2 ArticuloDeInvestigación1-1

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

C'.

opyright © I ~96 IFAC 4c-042


13th Triennial World Congress, San Francisco, USA

History of Automatic Control to 1960: an overview

Stuart Bennett

Department of Automatic Control & Systems Engineering, The University of Sheffield


Mappin Street, Sheffield SI 3m, UK, email: s.bennett@Sheffield.ac.uk

Abstract: An outline of the development of control devices and control theory from the
earliest times is given. Devices discussed include the steam engine governor. process
controllers, and servomechanisms. The impact of the second world war on the devel-
opment of classical design methods and on the introduction of the stochastic approach are
covered.

Keywords: Control applications, control education, control history, control technology.

four main periods as follows:


I. INTRODUCTION
• Early Control: to 1900
Feedback control systems have been used for over 2000 • The Pre-Classical Period: 1900-1940
years, however, the word 'feedback' is a neologism intro- • The Classical Period: 1935-1960
duced in the 1920s by radio engineers to describe para- • Modern Control: post 1955.
sitic, positive feeding back of the amplifier output signal to This paper is concerned largely with the first three peri-
the input circuit, entering into common usage in the Eng- ods, other papers in the session ieal with the more recent
lish speaking world during the latter half of the century. period.
Automatic feedback is found in a wide variety of systems;
2. EARLY CONTROL: TO 1900
Oldenburger (1978), recalling the foundation of IFAC
commented on both the name and the breadth of the sub- Knowledge of Hellenic contra] systems was preserved
ject: within the Islamic culture being rediscovered in the West
I felt that the expression 'automatic control' covered all towards the end of the Renaissance. New inventions and
systems, because all systems involve variables, and one applications of old principles ,'merged during the 18th
is concerned with keeping these variables at constant or century-e.g., Rearnur (1683-1757) proposed several
given varying values. This amounts to concern aoout automatic devices for control1in~; the temperature of incu-
control of these variables even though no actual auto- bators based on an invention of Drebbel (1572-1663). The
matic control devices may be intentionally or otherwise temperature was measured by Ihe expansion of a Jiquid
incorporated in these systems. I was thinking of bio- held in a vessel connected to a V-tube containing mercury.
logical, economic, political as well as engineering sys- A float in the mercury operated an arm which controlled
tems so that J pictured the scope of IFAC as a very the draft to a furnace and hence the rate of combustion.
broad one. Bonnemain (circa 1743-1828) d"vised improved tempera-
This diversity poses difficulties for historians of the sub- ture control systems and based his sensor and actuator on
ject (and for editors of control journals) and the paper does the differential expansion of melals. During the 19th cen-
not attempt to cover all application areas. tury many thermostatic devices were invented, manufac-
tured and sold. These devices WI!re, predominantly, direct
The history of automatic control divides conveniently into

3008
acting controllers, i.e" the power required to operate the his colleague Stodola (1859-1942) who was working on a
control actuator was drawn from the measuring system, turbine control problem.
The most significant 18th century control development was Most 19th century applications were concerned with con-
the steam engine governor whose origins lie in the lift- trolling temperatures, pressures, liquid levels and speed of
tenter mechanism used to control the gap between the rotating machinery: the desire was for regulation and for
grinding-stones in both wind and water mills. Watt (1736- stability. However, growth in the size of ships and naval
1819) realised it could he adapted to govern the speed of guns, and introduction of weapc'I1S such as torpedoes, re-
the rotary steam engine. The Watt governor (1789) had sulted in using steam, hydraulic, and pneumatic power for
several disadvantages: it had only proportional control and position control systems. EngiIlfers began to devise pow-
hence controlled speed at only one operating condition ered steering engines to assist file helmsman. The fIrst of
(leading to comments that it was <a moderator not a con- powered steering engine, designed by Sickels (1853) was
troller'), it had only over a small speed range, and it re- an open loop system. A closed loop steering engine (1866)
quired careful maintenance. Many sought to avoid the off- was designed by J McFarlane Gray for Brunei's steamship
set inherent in the Watt governor and thousands of patents the Great Eastern. ]n France, mound the same time, Far-
were granted.. Typical were the designs of William Sie- cot designed a range of closed loop position control sys-
mens (1823-1883) who substituted integral for proportional tems, He suggested the names 'serva-moteur' or 'moteur
action, producing 'floating' controllers, Porter's loaded asservi' and hence our terms 'servomechanisms' and
governor (1858) could be run at higher speeds and hence 'servomotors' .
develop greater forces to operate an actuator. A little later
Use of electricity brought with it further requirements for
Pickering (1862) and Hartnell (1872) invented spring
control systems. For example, ~(]'c lamps need a constant
loaded governors which operated at higher speeds than the
gap hetween the electrodes, and generally it was helpful to
Watt governor and which were smaller than the Watt and
all users if either the voltage or the current of the electric-
Porter governors.
ity supply was kept constant. Electricity also provided
Reports of 'hunting' of governors appeared early in the additional tools-for measurement, for transmission and
19th cenrury and there were attempts to determine the manipulation of signals, and fc,r actuation. The electric
conditions for stable (non-hunting) operation. Poncelet relay which provided high gain power amplification and
(1788-1867) in 1826 and 1836, and Airy (1801-1892) in the spring biased solenoid which provided (crude) propor-
1840 and 1851 used differential equations to descrihe dy- tional control action were important devices.
namic motion but both met difficulties in fmding the con-
3. PRE-CLASSICAL PERIOD (1900-1935)
ditions for stable hehaviour. Airy (1851), stated the condi-
tions but his report is so terse that one cannot determine Early in the 20th century there 'Nas rapid and widespread
how he arrived at these conditions. Maxwell (1831-1879) use of feedback controllers for voltage, current and fre-
in his famous paper entitled 'On governors' (1868) de- quency regulation; boiler control for steam generation;
scribed how to derive the linear differential equations for electric motor speed control; ship and aircraft steering and
various governor mechanisms At this time mathematicians auto stabilisation; and temperature, pressure and flow
and physicists knew that the stability of a dynamic system control in the process industric!s. Sales of instruments,
was determined by the location of the roots of the charac- particularly measuring, indicating and recording devices
teristic equation, and that a system became unstable when grew rapidly. The variety of control systems was exten-
the real part of a complex root became positive; but how sive, however, most were desigled without any clear un-
to determine the location of the real parts of the complex derstanding of the dynamics of both the system to be con-
roots without fmding the roots of the equation? Maxwell trolled, and of the measuring aId actuating devices used
showed, for 2nd, 3rd and 4th order systems, how to do for control. Some systems were complex, involving
this but for fifth order equations he was only able to give complicated control laws, e.g., the automatic ship steering
two necessary conditions. The paper was little noticed at mechanism devised by Sperry incorporated PID control
the time and it was not until the early years of this cenmry and automatic gain adjustment tt) compensate for the dis-
that the work was assimilated as engineering knowledge. turbances caused when the sea conditions cbanged. In the
The problem was taken up by Routh (1831-1907) whose 1920s, the electricity supply ccmpanies, concerned with
first results were published in 1874. In 1877 he produced economic operation of steam ge aerating boilers, began to
an exteoded treatise on the 'Stability of Motion' in which introduce automatic boiler control systems. These were
he expounded what we now know as the Routh-HuTWitz complex multivariable systems which sought to control
stability criteria. In 1895, the Swiss mathematician Hur- feedwater level, boiler pressuremd CO2 level by manipu-
witz (1859-1919) derived the criteria independently. He lating boiler draught, coal and w;lter feed.
had been asked for help with the mathematical problem by

3009
As use spread two major problems became apparent: (i) a component drift could be reduc<d. On 2 August 1927, he
lack of theoretical understanding and no common lan- sketched a circuit for a negatiVE feedback amplifier. Fol-
guage, and (ii) no simple, easily applied analysis and de- lowing extensive development work practical trials took
sign methods. The available analysis tool was the differ- place in 1930 and the amplifier was used in 1931. Black
ential equation and the application of the still not widely was assisted by Nyquist (1)89-1976) whose paper
known Routh-Hurwitz stability test. A laborious process, 'Regeneration Theory' (1932) l"id the foundations of the
and one which gives no guidance to the designer on the so called Nyquist analysis. This work provided a practical
degree of stability, or what to do to malee the system sta- device-the negative feedback amplifier-and led to a
ble. Engineers became puzzled and confused: controllers deeper understanding of the benefits of negative feedback
which worked well for one application, or for one set of in systems. It also, eventually, led to a method of analys-
conditions, were unsatisfactory when applied to different ing and designing control systems which did not require
systems or different conditions. Problems arose when a the derivation and manipulation of differential equations.
change resulted in a change in the major time constant. and for which experimental data- measured frequency
This frequently caused instability in what had previously response-couid be combined with calculated data; from
been, or seemed to have been. a stable system. Some the combined response the degree of stability of the system
acute observers noted that the best human operators did could be estimated and changes necessary to improve the
not use an on-off approach to control but used both antici- performance could be seen.
pation, backing-off the power as the controlled variable
Contemporaneously with Black's work, Mason of the
approached the set-point, and small, slow adjustments
Foxboro Company developed a pneumatic negative feed-
when the error persisted. Sperry tried to incorporate these
back amplifier. E H Bristol, a founder of the Foxboro
ideas into his devices and for many years Leeds resisted
Company had invented the fl',pper-nozzle amplifier in
attaching simple on-off control outputs to his recorders
1914. Early versions of the amplifier were highly non-
because he realised that this would not provide good con-
linear and extensive modificatic-ns only succeeded in in-
trol.
creasing its linearity to 7% of Ml range. In 1928, Mason
Minorslcy (1922) gave a clear analysis of position control began experimenting with feeding back part of the output
systems and formulated a control law, which we now refer of the amplifier and in 1930 pwduced a feedback circuit
to as three term or PlD control, by observing the way in which linearised the valve operation. This circuit also en-
which a helmsman steered a ship. This work was not abled integral (or reset) action to be easily introduced into
widely known until Minorsky's series of articles appeared the system. In 1931, Foxboroxgan selling its Stabilog
in The Engineer in 1937. But even if designers had been pneumatic controller which offered both linear amplifica-
aware of Minorslcy' s work they would still have lacked tion (using negative feedback) and integral (reset) action.
suitable linear, stable, amplification devices to convert the There was initial market resistan:::e 10 it, on the grounds of
low power signals obtained from measuring instruments to cost and because its behaviour was not understood. Fox-
a power level suitable to operate a control actuator, Slide boro responded with a bulletin explaining the principles of
and spool valves developed during the early pan of the the system in clear and simple It:nns, and stressing how it
century were beginning to provide the solution for hydro- differed from 'narrow band' comrollers.
mechanical systems, although valve overlap and stiction
The electronic and pneumatic ne.gative feedback amplifiers
were problems that had to be overcome, however, there
were the outcomes of work on industrial problems. During
was an impasse with respect to amplifiers for electronic
the same period extensive work was done on analogue
and pneumatic systems.
calculating machines under the direction of Vanevar Bush
By 1920 the amplification problem was a serious obstacle at the MIT. This led to the design of a high performance
to the further development of long distance telephony. servomechanism by Hazen (1901-1980) and his students.
Improvements in cable design and use of impedance load- Hazen also undertook the first major theoretical study of
ing had extended transmission distances but the transconti- servomechanisms. His papers, published in 1934, provided
nental service in the USA was still dependent on amplifi- the starting point for the next ge:1eration of control system
cation. Electronic telephone repeaters in use introduced specialists.
distortion and hence limited the number that could be
4 THE CLASSICAL PERIOD 1935-1950
placed in series. Expansion of network traffic also caused
problems since it meant an increase in bandwidth of the The period 1935-1940 saw adv:mces in understanding of
lines and hence increased transmission loss. Black (1898- control system analysis and design made independently by
1983) working on the problem realised that if some of the several groups in different countries. The best known and
amplification of a high gain amplifier was sacrificed by most influential work came from three groups in the USA.
feeding back part of the output. distortion due to noise and The AT & T Company continued its work on the tele-

3010
phone system and was seeking to extend the bandwidth of brought together mechanical. el,octrical and electronic en-
communication systems and to obtain good frequency re~ gineers and an outcome of thi~ cross-fertilisation was a
sponse characteristics for voice communication channels- recognition that neither the frequency response approach
ideally constant gain over a wide bandwidth with a sharp used by the communication engineers nor the time domain
cut-off and with a smaII phase lag. Engineers working on approach favoured by the mechanical engineers were,
this problem found that if they achieved the desired gain separately. effective design approaches.
characteristic then the phase lag was too large. [n 1940.
Brown at MIT showed how mechanical and electrical sys-
Bode showed that no definite and wtiversal attenuation and
tems could be represented and manipulated using block
phase shift relationship for a physical structure exists but
diagrams. Hall (1943) showed that by treating the blocks
that there is a relationship between a given attenuation
as transfer functions (using Lapl:iI.ce transforms) the system
characteristic and the minimum phase shift that can be
transfer locus could be drawn, and hence the Nyquist test
associated with it. He also introduced the concept of gain
for stability could be used. More importantly the gain and
and phase margins. and the gain-bandwidth limitation.
phase margin could be determined and he introduced M
Mechanical engineers and physicists working in the proc- and N circles which enabled of the closed loop time do-
ess industries began a systematic theoretical study of con- main behaviour to be estimated. Another group working at
trol systems being used, sought to establish a common the Radiation Laboratory at MIT designed the SCR-584
terminology. and tried to develop design methods. They radar system which. linked with the M9 director. was de-
persuaded the ASME to form an Industrial Instruments ployed in south east England and had a high success rate
and Regulators Committee (1936). Several members of against V 1 rockets. The M9 director was designed a BTL
this loose grouping were aware of developments in Ger- group led by Bode and including Blackntan. Lovell and
many and in England. During this period the manufactur- Shannon. The SCR-584 system design led to the Nichols
ers of pneumatic controllers continued to improve and chart design method, work by Phillips on noise in servo-
develop their instruments and by 1940 field adjustable in- mechanisms, and Hurewicz's work on sampled data sys-
struments with PID control were available, e.g.) an im- terns. After the war details of the work were published in
proved version of the Stabilog and the Taylor Fu/scope. In the seminal book Theory of Servomechonisms (James et.
1942. J.G. Ziegler and N.B. Nichols of the Taylor In- al. 1947). The Radiation Laboratory group used phase
strument Companies showed how to fmd the optimum advance circuits in the forward loop to modify the per-
settings for PI and PID control-the so called Ziegler- formance of their control system. Other workers, particu-
Nichols tuning rules. These were extended in the mid- larly in the UK. used minor loop feedback to modify sys-
1950s by Coon. tem response and hence found tile Nyquist approach diffi-
cult. Whiteley (1942) of the British Thomson Houston
The third grouP. located in the Electrical Engineering De-
Company proposed an approach based on plotting the in-
partment of MIT. was led by Hazen and Brown. They
verse functions on a Nyquist diagram, in the same year
used time-domain methods based on operator techniques,
H. T. Marcy independently propc,sed a similar method.
introduced block diagrams. and used the differential ana-
lyser to simulate control systems. Scholarly interchanges Anti-aircraft control raised systc:m design problems since
between MIT and the University of Manchester led to a several units. often designed anc! manufacrured by differ-
differential analyser being built at Manchester and, in ent groups, had to be integrated; overall performance was
1936. Hartree and Porter with Callender of ICI used the dependent not on the performance of individual units but
machine to simulate an industrial control system. on how well they worked together. Difficulties of making
units work together led to a deepening understanding of
The second world war concentrated control work on a few
bandwidth. noise. and non-lineHities in systems. By the
specific problems. Important among these was aiming of
end of the war people such a~ Tustin in England, and
anti-aircraft guns. This is a complex problem involving
PhilIips. Hurewicz. McColl. Minorsky and Stibbitz in the
detection of the position of the aeroplane, calculation of its
USA were concentrating on nail-linear and sampled data
furure position and precise control of a heavy gun. The
systems. Another major development to emerge from the
operation needed up to 14 people to carry out complicated
fire control work during the war was the study of stochas-
observation and tracking tasks in a co-ordinated way and it
tic systems: Wiener (1894-19641 wishing to contribute to
became clear in 1941 that the cumbersome system of re-
the war effort proposed tackling the problem of predicting
laying manually infonnation obtained from radar devices
the furure position of an aircraft. He worked with Bigelow
to gun controllers was not adequate to combat fast aircraft.
on implementing his prediction method and he was disap-
A system was needed in which the radar system was di-
pointed that in the end his system was only able to achieve
rectly linked to the gun director. which was in turn linked
a marginal improvement (less than 10 %) over the system
to the gun position controller. This 'systems' problem
developed empirically by the Bdl Telephone Laboratory.

3011
The work led to the report 'The extrapolation, interpola- evaluating a whole range of performance indicators inclnd-
tion and smoothing of stationary time series with engineer- ing IAE, lSE, ITAE, and ITSE. Sterile arguments devel-
ing applications' (OSRD Report 370, I Feb. 1942), known oped about which the performana:e indicator was the 'best'
as 'the yellow peril' due to its yellow covers and the for- until it was accepted that what was importaut was the
midable difficulty of its mathematics-which was eventu- choice of an appropriate performance indicator for a par-
ally published in the open literature in 1949. ticular application. Additional to error fWlCtion based per-
formance criteria there was, for certain classes of system,
By the end of the war the classical control techniques-
interest in minimising the time to reach a set-point (e.g.
with the exception of the Root Locus design method of
military target seeking servomechanisms and certain
Evans-had been established. The design methndologies
classes of machine tools). McDondald's 'Non-linear tech-
were for linear single input, single output systems. The
niques for improving servo performance' (1950) was fol-
frequency response techniques, using Nyquist, Bode,
lowed during the 1950s by exwnsive work on the time-
Nichols and Inverse Nyquist charts, assessed performance
optimal problem relating to the single controlled variable
in terms of bandwidth, resonance, gain and phase margins
with a saturating control. The problem was studied by
and provided a pictorial view of the system behaviour.
Bushaw (1952), Bellman (1956) aud in 1960 LaSalle gen-
The alternative approach based on the solution of the dif-
eralised all the previous results showing that if optimal
ferential equations using Laplace transform techniques
control exists it is unique and tang-bang. Progress made
expressed performance in terms of rise time, percentage
in this area is summarised in Oldenburger's book Optimal
overshoot, steady-state error and damping. Many engi~
Control (1966). The more difficult problem was how to
neers preferred the latter approach because the perform-
choose the control structure which would give the best
ance was expressed in 'real' terms, Le. the time behaviour
performance and how to define this ~best' performance?
of the system. The disadvantage, of course, is that until
A model of the plant was needed: either a physical-
the development of the root locus method there was no
mathematical balance equations of mass, energy etc. in
simple and easy way in which the designer could relate
which the parameters are functi('ns of the physical data of
parameter changes to time behaviour changes.
the process or 'black box' modds based on experimental
The achievements of the classical era were consolidated measurements, e.g., frequency response in which the pa-
and disseminated in books published during the 1940s and rameters are not directly related the physical data of the
early 1950s. The lEE held a conference in London in 1946 systems.
on radar followed by a further conference on servo-
Work on frequency response ideas and design methods
mechauisms in 1947. The USA govermnent paid selected
continued throughout the 19508. Design methnds for sys-
staff for six months after the war to write up their work.
tems containing non-linearities were developed as were the
An outcome was the Radiation Laboratory Series of
theoretical foundations of sampled-data systems. The
books. The conference on 'Automatic Control' (1951) at
teaching of servomechanisms and control theory spread,
Cranfield, England and the 'Frequency Response Sympo-
initially through special courses run for practising engi-
sium' (1953) in New York marked the start of the transi-
neers and graduate students and then through incorporation
tion period. leading to modern control.
within the standard syllabus of many engineering courses,
Wartime experience demonstrated the power of the fre-
5 MODERN CONTROL
quency response approach: it also revealed the weakness
of design methods which assumed linear, deterministic Some post-war work was influmced by the insights and
behaviour. Real systems are non-linear, real measure- new understandings developed during the war but the tra-
ments contain errors and are contaminated by noise, and in jectory of development was, MacFarlane (1979) argues,
real systems both the process and the environment are largely determined by two factors: (i) the problem which
uncertain. But what design techniques can be used which govermnents saw as important, the launching, manoeu-
allow the designer to consider non-linear and non- vring. guidance and tracking of missiles and space vehi-
deterministic behaviour and to allow for measurement er- cles; aud (ii) the advent of the digital computer. The first
rors and noise? The design problem changed from simply problem was essentially contra of ballistic objects and
achieving a stable controller to achieving the ~best' con- hence detailed physical models could be constructed in
trailer. But what is the 'best' controller? Ziegler & terms of differential equations, i:oth linear and non-linear;
Nichols had shown how to choose the parameters of a also measuring instruments and other components of great
given type of controller to obtain an 'optimum' perform- accuracy and precision could be developed and used. En-
ance of a given control structure. Similarly Whiteley' s gineers working in the aerospac., industries turned to for-
standard forms enabled designers to choose a particular mulating the general differential ~quations in terms of a set
performance for a range of systems. Work was done on of first order equations, hence th·~ 'state-space' approach.

3012
Bellman srodied the problem of determining the allocation hold the first conference in Mos<:ow in 1960. The Moscow
of missiles to targets so as to inflict the maximum damage. Conference was an important and highly visible symbol of
This led him to formulate the 'principle of optimality' and the change in direction that had been slowly developing
to Dynamic Programming. The name was determined by during the 1950s and it is fitting that at the conference
political expediency, the research was supported by the Kalman presented a paper, 'On the general theory of con-
Air Force but the then Secretary of Defense had an aver- trol systems' (Kltlman, 1960) tilat clearly showed that a
sion to the word research and it was assumed would have deep and exact duality existed between the problems of
an even greater aversion to mathematical research . multivariable feedback control and multivariable feedback
. Dynamic' is a word with positive connotations and fIltering and hence ushered in a new treatment of the op-
'programming' was thought to be more acceptable than timal control problem.
plarming. Bellman went on to srudy optimal control the-
BIBLIOGRAPHIC NOTE
my, using the calculus of variations and later, because of
the boundary value problem inherent in the approach, The following works have been .;onsulted in preparing this
sought to formulate deterministic optimisation problems paper.
such that they could be solved using Dynamic Program- Bennett, S. (1976). The emergence of a discipline: aut(}-
ming. His saw that by applying a particular control policy matic control 1940-1960. Automalica, 12, 113-121.
the system would reach a region in state-space with a Bennen, S. (1979). A History of Control Engineering
specified amount of time left. Formulated thus the prob- 1800-1930. Peter Peregrinus, Stevenage.
lem can handled as a multistage decision process. With Bennett, S. (1984). Nicolas Minorsky and the automatic
Dreyfus, he developed computer programmes for numeri- steering of ships. IEEE Control Systems Magazine,
cal solutions to a range of problems. A difficulty with Dy- 4, 10-15.
namic Programming is the dimensionality problem and Bennett, S. (1985). Harold Haz"n and the theory and de-
even now, with computing power far beyond anything sign of servomechanisms. 1nl. J. of Control 42, 989-
available to Bellman, we still need to use approximations 1012.
to handle complex systems. Bennett, S. (1992). The development of process control
instruments: the early years Transactions Newcomen
Dynamical problems involving minimising or maximising
Society. 63, 133-164.
some performance index have 'an obvious and strong
Bennett, S., (1993). A History of Control Engineering
analogy with the classical variational formulations of ana-
1930-1955. Peter Peregrinus, Stevenage.
lytical mechanics given by Lagrange and Hamilton'. The
Bissell, C.c. (1989). Stodola, Hurwitz and the genesis of
generalisation of Hamilton's approach to geometric optics
the stability criterion. Int. J. Control. SO, 2313-2332.
by Pontryagin (1956) in the form of his maximum princi-
James, H.J., Nichols, N.B., Phillips, R.S. (1947). Theory
ple laid the foundations of optimal control theory. This
of Servomeclumisms. McGraw-HiII, New York.
and Bellman's insight into the value and usefulness of the
MacFarlane, A.G.J. (1979). 111e development of fre-
concept of state for the formulation and solution of many
quency-response methods in automatic control. IEEE
control and decision problems led to extensive and deep
TrOJiS. on Automatic Control, AC-24, 250-265.
studies of mathematical problems of automatic control.
Mayr, O. (1970) The Origins ~f Feedback Control. MIT
And, the availability of the digital computer, during the
Press, Cambridge, MA.
late 1950s, made a recursive algorithmic solution possible
Mayr, O. (l971a). Victorian physicists and speed regula-
as opposed to the search for a closed form solution of the
tion: an encounter between science and technology.
classical approach.
Notes and Records of the Royal Society of London,
Atbans has placed origins of what is now referred to as 26, 205-228.
modem control theory as being 1956 and in September of Mayr, O. (I 971b). James Clerk Maxwell and the origins
that year an international conference on automatic control, of cybernetics. Isis, 62, 42$-444.
organised by the joint control committee of the VDI and Mayr, O. (1975). Yankee practice and engineering theory:
VDE was held in Heidelberg, Germany, During the con- Charles T. Porter and the dynamics of the high speed
ference a group of delegates met and agreed to form an steam engine. Technology "nd Culture. 16, 570-602.
international organization to promote progress in the field Oldenburger, R. (1978). IFAC, from Idea to Birth. Auto-
of automatic control. An organising group of seven people matica. 14, 53-55.
was charged with drawing up plans for an international Wiener, N. (1949). Extrapol,tion, interpolation and
federation. The organization, the International Federation smoothing of stationary time series with engineering
of Automatic Control (IFAC), was officially formed at a applicalions. MIT Press. Cambridge, Mass.
meeting held in Paris on September 11 and 12, 1957. At
this meeting the Russian delegate extended an invitation to

3013

You might also like