with our style of explanation of the principle of causality, what might work better for explaining big universal ideas is a style similar to that used by Late Professor Feynman in his lectures on physics (Feynman et al., 1989). It is now known that linking new ideas to students prior knowledge is an effective way to make the ideas stick. As emphasized under the topic of stability, many existing textbooks seem to miss out on this important opportunity to make that connection with prior knowledge.

Acknowledgements
The opinions expressed are those of the authors and do not necessarily reect the views of any entity mentioned in this article. The authors are grateful to General Physics Corp. and Ecopetrol, S.A. for their support. The authors are indebted to Professor Mark L. Fowler of the University of Rochester and Professor Steven Sloman of Brown University for sharing many useful insights on the topic of causality. Our special thanks to Lucero Ramirez for her support with translations.

5.

Conclusions and recommendations

From a learning standpoint, co-evolution of process control as a separate eld as well as a part of other disciplines makes it interesting and challenging. For simple ideas cross-discipline communication can be a challenge because each discipline or domain has developed its own vocabulary that at times is redundant or at other times inconsistent. Identifying similarities and differences is a well-established teaching strategy (Marzano et al., 2001, p. 16) and is appropriate for overcoming terminology related problems. However, the emphasis of the strategy needs to be on discerning subtle rather than gross differences between two labels or two phrases. Some concepts are intrinsically complex because they tie many other concepts together. Good practical examples serve as helpful learning aids for complex concepts. Finding such examples is not easy. Some ideas in process control are big ideas. There is lack of clarity in the process control textbooks and literature on the precise impact of these universal principles on the models or methods in process control. Our analysis of the state of process control textbooks and literature matches the following observations made by Bechhoefer (2005) while referring to the challenges physicists face in learning process control. They (engineering textbooks) are long. 800 pages are typical with the relevant information often scattered in different sections. . . .They often cloak concepts familiar to the physicist in unfamiliar language and notation. . .. The other alternative, more mathematical texts. . ., are terse but assume the reader already has the intuitive understanding of the subject. The availability of software tools over the past decade should have in principle shifted the emphasis of learning from

Appendix A. Supporting data.


In July 2011, the authors conducted an online survey of 46 engineers. Over 70% of the engineers had a chemical engineering degree. Twenty percent were electrical and instrumentation engineers and the rest other types of engineers. Bulk of the engineers had more than three years of practical experience. About a fth of the respondents considered themselves to be process control specialists. The questions were not scientifically tested for validity and reliability. So, while the specic percentages may not accurately represent the whole population of practicing engineers the authors are condent that the observed trends are valid and reliable. Majority of chemical engineers (79%) linked the word phase to solid, liquid and gas. Majority of the electrical engineers (66%) linked it to phases in project management. Six out of 43 engineers correctly identied the meaning of anti in anti reset-windup and only one-third of the respondents correctly identied the implications of a non-minimum phase response on pole-zero plots. Half of the respondents incorrectly supported the claim that ODEs written in the standard form embody the principle of causality. Forty-three percent missed the question

e6

education for chemical engineers 7 ( 2 0 1 2 ) e1e6

about applicability of unilateral Laplace transform to noncausal systems. One out of 44 respondents correctly matched zero-state stability with BIBO. While over 50% correctly linked asymptotic to the word eventually only two out of 41 respondents linked local and global correctly with initial position and only four out of 41 correctly associated the adjective uniform with for all starting times.

References
Bechhoefer, J., 2005. Feedback for physicists: a tutorial essay on control. Rev. Mod. Phys. 77, 783836. Bequette, B.W., 2003. Process Control. Prentice-Hall International, Inc., Upper Saddle River, NJ. Brogan, W.L., 1991. Modern Control Theory, third ed. Prentice-Hall, Inc., Upper Saddle River, NJ. De Silva, C.W., 2005. Mechatronics An Integrated Approach. CRC Press, LLC, Boca Raton, FL. DiStefano, J.J., Stubberud, A.R., Williams, I.J., 1995. Feedback and Control Systems. McGraw-Hill Book Co., New York, NY. Feynman, R., Leighton, R.B., Sands, M.L., 1989. The Feynman Lectures on Physics: Commemorative Issue, vol. II. California Institute of Technology, Pasadena, CA. Franklin, G.F., Powell, J.D., Emami-Naeini, A., 1994. Feedback Control of Dynamic Systems, sixth ed. Prentice-Hall, Inc., Upper Saddle River, NJ. Gajic, Z., 2003. Linear Dynamic Systems and Signals. Prentice-Hall, Inc., Upper Saddle River, NJ. Gleick, J., 2011. The Information. Pantheon Books, New York, NY. Gopal, M., 2008. Control Systems Principles and Design, third ed. Tata McGraw-Hill, West Patel Nagar, New Delhi.

Hsu, H.P., 1995. Signals and Systems. The McGraw-Hill Companies, Inc., New York, NY. Kailath, T., 1980. Linear Systems. Prentice-Hall, Inc., Upper Saddle River, NJ. Krathwohl, D.R., Bloom, B.S., Masia, B.B., 1973. Taxonomy of Educational Objectives, The Classication of Educational Goals. Handbook II: Affective Domain. David McKay Co., Inc., New York, NY. Krener, A.J., 1980. Boundary value linear systems. Asterique 75-76, 149165. Liptak, B.G., Liptk, B.G. (Eds.), 2005. Instrument Engineers Handbook: Process Control and Optimization. , fourth ed. Taylor and Francis, Boca Raton, FL. Macia, N.F., Thaler, G.J., 2005. Modeling and Control of Dynamic Systems. Thomson Delmar, Clifton Park, NY. Martin, T.E., 2000. Process Control, second ed. McGraw-Hill Book Companies, Singapore. Marzano, R.J., Pickering, D.J., Pollock, J.E., 2001. Class Room Instruction that Works: Research-based Strategies for Increasing Student Achievement. ASCD, Alexandria, VA. Murray, R.M., Li, Z., Sastry, S.S., 1994. A Mathematical Introduction to Robotic Manipulation. CRC Press, LLC, Boca Raton, FL. Norman, M.F., 1981. Lectures on linear systems theory. J. Math. Psychol. 23, 189. Riggs, J.H., 2001. Chemical Process Control. Ferret Publishing, Lubbock, TX. Seborg, D.E., Edgar, T.F., Mellichamp, D.A., 2004. Process Dynamics and Control, second ed. John Wiley & Sons, Inc., Hoboken, NJ. Shmaliy, Y.S., 2007. Continuous-time Systems. Springer, Inc., The Netherlands. Smith, C.L., 2009. Practical Process Control: Tuning and Troubleshooting. John Wiley & Sons, Inc., Hoboken, NJ.

Anda mungkin juga menyukai