Pneumatic systems offer a number of advantages: typically they are fast,
rugged, simple to maintain and low cost. To move a part from one fixed
position to another, pneumatic cylinders with fixed stops and switching
valves are a natural choice. This situation changes if a stop in between the
end-positions is required. This situation arises for instance in material han-
dling when fault parts have to be ejected to another container than OK
parts. This task requires either a multi-position cylinder or a closed-loop
positioning system.4575
Due to fact that electric DC motors with ballscrew drives can be very
easily controlled they have been used in the past. Today there are more
electric alternatives: controlled AC motors are available as off-the-shelf
products as are electric linear drives. Due to the compressibility of air, ad-
verse friction characteristics and low damping, pneumatic drives require
sophisticated controllers. But they also offer advantages: the compressibil-
ity of air provides a “cushioning” effect which is important in applications
like blow moulding or glass forming (Han and Alleyne 2000). They can
also be used in a vacuum without generating heat; a situation that arises in
semiconductor production (Kagawa et al. 2000b). The reported accuracy in
that application is ± 0.2 µm.
First theoretical and experimental results of closed-loop, position con-
trolled pneumatic systems were presented in 1954 by Shearer. He did a
theoretical analysis of a pneumatic cylinder, a proportional directional
control valve and a mechanical proportional controller. To stabilise the
system, he used gas tanks which had laminar resistances at the inlet ports.
Parallel to the theoretical and experimental work he made a number of
simulation runs on an analogue computer. His system worked with a much
higher pressure than used in industrial applications today and is in this re-
spect different from today’s drives.
Burrows and Webb (1966) extended Shearer’s analysis using the root-
locus technique from linear control theory.
In 1969 Burrows described a system structure that was very similar to
those used for later work by others: “The valve is operated by a torque
motor fed by a controller which sums the effect of position, velocity and
transient pressure feedback”. This means there was an electrically operateddirectional control valve and a feedback of the control signal position and
the additional signals velocity and pressure change. The next important
contribution was published by Barker (1976). He used the Luenberger ob-
server to reconstruct the velocity and acceleration signals from the meas-
ured position signal and could thus omit two sensors. The additional feed-
back of these two signals significantly improved the performance of his
system, a linear pneumatic actuator in a missile flight control system.
Heinen (1976) studied electro-pneumatic control loops. He used a com-
mercial hydraulic servo valve and a hydraulic cylinder with a labyrinth
seal because dedicated pneumatic components were not available at that
time. He found that the single loop proportional position control could be
dramatically improved by the use of additional signals, e.g. the load pres-
sure.
In 1983 Schwenzer did a systematic study of the then known control
laws for pneumatic drives. Starting with single loop feedback, he added
additional feedback signals. Finally, he studied state space controllers
where he looked both at full order state observers and differentiating cir-
cuits to reconstruct the piston velocity and acceleration. Several other theo-
retical and experimental studies followed. Using different approaches, the
best achieved accuracy was 0.01 mm (Schwenzer 1983; Nguyen 1987:76; 气动系统英文文献和翻译:http://www.751com.cn/fanyi/lunwen_1308.html