You are on page 1of 9

A Bri.

ef History of Automatic Control


Stuart Bennett
hence the rateof combustion and heat output. Improved temperautomatic feedback corltrol systems have been known and by Bonnemain(circa1743used formore than2000 years; some the earliest examples turecontrolsystemsweredevised of his sensorandactuator are water clocks describedby Vitruvius and attributed to Ktesi- 1828).whobased on the differential bios (circa 270 B.C.). Some three hundred years later, Heron expansion of different metals. During the 19th century an extenof Alexandria described a range of automata which employed a sive range of thermostatic devices were invented, manufactured, variety of feedback mechanisms. The word feedback 20th and sold. These devices were, predominantly. direct-acting conis a century neologism introduced in the 1920s by radio engineers to trollers; that is, the power required to operate the control actuator describe parasitic, pcsitive feeding back of the signal from the was drawn from the measuring system. The most significant control development during the 18th output of an amplifier to the input circuit. It has entered into century was the steamengine governor. The origins device of this common usage in the English-speaking world during the latter half of the century. lie in the lift-tenter mechanism which usedto control the gap was Automatic feedback is found in a wide range of systems; between the grinding-stones in both wind and water mills. MatRufus Oldenburger, in 1978, when recalling the foundation of thew Boulton (1728-1809) described the lift-tenter in a letter IFAC. commented on both the name and the breadth of the (dated Ma>- 28, 1788) to his partner, James Watt (1736-1819), who realized it could be adapted to govern the speed the rotary of subject: I felt that the expression automatic control covered steam engine. The first design producedin November 1788, was all systems. because all systems involve variables. and one is and a governor was first used early in 1789. The original Watt concernedwithkeepingthesevariables at constantorgiven varying values. This amounts to concern about controlof these governor had several disadvantages: it provided only proporof speedatonlyone variables even though no actual automatic control devices may tionalcontrolandhenceexactcontrol to a be intentionally or otherwitbe incorporated in these systems. operating condition (this led comments that it was moderaI was thinking of biological, economic, political well as engineering tor, not a controller); it could operate only over a small speed as systems so that I pictured the scope IFAC as a very broad one. range: and it required careful maintenance. of The first70 years of the 19th century saw extensive efforts This diversity poses difficulties for historians of the subject to improve on the Watt governor. and thousands of governor patents (and for editors of control journals), and this article does not attempt to cover allappkation areas. were granted throughout the world. Many were for mechanisms The history of automatic control divides conveniently into designedtoavoidthe offsetinherentintheWattgovernor. four main periodsas follows: S p i c a 1 of such mechanisms were the governors patented by William Siemens (1823-1883) in 1846 and 1853, which substiEarly Control: To 19130 tuted integral action for proportional action and hence produced The Pre-Classical Period: 1900-1940 floating controllers with no fixed set point. Practical improveThe Classical Period: 1935-1960 ments came with the loaded governor of Charles T. Porter (1858): Modern Control: Pos8t-1955 his governor could be run at much higher speeds, and hence This article concerned with the first three of the above; other is greater forces could be developed to operate an actuator. A little articles in this issue deal with the more recent period. laterThomasPickering(1862)andWilliamHartnell(1872) invented spring-loaded governors, which also operated at higher Early Control: To 1900 Knowledge of the control systems of the Hellenic period was speeds than the Watt governor and which had the added advanpresened within the Islamic culture that was rediscovered in the tage of smaller physical size than the Watt and Porter governors. From theearly pears of the 19th century there were reports of West towardtheend of theRenaissance.Newinventionsand problems caused by governors hunting, and attempts to anaapplications of oldprinciplesbegan to appearduringthe18th century-for example. RenbAntoine Ferchault deRCamur (1683- lyze the governor mechanism to determine the conditions for stable (non-hunting) operation were made.J.V. Poncelet (17881757) proposed several automatic devices for controlling the temand perature of incubators.Thesewerebased on an invention of 1867) in 1826 and 1836. G.B. Airy (1801-1892) in 1840 and 1851 produced papers that showed how dynamic motionof the Cornelius Drebbel(1572-1663). The temperature was measured by the expansion of a liquid held in a vessel connected to a U-tube governor could be described using differential equations, but to determinethe containing mercuy. A float. in the mercury operated an m which, both metdifficultieswhentheyattempted a r conditions for stable behavior. Airy, in 1851. stated the condithrough a mechanical linkage, controlled the draft to a furnace and tions for stable operation. but his report is so terse that it is not

June 1996

0272-1708/96/$05.0001996IEEE

17

stability of a d> namic system was determined b> the location of The Pre-Classical Period (1900-1935) the rootsof the characteristic equation, and that a system became of unstable u-hen the real partof a complex root became positive; The early years the 20th century saw the rapid and widespread the problem washow to determine the location of the real parts application of feedbackcontrollers for voltage,current,and of the complex roots without finding the rootsof the equation. frequencJ.regulation;boilercontrolforsteamgeneration; Maxwell showed. for second-, third-. and fourth-order systems, electric motor speed control; ship and aircraft rteering and auto that by examining the coefficients of the differential equations stabilization; and temperature, pressure. andflow control in the the stability of the system could be determined. He was eble to process industries. In the twenty years between 1909 and 1929, sales of instruments grew rapidly as Fig. 1 shows. The majority gil-e the necessary and sufficient conditions only for equations up to fourth order; for fifth-order equations he gave neces- of theinstrumentssoldweremeasuring,indicating,and t\+o sary conditions. Maxn.ells paper, now seen as significant. was recording devices. but toward the end of the period the salesof of devices designed, little noticed at the time, and it was not until the early years of controllers began to increase. The range this centurythat the work began to be assimilated as engineeringbuilt, and manufactured was large; however, most were designed without any clear understanding of the dynamics both of the knowledge. system to be controlledand of the measuringandactuating Theproblemformulated by Maxwell was takenupby delices used for control. The majority of the applications were Edward .I. Routh (1831-1907), whose first results were pubthis lished in 1874. In 1577 he produced an extended treatise on concemed uith simple regulation, and in such cases lack of the Stability of Motion in which, drawing on the work of understanding was not a serious problem. However, there were some complex mechanisms involving complicated control laws Augustin-LouisCauchy(1789.1857)andCharlesSturm example:theautomaticship-steering (1 803- 1855), he expounded what we now know as the Routh- being developed-for mechanism devised by Elmer Sperry (1911) that incorporated Hurwitz stability criteria. I11 1895, the Swiss mathematician AdolfHurwitz(1859-1919)derivedthecriteriaindeP D control and automatic gain adjustmentcompensate for the I to pendently (basing his work on some results of C. Hermite). disturbances caused when the sea conditions changed. Another example is the electricity supply companies concerned about He had been asked for help with the mathematical problem achiei-in_g economicoperation of steam-generatingboilers. by his colleague Aurel Boleslaw Stodola (1859-1942), who Boiler control is of course a multimriable problem in that both was working on a turbine control problem. Most of the inventions and applications of this period were uater le!.el and steam pressure have to be controlled. and for concerned with the basic activities of controlling temperatures, efficient combustion the draught to the boiler has also to be pressures, liquid levels, and speed of rotating rnachineq: the controlled.During the 1920sseveralinstrumentcompanies the desire was for regulation and for stability. However, growth in develop complete boiler control systems. the size of ships and naval guns. and introductionof new weaponssuch as torpedoes,resultedintheapplication of steam; ,4s control devices and systems began to be used in many hydraulic.andpneumaticpowersystemstooperateposition differentareas of engineering,twomajorproblemsbecame control mechanisms. In the United States, Britain, and France, apparent: (1) there was a lack of theoretical understanding with engineers began to work on devising powered steering engines to assist the helmsman; on large ships the hydrodynamic forces on the rudder s e r e such that large gear ratios between the helm and the rudder were required hence moving the rudder took and a long time. The first of powered steering engine. designed by Frederick Sickcls in the U S . (patented 1853) was an opeu-loop system. The first closed-loop steering engine (patented 1566) was designed by J. McFarlane Gray for Brunels steamship the 3 r w N N N Great Enytern. In France? around the same time: Jean Joseph m r r m m r Farcot designedarange of steering engines and other closed-loop positioncontrolsystems.Hesuggestednaming his devices servo-moteur or moteur asservi, hence our terms servomechanisms and servomotors. Furtherapplicationsforcontrolsystemsbecameapparent with the growth in knowledgeof electricity and its applications. For example, arc lamps required the gap berazeen the electrodes to be kept constant, and generally it was helpful to all users if either the voltage the current the electricity supply was kept or of constant. Electricity also provided additional tools-for measurement, for transmission and manipulation of signals, and for actuation-which engineersbeganto use. Theelectric relay, which provided hish gain power amplification. and the spring Fig. 1. in) Ratio of instrument to machine5 sales in the United biased solenoid. which provided (crudej proportional control States, 1918 to 1936 11921 = 100). ib) Index of instrument sales i n action, were significant devices. the UnitedStutes, 1909 to 1936 11921 = 100).

18

IEEE Cmtrol Systems

to beused withinAT&T in 1931. Information about the amplifier 1 0 common language in which to discuss problems,(2) there 1 and was not published in the open literature until 1934. In developing were no simple. easily applied analysis and design methods. The only available analysis tool was the differential equation and the practical amplifier and in understanding its behat-ior. Black the application of the still not widely known Routh-Huraitz stability was assisted by Harry Nyquist (1889-1976); whose paper "Regeneration Theory" laid down the foundations of the so-called test. This is a laborious process. dependent on being able to obtain values for the parameters, and one that gives no guidance Nyquist analysis and was published in 1932. to the designer on the degree of stability, or what to do make to This work provided a practical device-the negative feedthe system stable. back amplifier-and led toa deeper understanding the benefits of of negative feedback in systems. It also, eventually, led to a As applications ~multiplied, engineers became puzzled and confused: controllers that worked satisfactorilq for one applica- method of analyzing and designing control systems which did not require the derivation and manipulation differential equaof tion, or forone set ofconditions, "ere unsatisfactorywhen tions, and for which experimental data-the measured frequency applied to different systems or different conditions: problems urith calculated data: from the arose when a change in part of the system (process. control- response-could be combined orle of ler, measuring system, or actuator) resulted in a change in the combined response the degree stability of the system could be major time constant thaltpart. This frequently caused instabil- estimated and a picture of changes necessary to improve the of ity in what had previously been, or seemedhave been, a stable performance could be deduced. to system. Some acute observers, for example Elmer Sperry and Contemporaneously with Black's work. Clesson Mason of E. Morris E. Leeds. noted that the best human operators did not use Foxboro Company developed a pneumatic negative feedback the an on-off approach to conlrol but used both anticipation, backing amplifier. Edgar H. Bristol, oneof the founders of the Foxboro off the power the controlled variable approached the set-point, as Company, had invented the flapper-nozzle amplifier in 1914. and small. slow adjustmlmts when the error persisted. Sperry The early versions of the flapper-nozzle amplifier were highly tried to incorporate these ideas into his devices, and for many non-linear (effectively on-off behavior), and during the 1920s years Leeds resisted attaching simple on-off control outputs to extensive modifications had only succeeded in increasing its his recorders because he realized that this would not provide linear range to about 7% of full range. In 1928, Mason began good control. experimenting with feeding back part of the output movement of the amplifier, and in 1930 produced a feedback circuit that In 1922. Nicholas Minorsky (1885-1970) presented a clear analysis of the control involvedin position control systems and linearized the valve operation. This circuit enabled integral (or of the formulated a control law that we now refer to as three-tem or reset) action to be easily introduced into the behavior PID control. He arrived his law by observing the way in whichsystem. In 1931, the Foxboro Company began selling the Staat bilog pneumatic controller which incorporated both linear ama helmsman steered. a ship. This work did not become widely plification(basedonthenegativefeedbackprinciple)and knou-n until the late 1930s, after Minorsky had contributed a series of articles toThe Engineer. But evenif designers had been integral (reset) action (Fig. 2). There was some initial market on aware of hlinorsky's work they would still have lacked suitable resistance to this device, the grounds of cost and because its linear. stable. amplrfication devices to convert the low power behavior was not understood. Foxboro responded by producing, in 1932, a bulletin explaining the principles the systemin clear of signals obtained from measuring instruments to a power level and simple terms and stressing how the behavior was different suitable to operate a control actuator. Slide and spool valves developed during the early partof the 20th century were begin- from what it termed "narrow-band" controllers, that is. those with limited linear range. ning to providethesolutionforhydro-mechanicalsystems, The electronic negative feedback amplifier and the pneumatic although valve overlap that resulted in dead space and stiction were problems that had to be overcome. However, there was an controller were the outcomes of Rork on industrial problems. impasse with respect to amplifiers for electronic and pneumatic During the same period, extensive work was being carried out on systems. As earl>-as 1920 the amplification problem was proving analog calculating machines under the direction of Vanevar a serious obstacle to the further development of long-distance Bush at the Massachusetts Institute of Technology. This w-ork of telephony. Improvements in cable design and the use imped- resulted in the differential analyzer, which provided a means of simulating the behavior of dynamic systems and of obtaining ance loading had extended the distance over which telephone numerical solutions to differential equations. It also led to the transmissions could take place without amplification, yet the transcontinental service in the S . was dependenton amplifica- study and design of a high-performance servomechanism by U tion. Telephone repeater's based on electronic amplification of Harold Locke Hazen (1901-1980) and his students.In addition the the signal a w e used around 1920. but the distortion they intro- to designing sae n 0 system, Hazen also undertook first major duced limited the number that could be in series. Expansion theoretical study of servomechanisms. His papers. published in used 1934,providedthestarting pointforthenextgeneration of of traffic on the network was also causing problems since it control system specialists. necessitatedanincreaseinbandwidth of thelineswiththe consequent increase in transmission loss. Harold Stephen Black The Classical Period: 1935-1950 (1898-1983) began work on this problem in the early 1920s. He realized thatif some of the amplificationof a high-gain amplifier During the period 1935-1940, advances in understanding of were sacrificed by feeding hack part of the output signal: the control system analysis and design were made independently by distortion due to .se and component couldbe reduced. On severalgroups in severalcountries. The bestknownand most no drift influential work came from three groups working the U.S. The in August 2; 1927, hp sketched a circuit for a negative feedback amplifier.Followingextensivedevelopmentwork.full-scale development in Europe and in Russia during this period followed a practical trials were carried out in 1930, and the amplifier began somewhat different path deriving from Vyschnegradsky's work in

Julze 1996

19

Russia and then Barkhausen's work in Germany, followed by developments due to Cremer. Leonhard, and Mikhailov. AT&T continued with its attempts find ways of extending to the bandwidth its communication systems. and upon obtaining of good frequency response characteristics. The ideal which they were seeking was a constant gain over a wide bandwidth with a sharp cut-off and with a small phase lag. Engineers in the Bell Telephone Laboratories worked extensively on this problem. but found that if they achieved the desired gain characteristic then the phase lag was too large. In 1940. Hendrik Bode, who had beenstudyingextensions to thefrequencq-domaindesign method, showed that no definite and universal attenuation and phase shift relationship for a physical structure exists, but that there is a relationship between a gi\-en attenuation characteristic and the minimum phase shift that can be associated with it. In the same paper he adopted the point (-1,O) as the critical point rather than the point (t1,O) used by Nyquist, and he introduced the concept of gain and phase margins, and the gain-bandwidth limitation. Full details of Bode's work appeared in 1945 in his book Xetwojork ilnul),sis and FeedbackArnpl(fierDesign. became clear during1941 that the cumbersome systemof relayThesecondimportantgroup.mechanicalengineersand to physicists working in the process industries in the U.S., encour- ing manually the information obtained from radar devicesthe gun controllers w-as not adequate to combat the threat of fast aged by Ed S. Smith of the Builders Iron Found? Company, began systematically developing a theoretical understanding of aircraft and that there was a need to develop a system in which the control systems they used. They sought to establish a com- an automatic tracking radar system was directly linked to the gun director, which was in turn linked the gun position controller. to mon terminology and tried to develop design methods. They Work on this "systems" problem brought together mechanipersuaded the American Society of Mechanical Engineers to of cal. electrical, and electronic engineers. and an outcome this formanIndustrialInstrumentsandRegulatorsCommitteein 1936. rhus becoming the first major professional body to form cross-fertilization of ideas was a recognition that neither the a section specificallyto deal with automatic control. Several mem-frequency response approach used by the communication engineers nor the time domain approach favored b) the mechanical bers of thisloosegrouping vr-ere aware of developmentsin Germany and in England. During this period the manufacturers engineers were, separately. effective design approaches for serof pneumatic controllers continued improve and develop their vomechanisms. What was required was an approach that used to instruments, and by 1940 field-adjustable instruments with PID the best features of each. control wereavailable-for example. an improved versionthe of LVork by Gordon S. Brown and his students at >TIT showed Stabilog and the Taylor Fulscope. In 1942, J.G. Ziegler and N-.B. how many mechanical and electrical systems could be repreNichols of the Taylor Instrument Companies published papers sented and manipulated using block diagrams. Albert C. Hall describing how to find the optimum settings for PI and PID showed, in 1943, that treating the blocks as transfer functions by control-the so called Ziegler-Nichols tuning rules. These ere (he used the Laplace transfonn approach) the system transfer a extended in the mid-1950s by Geraldine Coon (Taylor Instrulocus could be drawn, and hence the Nyquist test for stability menti. could be used. More importantly theand phase margin could gain be determined, and he introduced the use of M and N circles The third group was located in the Electrical Engineering which enable estimates the closed loop time domain behavior of Department of MIT and was led by Harold L. Hazen and Gordon S. Brown. They used time-domain methods based on operator tobemade.Anothergroupworkingthe so calledRadiation deT.eloptechniques, beganto develop the of block diagrams,and used Laboratory at MIT (this laboratory was concerned with use ingradarsystemsforthedetectionandtracking of aircraft) the differential analyzer to simulate control systems. Scholarly the of interchanges betweenhIIT and the University Manchester led designed the SCR-583 radar system, which, linked with419 director,wasdeployed in southeastEngland and had a high to a differential analyzer being built at Manchester University and, in 1936, Douglas Hartree and .4rthur Porter assisted A . success rate against Vl rockets. The M9 director was designed Callender of IC1 to use the machine to simulate an industrial by a group led by Bode and including Blackman. C . 4 . Lovell. and Claude Shannon, working in the Bell Telephone Laboratory. control system and to derive design charts for the system. Out of the work on the SCR-584 came the Nichols chart design The advent of the second world war concentrated control system work on a few specific problems. The most important method, workby R.S. Phillips on noise in servomechanisms, and of these was the aiming of anti-aircraft guns. This is a complex W. Hureaicz's work on sampled data systems. After the war, problem that involves the detection of the position of the air- details of the work were published in che seminal book Theory plane, calculation of its future position, and the precise control qf Servomechanisnls. of the movement of a heavy gun. The operation required up to The Radiation Laboratory group used phase advance circuits 14 people to carry out complicated observation and tracking in the forward loop to modify the performance of their control tasks in a coordinated way. The design of an adequate servo- system. Several other u-orkers, particularly in the U.K.. used mechanism to control the position wasa difficult task. It also minor loop feedback to modify system response gun and hence found

20

the Nyquist approach difficult. In 1942. A.L. John Whiteley MacColls Fundutnentul Theoq of Sewomechat~isnzs.began to of the British Thomson Houston Company proposed an approach set out the new approaches. Encouraged by the British governbased on plotting the inverse functions on a Nyquist diagram; ment, the Institution of Electronic Engineers held a conference in thesameyear H.T. Marcq (Kellog Company)independently in Londonin 1946 on radar, and the interest shown papers in the proposed a similar method. relating to servomechanisms resulted in a further conference The problemsraised Iby anti-aircraftcontrolweresystem devoted to control held in 1947. In the United States the governdesign problems in that several different units. often designed ment agreed to continue paying key people for a period of six and manufactured by different groups, had to be integrated; the months after the endof the war to enable themto write up their overall performancewas dependent notso much on the perform- work.OneoutcomewastheRadiationLaboratory Series of ance of theindividualunitsbutonhowwelltheyworked books, including Theor) of Servo~nechanisnt. together. Difficulties experienced in getting units to work toThe conference on Automatic Control held in July 1951 at gether led to a deeper understanding of bandwidth, noise, and Cranfield, England, and the Frequency Response Symposium non-linearities in sy:;tems. By the end the war people such as held in December 1953 in New York marked the beginnings of of ArnoldTustin (1899-1994) in England and R.S. Phillips. W. the transition period leading to modern control theory. The first Hurewicz, L. McColl, N. Minorsky, and George Stibbitz in the of these, organized the Departmentof Scientific and Industrial by U S . were concentrating on nonlinear and sampled data systems. Research. with the assistance of the IEE and the IMechE, was The other major development to emerge the fire control the first major international conference on automatic control. from work during the war was the study of stochastic systems: Norbert Arnold Tustin chaired the organizing committee, and 33 papers Wiener (1894-1964: wished to contribute to the war effort and werepresented,16ofwhichdealtwithproblems of noise, proposed tackling tl-e problem of predicting the future position non-linearity or sampling systems. There were also sessions on of an aircraft. His proposal was based on the work he had done analog computing and the analysis the behavior of economic of in the 1920s on genPralized harmonic analysis (Wiener. 1931). systems(thislatterreflectingboththeparticularinterest of He worked with Jokn Bigelow on implementing his prediction Arnold Tustin and the growing interest in applications of feedsystem. and they succeedEd in developing an electronic system back theory). forprediction.Wienerwasdisappointedthat in theendhis The wartime experience demonstrated the power the freof system was only ablk to achieve a marginal improvement (less quency response approach to the design of feedback systems: it than 10%) over the system developed empirically by the Bell also revealed the weakness of any design method based on the Telephone Laboratory. The work did lead to Wiener producing assumption of linear, deterministic behavior. Real systems are the report The Extrapolation, Interpolation and Smoothing of non-linear; real measurements contain errors and are contamiStationary Time Series w t h Engineering Applications (OSRD nated by noise: and in real systems both the process and the Report370.February 1. 1942)>knownastheyellowperil environment are uncertain. But what design techniques can be because of its yellow covers and the formidable difficulty of its mathematics. It was eventually published in the open literature used that allow the designer to consider non-linear and non-deterministic behavior and to allow for measurement errors and in 1949. noise? Also, the design problem changed from that of simply By the endof the war the classical control techniques-with achieving a stablecontrollertothatofachievingthebest the exception of the root locus design method of Walter Evans controller. Rut what is the best controller? (1948, 1950)-had been established. The design methodologies Ziegler and Iiichols had shown how to choose the parameters were for linear single-input systems-that is. systems that can of be described by linear differential equations with constant coef- of a given type controller to obtain an optimum performance of agivencontrolstructure (PI, PID).Similarly,Whiteleys ficientsandthathaveasinglecontrolinput.Thefrequency a particular performstandard forms enabled designers to choose response techniques, based on the use of Nyquist, Rode. Nichols. a and Inverse Nyquist charts, assessed perfomlance in terms of ance for a range of systems. Work was done on evaluating wholerange of performanceindicatorsincludingIAE. ISE, bandwidth, resonance, and gain and phase margins and provided ITAE, a graphical. pictorial view of the system behavior. The alternative and ITSE (Graham and Lathrop. 1953). Sterile arguments was approach based on the sollution ofthe differential equations using developed about which the performance indicator the best that what was important was the choice of Laplace transformtechniques expressed performance in terms until it was accepted of an rise time, percentage overshoot. steady-state error. and damping. appropriate performance indicator for a particular application. In addition to performance criteria based on minimizing some Many engineers prefemed the latter approach because the peris. the formance was expressed in real terms. that time behavior error function there was, for certain classes of system, interest in minimizing the time to reach a set-point (obvious applications of course.isthatuntilthe of thesystem.Thedisadvantage, development of the root LOCUSmethod there was no simple and are military target seeking servomechanisms and certain classes easy way in which line designer could relate parameter changes of machinetools).DonaldMcDondaldsNon-LinearTechniques for Improving Servo Performance (1950) was followed to time behavior changes. by The achievements of the classical era began to be consoli- during the 1950s extensive work on the time-optimal problem relating to the single controlled variable with a saturating control. dated and disseminated in books published during the 1940s and by Bellman early 1950s. The first book dedicated to control systemswas Ed The problem was studied by Bushaw (1952) and S . Smiths Automatic Control Engineering. published in 1942; (1956). In a definitive paperJ.P. LaSalle (1960) generalized all however, this book had a prewar feel to and it did not reflect the previous results and showed that if optimal control exists it it in thisareais the changes in approach that were developing from the wartimeisuniqueandbang-bang.Theprogressmade summarized in Oldenburgers bookOptimal Control (1966). work. The later books. Bodes book (referred to above) and Leroy

June 1996

21

The more difficult problem mas how to choose the control structure that would i w the best performance and how g to define this best performance. To do this; a model of the plant was needed: either physical-mathematical balance equationsof mass, energy, etc., in which the parameters are functions ofthe physical data of the process, or black box models based on experimental measurements-for example, frequency- response in which the parametersarenotdirectlyrelatedthephysical data of the s) stems. Work on developing frequency response ideas and design methods continued throughout the 1950s. Design methods for systems containing non-linearities were deceloped,as were the theoretical foundations of sampled-data systems. The teaching of servomechanisms and control theory spread, initiallq through special courses run for practicing engineers and graduate students and then through incorporation within the standard syllabus of many engineering courses.

reach a region in state-space and there would be a specified amount of time left. Formulated in this way, the problemcan be treated as a multistage decision making process. Working with Stuart Dreyfus. Bellman developed computer programs to produce numerical solutions a range of problems, and the results to a e r e published in 1962. The principal difficult!. with dynamic programming is the dimensionality problem, and even though u-e now have computing power far beyond anything available to Bellman and Dreyfus we stillneed to use approximations to handle complex systems. As well as involving positional accuracy. performance requirements also involve constraints expressibleas optimization requirements; for example, reaching a specified position in minimum time, or carrying out ofmaneuvers withminimum fuel a set consumption. Consequently, attention once again focused the on differentialequationapproachtotheanalysisanddesign of control systems. Dynamical problems that involve minimizing or maximizing some performance index h a l e an ob\-ious and stronganalogywiththeclassicalvariationalformulations of Modern Control Although the direction of some post-war work influenced analytical mechanics given by- Lagrange and Hamilton. The was generalization of Hamiltons approach to geometric optics by by the insights and new understandings developed during the war, the trajectory of development. Alistair J.G. MacFarlane Pontryagin (1956). in the form of his maximum principle. laid (1979) argues, was largely determinedby two factors: first, the the foundations of optimal control theory. This and Bellmans insight into the value and usefulness of the concept of state for problemthatgovernmentssaw as important.thelaunching, of maneuveling. guidance, and tracking of missiles and space ve- the formulation and solution many control and decision problems led to extensive and deep studies of mathematical problems hicles; and second; by the advent of the digital computer. The of automatic control. And the growing availability the digital of first problem was essentially control of ballistic objects. and of hence detailed physical models could be constructed in terms computer during the late 1950s made a recursive algorithmic differential equations, both linear non-linear; also measuring solution possible (as opposed to the search for a closed-form and instruments and other components of great accuracy and preci- solution in the classical approach). Michael Athans has placed the origin of what is now referred sion could be developed and used. Engineers working in the aerospace industries, following the example set by Poincark, to as modem control theory as 1956, and in September of that turned to formulating the general differential equations in terms year an international conference automatic control, organized on by of was held in of a set of first-order equations, and thus began the approach that the joint control committee the VDI and VDE, became known as the state-space approach. Heidelberg, Germany. During the conference a group of deleBetween 1948 and 1952 Richard Bellman, working in the gates agreed to form an international organization to promote mathematics department of the RAND Corporation, studied the progressinthefieldofautomaticcontrol.Anorganizing problem of determining the allocation missiles to targets so as group-Broida (France), Chairman, Grebe (Germany), Letov of (U.S.) FVelbourn to inflict the maximum damage. This work led him to formulate (USSR),Nowacki(Poland).Oldenburger as the principleof optimality and to dynamic programming. The l&J.K.),with Ruppel (Germanyj Secretary-was charged with choice of name was, according toan account published in 1981, drawing up plans for an international federation. The organizadetermined by political expediency. The research was supported tion, the International Federation of Automatic Control (IFAC), bytheAirForcebutthethen-Secretary of Defense had an as officially formed at a meeting held in Paris 011 Sept. 11 and as aversion to the word researchand it was assumed he would have 12, 1957. Also chosen were attendee Harold Chestnut the first an even greater aversion to mathematical research. m i c was, president, with A M Letov and V. Broida elected as vice presiQna dents. G. Ruppel as secretary, and G. Lehmann as treasurer. 4 t and still is. a word with positive connotations, andprogramming was thought to be more acceptable that planning. (Sames are this meeting the Russian delegate extended an invitation to hold important, and looking back over years it does seem that the the first conference in Sloscou- in 1960. 50 use of the names control engineering. automatic control, and The Moscow Conference was important and highly visible an syurms engineering have not achieved for our subject the rec- symbol of the change in direction that had been slowly developognition that might have been expected. Sames such as cyber- ingduringthe 195Os, and it is fittingthat at theconference netics a n d Tobotics command a greaterdegreeofpublic Kalman presented a paper. On the General Theory of Control recognition and apparent understanding.) Systems,thatclearlyshowedthatadeepandexactduality of In the latter part of the 1950s Bellman began working on existed between the problems multivariable feedback control optimal control theory, at first using the calculus of variations and multiwiable feedback filtering, hence ushering in a new but later, because of the boundary value problem inherent in the treatment of the optimal control problem. calculus of variations approach, seeking to formulate determiAn importantstep was Kalmanstreatment of thelinear nistic optimization problems in a way in which they could be multivariable optimal control problem with a quadratic perfonnsolved by using dynamic programming. His insight was to see ance index, and in particular the provision a synthesis proceof that by applying a particular control policy the system would dure. Further impetus to the state-space approach given with was

22

IEEE Control Systems

Fig. 3. The Executive Colrncil of IE4C 1959 (reproduced,from Automatica ~801. 7,p , 55, 1971 1.

Kalmans work on the concepts of observabi1it)- and controllability, and with Rosenbrocks idea modal control, which led of to extensive work on polt: shifting. A further important result was Wonhams proof that a sufficient conditionall the closedfor loop characteristic frequencies of a controllable system to be arbitrarily allocatablt: undt:r feedback is that all the statesof the system are accessible:. Thefinal triumph of time-responsemethodsappearedto come when Kalman and Bucy attacked the filtering problem. Their work, a5 well as producing the Kalman-Bucy filter, demonstrated the basic role of feedback in filtering theory and the duality that existed between the multivariable control problem andmultivariablefeedbackfiltering.FollowingtheMoscow conference: the state-space approach dominated the subject for almost two decades, leading Isaac Horowitz, m-ho continued to work on frequency response ideas, to a feeling of isolation and to a lament written in 1984 that .modern Ph.D.s seem to have poorunderstanding of evensuch a fundamentalconceptof Conclusion bandwidth and not the remotest idea of its central importance in The conferences of 1951 and 1953, together with the publifeedback theory. It i!; amazing how many are unaware that the cation of numerous textbooks: articles such as Tustins in Engiprimary reason for feedback in control is uncertainty. neering in 1950 and Brow-nr in Scientific American in 1951; and There was a rapid realization that the powerful optimal con- numerous articles on control topics in hchanical Engineering trol methods could not be used on general industrial problems during the early1950s brought automatic control to the attention because accurate plant models were not awilable and in man)- of engineers. The publicationof Wieners book The Huntan Use cases not achie\ able. 4 s Karl Astrom and P. Eykoff, writing in of Human Beings and a series of articles published in Scienrific 1971, remarked. a strength of the classical frequency response American in 1952 attracted the attention of a wider technical approach is its very- powerful technique for systems identifica- community. By themid-1950sthere a a s a growinggeneral tion, i.e., frequency analysis through which transfer functions awareness of the potential of automatic control. Many books on can be found accurately for use in the synthesis technique. In the subject intended for the general reader were published, and modern control the models used are )pcrrnrmtric models in terms theBritishgovernmentquietlyencouragedadebate on the of state equations, and this has led to interest in parameter subject. The emphasis in these popular and semi-popular works estimation and related tec miques. was on automation in the sense of mechanization and remote Further problems arose attempting to applying the state-space control of production lines and other assembly processes. There in of approach to industrial problems, one being the formulation of an was also great interest in the possibilities numerical controlof appropriate performance irtdex, not always obvious, and the other machine tools. Central to this debate were issues that many of the engineers being the complexity of the controller resulting from the design method. for example, the incorporationof a Kalman-Bucy filter in and administrators involved in control system work during the thecontrolsystemsresults i n thecontroller haling a dynamic war had anticipated-control systems had moved beyond feedcomplexity equivalent to that of the plant being controlled. As a back amplifiers and single-loop serITomechanisms and had beconsequence there was a revival of interest in the frequency-re- comeconcerned withlarge-scale.complexsystems.Gordon

sponseapproach, and a systematicattackon the problemsof developingfrequencyresponsemethodsformultivariable systems began in 1966 with a paper by Howard Rosenbrock. Turning to MacFarlanes second influence 011 the development of modern control-the digital computer--we find that the main impact during the 1950s and 1960s was to support theoretical investigations and particularly (using Wonhams definition)synthesis.Thedesign andimplementation of practical systems were much more strongly influenced through the replacement of electronic tubes by semiconductors suchdiodes. as transistors and thyristors in the fifties, Gerecke commented, as and the replacement of mechanical and electrical components by solid-state and microelectric devices. By the early 1960s the digital computerhadbeen usedon-linetocollectdata, for optimizationandsupervisorycontrol(MonsantoChemical Company.Luling!La.,in1960)andinalimitednumberof applications for direct digital control, for example, at an ICIplant at Fleetwood in the U.K. in 1962. However.its widespread use for on-line control did not occur until the earl)- 1970s. A leading advocate for the useof the digital computer in the process industrieswas Donald P. Eckman. who in the early 1950s persuadedseveralcompaniestosupport a researchprogram based at theCase Institute of Technology, Cleveland, Ohio. The program, originally entitled Process Automation, was renamed Control of ComplexSystemsbecauseEckmanwished to distinguish what he was doing from the popular imageof automation, meaning the mechanization of manufacturing and the displacement of labor. By the end of the decade Eckman was arguing in support of Systems Engineering with the idea that what industry needed was engineers with a broad background [cutting] across conventional boundaries of the physical engineeringandmathematicalsciencesandwithanability to approach problems analytically. to reduce physical s>-stems to an appropriate mathematical model to which all the power of mathematical manipulation, extrapolation, and interpretation can be applied.

Juue 1996

23

Brown and Duncan Campbell, in 1949, laid out clearly what the5 saw as the areas of application of control in the future: Improved automatic control is the co-ordinated designof ... plant, instruments, and control equipment. have ill mind more lye a philosophic evaluation of systems which might lead to ?he irnprosement ofproduct qua&, to better eo-ordination ofplant operation, t o n clarification of fhe economics related to newplanr design, and to rhe safe operation of plants in our composite social-industrial communi& These general remarks are illustratedbymentionthatcertainindustriesoperating at large producliorz migkr show appreciable increase in economy and quality on standand production items by improved automatic control. The consewation o raw materiuls used i CI process f n f oftenpromptsreconsideration o control.Theexpenditure of power or energy in product manufacture is another important .factor related to control. The protecrionof heairh ofthe population adjacent to large industrial areas against atmospheric poisoning and water-streampollutionis G su7ficientlyserious for stud} and problem ro keep us constuntly alert advances in the technique of automatic control. not only because ofthe human aspect bur because of the economy aspect. This they viewed a long-term as program with many technical and human problems that may take a decade or more to resolve. Since Brown and Catnpbell wrote these words, the penetration of control systems into everyday life has gone further than they perhaps expected. The complexityof what we now seek to control. the techniques that we have available. the power of and the technology-particularly the digital computer-place enormous responsibilities on us as engineers and as citizens.

Lauer, H., Lesnik, R., Matson, L., Servomechanism Fundamentals 1948 Brown, G.S., and Campbell, D.P., Principles ofSenomechanisms Oldenbourg, R.C., Sartorius, TheDynamics of Automatic Control &iener; X., Cybernetics: or Control and Communication in the Animal and the Machine 1949 Shannon, C.E.. Weaver, W . . TheMathematicalTheory o f Corninunication Wiener, X., Extrapolation, Interpolation, and Smoothing 0-f Stational? Time Series with Engineering Applications
1950 Porter, A , , Introduction to Servomechanisms

1931 Servomechanisms: Selected Government Research Reports .4hrendt, W.R., Taplin, J.F.,Automatic Feedback Control Behar. M . E , Handbook of Measurement and Control Chestnut, Harold, Mayer; R.W., Servomechanisms and Regulating System Design b l . I Famngton, G.H., Fundamenrals qfAzttomatic Control Fett, G.. Feedback Control S y s t e m Macmillan, R.H., A n Introduction to theT h e 0 9 of Control in Mechanical Engineering 1952 for Tustin, A , , Direct Current Machines Control Systems

Appendix: Books on Control Published Between 1940 and 1955


1942 Gardner, M,,4~, Barnes, J.L.,Transients rnLinear Systems and Smith, E.S., Automatic Control Eqineering

1953 Flugge-Lotz, I.,Discontinuous .4utomatic Control Haines. J.E.. Automatic Control of Heating a n d A i r Conditioning 1943 Jones, R.W.. Electric Control Systems Griffiths, R., Thermostats and Temperature Regulnring InNixon, F.E.,Principles ofdutomatic Conrrol struments Thaler, R.J., Brown, R.G., Servomechanism Analysis Hal1,A.C.. TlleA~~al~sisar~dSyrzthesisofLinearSeivomechaTustin, A , ,Mechanism of Economic Systems nism West, J.C.,Textbook of Se~vornechanisms 1944 i954 Oldenbourg. R.C.. Sartorius, H.. D y m n i k Selbsttutiges ReAhrendt? W.R.. Servomechanism Practice gelungen Profos. P., Ekrorielle Regeltheorie Evans, U!R., Control System Dynamics VDI, Regelungstechnik: B e y ( & und Bezeichnungen Fett, G.H., Feedback Control Systems Izawa, K., Introduction I O Automaric Control 1945 La Joy, M.H.. Industrial Autonzutic Coutrols Bode, H.W., YetworkAnalysisandFeedback Arnplzj5er Oppelt, W., Kleirles Handbuch Technisches Regelvorgange Design Peters, J., Einsch~.~.irtgvorgaige, Gegenkopplung, Stabilitat Eckmann, Donald P.. The Principles sf Industrial Process Profos, P., Ektorielle Regeltheorie (2nd edition) Control Soroka, V4.W.. Analog Methods in Computation and SimulakIacCol1, L.A., Fundamental Theory of Senomechanisms tion 1946 Takahashi. Y., The Theop ofAutomatic Colltinl Japanese) (in Ahrendt. \\,R.. Taplin. J.F.. Automatic Regulation Truxal, J.G.,Feedback Theory and Control System Synthesis Tsien, H.S.. Engineering CJbernefics 1947 James, H.J.; Nichols. N.B., Phillips, R.S., Theor). ofServoYoung, .4.J.,Process Control mechanisms Bruns. R.A.. Saunders, R.M., Analysis of Feedback Control Oppelt, W.. Grundgesetre der Regelung System, Seivomechanisms and Automatic Regulators

24

Systems

IEEE Contvol

1955
Chestnut. H.: M a y : R.\K.. Ser?;omechnnislnsandRegulnting Systems Design vol. 2
Thaler. G.J.. Elements of Se~onzecizailisilzs Truxal, J.G.. Automatic Control System S)mthesis Tsypkin, Y.Z., Theory ofRelay Control Systems (in Russian). Van Valkenburg: M E . . !Vetwork Arlnlysis Young, A.J., An Introduction to Process Control Systems Design

C.C. Bissell. Stodola, Hurwitz. and the Genesis of the Stability Criterion. International Journal of Control, vol. 50; pp. 2313-332. 1989. G.S. Brown and D.P. Campbell, Instrument Engineering: Its Gron th and Promise in Process-Control Problems, Fall Meeting of Industrial Instruments and Regulators Division of .4SME, September 1949. Published in Mechanical Engineering. vol. 7 2 , pp. 124-7 and 136: discussion pp. 587-589, 1950. M S . Fagen. ed., AHistoq ofhgineering Science in the Bell System: and J Bell Telephone Laboratories, The Early Years (1875-1925),Murray Hill, S : 1975. AT.Fuller, edition 0fE.J. Rouths Stobiliq qfkfotion,Taylor & Francis, London. 1975. E. Gerecke. Theory and hlathematical Methods of Automatic Control Until 1963,Automatica. ~ o l14. pp. 59-61, 1978. . I. Horowitn, Histoq of Personal Iniohement in Feedback Control Theory. Control Systents Magazine. \ol. 4 pp, 22-3. 1981. . . T.P Hughes. Elrner S p e r q : I n ~ e n t o l and Engineer. Baltimore: Johns Hopkins Press. 1971. R.E Kdlman. On the Genzral Theory of Control Systems. Procrrdings ofrhr FirstlFAC Congress in .vfoscow. vol. 1, pp, 481-492, London. Butterworth. 1960. AV. Khramoi, Histo? ofAurumurion in R~tssin Before 1917 (hIoscow) (in Russiani. 1956: English tranrlation, Jerusalem. 1969. I. Lefkowitn, Don Eckman and His Impact om Process Control, Control .System Magazine, vol. 4: pp. 32-3. 1981. A.G.J. MacFarlane, The Development of Frequency-Response Methods ~nAutomatic Control. IEEE Trans. opt Autontntic Controi, vol. .4C-24, pp. 250-265, 1979. Reprinted in Frequency-Response Methods in Control Svsreins, New York, IEEE Press, 1979. P.R. blasani, Norbert Wiener 1893-1964.Basel. Birkhauser. 1990. 0.May, The Origimof Feedback Coitirol, Cambridge. MA: MIT Press. 1970. 0. Magr. Yankee Practice and Engineering Theor).: Charles T.Porter and the Dynamics of the High Speed Steam Engine. Technologyand . Culture, ~ o l 16. pp, 570-602, 1975. 0. hlayr. Victorian Physicists and Speed Regulation: An Encounter Between Science and Technology, .Totes and Records o f f h e Royal Societx ofLondon. vol. 26. pp, 205.228, 1971. 0. h h y . James Clzrk \laxwell and the Origins of Qbernetics, Isis. vol. 62. pp, 425-134, 1971. D.C. McDonald. Non-Linear Techniques for Improk ing Servo Perfonnance,Proc. Nuti. Electronics Con$, vol. 6, pp. 100-21, 1950. R. Oldenburger. IF.4C, from Idea to Birth, Autonlatitn, \ol. 14, pp. 53-55, 1978. E.F. OKeill, A Histor?. rf Engineering and SrirnrP in the Be// S~stern: Tranmission Technolog) 1192.5-1975),Murray Hdl. NJ: ATKrT Bell Laboratories. 1985. 4 I. Postlethwaite and A.G.J. MacFarlane. . Complex Vcriohie .4pproach to theAnc~i)siro~Lineur.Llitlr~~nriabie FeedbockSymvns, LectureNotes in Control and Information Systems. Springer, 1979. K. Rorentrop, Enrwickhng der modernen Regleun,qsrechnik, Munich: Oldenbourg, 1971, P.H. Sydenham, Measuring I~%struments: of Knowledge and COIZTools irol. Sterenage: Peter Peregrinus, 1979. N. Friener, Generalized Harmonic Analysis. Acta Mafiwnatica, vol. 55. pp. 117-258. 1931. N. Wiener. Extrapolation, Iilterpolafion and Smoothing of Stational). Erne Series with Engineering Applications, MII Press. Cambridge. Mass.. 1949.

June 1996