a seminar on "Advanced Nonlinear PID-Based Antagonistic Control for Pneumatic Muscle Actuators" for control enginerring,It involves the control of pneumatic mucle actuators using Advanced nonlinear PID control.It is a model less control.These actuators are mainly used in humanoid robborts.this leads to more easier and robust control.
3. INTRODUCTION
β’ Advanced Nonlinear PID control is a model-less control
mainly preferred in industrial control applications.
β’ They are used to achieve
οΌ reference tracking.
οΌ disturbance cancellation.
β’ Nonlinearities in the system will lead to significant
increase in modeling complexities so we prefer
AN-PID.
16-02-2015 3
4. s.p e u y
SIMPLE BLOCK DIAGRAM OF A PROCESS WITH AN-PID
CONTROLLER
16-02-2015 4
AN-PID Plant
Nonlinear
adjustments
5. PNEUMATIC MUSCLE ACTUATORS
β’ Pneumatic Muscle Actuatorο PMA
β’ It is a tube like actuator.
β’ They are characterized by decrease in actuating length
when pressurized.
β’ PMA is a device that mimics behaviour of skeletal muscle
β’ It generate force in a nonlinear manner when activated
(pressurized).
β’ It replaced pneumatic cylinders.
16-02-2015 5
7. ADVANTAGES OF PMAs OVER PNEUMATIC
CYLINDERS:
οHigh force-to-weight ratio.
οNo mechanical parts.
οLower compressed air consumption.
οLow cost
οLight weight
οfaster
16-02-2015 7
8. COMPONENTS AND POPERTIES OF
PMA
β’ The basic PMA component is Festo muscle(test
PMA).
β’ Two Festo muscles having same properties of test
PMA are used form antagonist setup.
β’ They are clamped together and are connected with test
PMA via a pulley.
16-02-2015 8
10. β’ All three PMAs are on vertical position and their upper end is
clamped.
β’ Pressure regulators β to control and measure the compressed
air supplied to PMAs.
β’ A pressure sensor is integrated inside pressure regulator to
provide measurement accuracy.
β’ A distance sensor is used to measure the displacement of PMA
in the vertical axis.
β’ A load cell is used to measure force produced from PMA.
β’ Data acquisition βnational instruments USB-6251 DAC .
16-02-2015 10
11. NON LINEAR PROPERTIES OF PMA:-
β’ PMA is having double helix aramid netting, it is covered
by a neoprene threaded coating β tube like formation.
β’ Aluminium bearing are properly attached at the ends of
the aramid-neoprene fibre wrapping.
β’ Cross-sectional view of the PMA is shown in figure.
16-02-2015 11
12. β’ The dual material leads to a non-linear characteristics:-
οViscoelastic properties of the neoprene wrapping
οFriction phenomena between the aramid threads and
neoprene coating .
οIrregular deformation of the tubes.
β’ These properties result in complex hysteretic phenomena.
16-02-2015 12
13. CONVENTIONAL PID
β’ It is the most utilised controller.
β’ It features a feedback control action u(t).
β’ u(t) ο weighted sum of three control parameters
* Proportional term
* Integral term
* Derivative term
β’ They are mathematically formulated as:-
u(t)=π² π· π π +
π
π» π° π
π
π π π π + π» π«
π π(π)
π π
β¦β¦β¦β¦.(1)
16-02-2015 13
14. Contβ¦
β’ Where
πΎ π ο Proportional gain
ππΌ ο Reset time
π π· ο Rate time
e(t)ο Error signal
e(t) =π π (t)-x(t) β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦β¦..(2)
π₯ π(t) : Set point value
x(t) : Process value
β’ Controllers goal is to
* adjust the manipulated variable u(t)
* minimise the process error signal e(t)
16-02-2015 14
15. Contdβ¦
β’ Due to the positioning control problem in PMA actuated
applications different PID controllers were used.
β’ Apart from this a more efficient type of PID controller is
used here.
β’ That is Advanced nonlinear PID .
16-02-2015 15
16. AN-PID
β’ Conventional PID control is considered as ideal.
β’ In cases of highly nonlinear processes like PMA, there is a
need of modifying the conventional PID.
β’ This is to achieve advanced performance.
β’ For this AN-PID was formulated.
β’ Additional degrees of freedom and tuning parameters was
incorporated with conventional PID.
16-02-2015 16
17. FEATURES OF AN-PID
ο Increased flexibility
ο Advanced customizable properties of overall control behaviour
ο Trapezoidal integration and partial derivative action
ο Nonlinear adjustment of the integral action by anti-windup
switch function
ο Gain scheduling mechanism
ο Bumpless transition mechanism
16-02-2015 17
18. FUNDAMENTAL COMPONENTS OF
AN-PID
ο±DERIVATIVE KICK CANCELLATION & NONLINEAR
INTEGRAL ADJUSTMENT
β’ Sudden alteration in β π₯ π β value results in spikes in PID
output.
β’ This is due to response of the derivative term ο Derivative
kick
β’ To avoid the derivative kick derivative term is posed on the
process value βxβ instead of e(t).
β’ So derivative term π’ π(t) is expressed as π’ π·(t)
16-02-2015 18
19. Contdβ¦.
π’ π·(t) = πΎ π π π·
ππ₯(π‘)
ππ‘
β¦β¦β¦β¦β¦β¦β¦. (3)
β’ To avoid the overshoot ο a nonlinear term β h(t) β was
added for adjusting the integral term π’π(t).
β’ Then π’π(t) is denoted as π’πΌ(t).
β’ Where π’πΌ(t) = πΎ π
β(π‘)
π πΌ 0
π‘
π π‘ ππ‘ β¦β¦β¦β¦β¦β¦β¦.(4)
β’ With h(t) = (π₯2
π,πππππ ( π₯2
π,πππππ+ 10π2(t)))
π₯ π,πππππ - Range of set-point value
16-02-2015 19
20. Contdβ¦
ο±TWO DEGREE FREEDOM ERROR MODIFICATION
β’ Additional modes were introduced for the PID tuning
parameters.
β’ The modes selectors are f , q β¬ R.
β’ They are posed on proportional and derivative term is, whereas
integral term is remained unaffectedο to avoid steady state
error.
β’ The error signals are chosen as:-
π π(t) = fπ₯ π(t) - x(t)
ππΌ(t) = π₯ π π‘ - x(t)
π π·(t) = qπ₯ π(t) - x(t) ....(5)
16-02-2015 20
21. Contdβ¦
β’ In conventional PID multiple control demands were
satisfied by using the error mechanism in one- degree of-
freedom manner.
β’ Equation (5) formulates a two-degree-of-freedom for
AN-PID.
β’ This provided advanced flexibility for control design
ο helped in disturbance rejection.
16-02-2015 21
22. β’ Mode selector f β¬ [0,1] ο trade-off between noise rejection and
set-point tracking.
β’ f=1 β error effected action β control emphasis on tracking
reference signal.
β’ f = 0 β measurement effected action β emphasis on disturbance
cancellation.
β’ Mode selector q β¬ [0,1].
β’ q = 1β differentiation on error
β’ q = 0β differentiation on measurement βreduces derivative
kick.
16-02-2015 22
23. Contdβ¦..
ο± ADVANCED NONLINEAR ERROR FUNCTION
β’ To achieve good control behaviour in different error
magnitudes through auto-adjustable gain βnonlinear error
function error squared is used.
π πππππππ (t) =
π π Γ π(π)
π π ,πππππ
........... (6)
β’ This will increase the efficiency of PID algorithm against
low-frequency disturbances which cannot be removed from
the measurement signal.
16-02-2015 23
24. β’ This function drives the βπΎ πβ to lower values as error
decreases and vice versa.
β’ Equation (6) is again modified as
πANβPID(t) =
π(π‘)
π₯ π,πππππ
π Γ π₯ π,πππππ + (1 β π) π(π‘) ..(7)
π β linearity factor
β’ π β¬ R+
and is bounded in π[0,1].
β’ π accounts the increase in πΎ π with respect to error.
β’ Graphical representation between linear error and
modified squared-error is shown in figure.
16-02-2015 24
26. β’ In case of small values of e(t) , the effect of the term
π(π‘) Γ |π(π‘)| π₯ π,πππππ will become negligible β result in
minimum value of πANβPID(t).
πANβPID(t) = 0.3 e(t) ........(9)
β’ Comparing equation (5) and (7) error signals for proportional,
integral and derivative actions are given by
π π
ANβPID(t)=
[ππ₯ π βπ₯ π‘ ]
π₯ π,πππππ
Γ ππ₯ π,πππππ + 1 β π ππ₯ π π‘ β π₯(π‘) (10)
ππΌ
ANβPID
(t)=
π₯ π π‘ βπ₯(π‘)
π₯ π,πππππ
Γ ππ₯ π,πππππ + (1 β π) π₯ π π‘ β π₯(π‘)
16-02-2015 26
28. β’ Discretization of the integral term is by trapezoidal integration.
β’ This is for smoother integral action control during x or π₯ π
variation.
π’πΌ(n) = π’πΌ(n-1) + πΎ π
βπ
π πΌ
π πΌ
ANβPID(π)+ π πΌ
ANβPID(πβ1)
2
h(n) ..(14)
β’ Discrete equivalent derivative term is
π’ π·(n) = πΎ π
π π·
βπ
π π·
ANβPID π β π π·
ANβPID(π β 1) β¦β¦(15)
β’ Discrete control output is given by adding (13) (14) (15)
π ππβπππ(π) = π π·(n) + π π°(n) + π π«(n) .......(16)
16-02-2015 28
29. β’ π’ π΄πβππΌπ·(n) limits between π’ πππ₯ and π’ πππ .
β’ In case of constant error factors integral action drives the
control effort to its extreme values π’ πππ₯ or π’ πππ .
β’ This results in saturated condition β windup.
β’ Windup will cause overshoot phenomena ,this is avoided
by a switch function β by using anti-windup switchβαΊβ.
β’ αΊ will be enhanced with equation (14).
π’πΌ(n) = π’πΌ(n-1) +αΊ πΎ π
βπ
π πΌ
π πΌ
ANβPID(π)+ π πΌ
ANβPID(πβ1)
2
h(n) ... (17)
16-02-2015 29
30. ο± GAIN SCHEDULING
β’ For highly nonlinear process, efficient control
performance is being required throughout their operating
range β gain scheduler.
β’ Then gain scheduler must be incorporated AN-PID loop.
β’ This scheduling has the ability to control parameters πΎ π,
ππΌ and π π· β according to region of operation specified
by π₯ π .
β’ An additional switching signal βiβ is introduced.
16-02-2015 30
31. β’ Switching signal (i) rules the previous switching values of
gain constants.
β’
πΎ π
ππΌ
π π·
=
πΎ π,π
ππΌ,π
π π·,π
for i = 1,2,3,β¦..,N
N β Maximum number of operating regions
ο± BUMPLESS TRANSITION
β’ Bumpless transition is used for smooth transition between
areas of operation .
β’ It act as integral sum of adjustment function.
16-02-2015 31
32. β’ Here π’ π + π’πΌ is kept in kept invariant to parameter
alterations.
β’ Inorder to ensure the invariance during such changes
integral action π’πΌ(n) is being altered.
π’πΌ(n) = π’πΌ(n-1) + πΎ π(n-1)π π
ANβPID(n-1) - πΎ π(n) π π
ANβPID(n)
π’ANβPID
Bumpless
(n) = πΎ π(n-1)π π
ANβPID
(n-1) + πΎ π(n) Γ
[αΊ
βπ
π πΌ
π=1
πβ1
[
π πΌ
ANβPID(π)+π πΌ
ANβPID(πβ1)
2
]h(i-1)
+
π π·
π πΌ
[π π·
ANβPID(n) - π π·
ANβPID(n-1) ] β¦..(18)
16-02-2015 32
34. CONCLUSION
β’ Presented about the advanced and highly adjustable
performance of AN-PID.
β’ This control helps for smooth functioning of PAMs.
β’ In future PMA-actuated applications will be used to
perform various operations (e.g., aligning, pressing,
drilling, gripping, clamping, handling, transporting) .
16-02-2015 34
35. REFERENCES
[1] George Andrikopoulos ,βAdvanced Nonlinear PID-Based Antagonistic
Control for Pneumatic Muscle Actuatorsβ ,IEEE Transactions on industrial
electronics, VOL. 61, NO. 12, DECEMBER 2014.
[2] A. B. Corripio, βTuning of Industrial Control Systemsβ, 2nd ed. Raleigh,
NC, USA: ISA,Jan 2000
[3] K. J. Γ strΓΆm and T. Hagglund, βPID Controllers: Theory, Design and
Tuningβ, IEEE Control Engineering USA:ISA Dec 1995
[4] S. Bennett, βA History of Control Engineeringβ,IEEE Control Engineering
U.K.: IET Jun.1986
16-02-2015 35