Disturbance Rejection and
Insensitive Controllers
Presented by-
Supriya Kumari
Roll no – 260
ME-I
ACR
Key terminology
• Controllers
• Set point tracking
• Controller tuning
• Sensitivity
• Disturbance rejection
Desirable Controller Features
• Stable
• Quick responding
• Adequate disturbance rejection
• Insensitive to model, measurement errors
• Avoids excessive controller action
• Suitable over a wide range of operating
conditions
What is Disturbance?
• Disturbance signals represent unwanted inputs
which affect the control-system’s output, and
result in an increase of the system error.
• eg : Wind, Gradient
Block Diagram
Disturbance Rejection
• To address unexpected forces that cause the
plant to move away from the target value, the
controller uses a function known as disturbance
rejection, which processes the disturbance and
provides commands that correct for these
unknown forces or conditions
What is the Ziegler-Nichols
Rule?
• The Ziegler-Nichols rule is a heuristic PID
tuning rule that attempts to produce good values
for the three PID gain parameters:
• Kp - the controller path gain
• Ti - the controller's integrator time constant
• Td - the controller's derivative time constant
• given two measured feedback loop parameters
derived from measurements:
• the period Tu of the oscillation frequency at the
stability limit : ultimate period
• the gain margin Ku for loop stability: Ultimate
gain
with the goal of achieving good regulation
(disturbance rejection).
Disturbance rejection in control
tuning
• In control tuning, the proportional gain (kp)
determines the amount of disturbance rejection
• The lower the overshoot and settling time for
reference tracking, the more sluggish the disturbance
rejection will be, and vice-versa.
• Tuning the controller involves finding the appropriate
balance between the overshoot and settling time of
the reference tracking function and the response
time of the disturbance rejection function
Zero steady state error
• The effectiveness of a system in disturbance signal
rejection is readily
studied by above topology.
• The response Y(z) to disturbance W(z),
can be found from the closed loop transfer
function
Now we introduce S(z) = Y(z)/W(z)
• S(z) is Sensitivity function of the control system
• To reduce the effect of disturbances, S(exp(jwT))
must be made small over the frequency band of
disturbances
• If constant disturbances are to be suppressed
then S(1) should be made small
• If D(z)GhoG(z) includes an integrator( means
D(z) or GhoG(z) has a pole at z=1) then steady
state error due to constant disturbance is zero.
• This may be seen as follows
• Since for a constant disturbance of amplitude A,
we have
Steady-state value of the output is given by
• Which is equal to zero if D(z)GhoG(z) has a pole
at z=1
Non zero steady state error
• Point where disturbance enters the system is
very important in adjusting the gain of
D(z)GhoG(z)
• The closed loop Transfer function for the
disturbance is
• In this case the steady state error due to
constant disturbance W(z) is not equal to
zero when GhoG(z) has pole at z=1
• This may be seen as follows:
• Let
• Where Q(z) is rational polynomial of z, such that
Q(1) is not equal to zero and Q(1) not equal to
infinity
• D(z) is a controller which doesn’t have pole at
z=1
• Magnitude of error can be reduced by
increasing the controller gain
Disturbance entering through
feedback link
• Above block diagram shows the situation
where measurement noise Wn(z) enters
the system through the feedback link
• The closed loop Transfer function for the
disturbance is
• Measurement noise is transferred to the
output whenever I D(z)GhoG(z)>1
• Large gains of D(z)GhoG(z) will lead to large
output errors due to measurement noise
• This is in conflict with the disturbance
rejection property w.r.t previous configurations
• To solve this problem, generally examine the
measuring instrument and modify the filtering,
so that is satisfies the requirement of control
problem
Disturbance rejection in MATLAB
Plant=0.3s2+0.1s
G = tf(0.3,[1,0.1,0]); % plant model C = pidtune(G,'PI');
pidTuner(G,C)
Insensitive controllers
• Assume that the structure and order of the
process model are chosen exactly, and they
don’t change with time
• Parameter changes wrt nominal parameter
vector θn are assumed.
• Closed loop behaviour for parameter
vector
• θ=θn + ∆θ
• If parameter changes are small then
sensitivity method can be used
• For controller design, both good control
performance and small parameter
sensitivity are required
• Resulting controllers are known as
insensitive controllers
• For large parameter the sensitivity design
is unsuitable
• For the design of insensitive controller the
situation is similar to disturbance signal
rejection
• The larger the gain of the feedback loop
around the parameter
• Lower the sensitivity of the closed loop
transfer function to changes in that
parameter
• The closed loop transfer function corresponding
to nominal parameter vector described by
• Process parameter now changes by an
infinitesimal value ∆θ
• This sensitive function shows relative
changes of input/output behaviour of a
closed loop
• Depend on changes of the process transfer
function
• Small parameter-sensitivity of the closed
loop behaviour can be obtained by making
small in the significant frequency range
References
• [ 1] M.Gopal “Digital Control and State
variable Methods”, McGraw Hill
Education
• Mathwork
Thank You
