Control Using Wireless Measurements
PIDPlus vs an Observer with PID
Presenters
 Terry Blevins
 Mark Nixon
Introduction
In this workshop we show how an observer may be modified for
use with a wireless measurement and compare control
performance to that achieve using DeltaV PIDPlus.
 Background on observers and development of PIDPlus
 Modification of Kalman filter and Smith Predictor for use with
a Wireless Measurement in Control applications
 Test Results –PIDPlus, Modified Kalman Filter and Smith
Predictor
 Summary
 Where To Get More Information
Traditional Approach in Control
 Controllers are commonly designed to over-sample the
measurement by a factor of 2-10X.
 To minimize control variation feedback control is executed 4X
to 10X times faster than the process response time i.e.
process time constant plus process delay.
 However, the underlying assumption in the control design
(using z transform, difference equations) and digital
implementation of the PID is that the algorithm is executed on
a periodic basis.
Control Data Sample Rate
Control Using Wireless Measurements
To minimize the power consumed in communicating new measurement values
the transmitter may use Window communication according to the following
rules:
 The transmitter will periodically sample the measurement 4-10x faster than
the process response time.
 If the magnitude of the difference between the new measurement value
and the last communicated measurement value is greater than a specified
resolution or if the time since the last communication exceeds a refresh
time the new value is communicated.
When the measurement is not updated, the calculated reset and derivative
action may not be appropriate.
PIDPLUS for Wireless Control
 The PID is
restructured to
reflect the reset
contribution for the
expected process
response since the
last measurement
update.
 The rate contribution
is recomputed and
updated only when
a new measurement
is received - using
the elapsed time
since the last new
measurement.
Use of Observers in Wireless Control
The DeltaV Future architecture team has investigated
whether an observer can be used for PID control with a
wireless measurement. Two types of observers were
considered:
 Kalman Filter
 Smith Predictor
It was necessary to modify these observers to work correctly
in control using a PID and wireless measurement
Kalman Filter - Modificatons for Noise
with Non-zero Mean
 Composite is available
through Application
exchange for use with a
continuous measurement
 See workshop 124381,
Emerson Exchange 2013,
“Addressing Control in the
Presence of Process and
Measurement Noise”
Kalman Filter - Modificatons for Wireless
Measurement and Noise with Non-zero Mean
 The calculation of
the residual must be
modified to account
for the slow, non-
periodic update of
the process
measurement
Application of Smith Predictor With PID
 Smith Predictor is a
standard module in
the DeltaV library.
 Most commonly used
in continuous process
control applications
that are dominated by
large deadtime.
Smith Predictor – Modification for Wireless
Measurement
 The calculation of the
residual must be
modified to account for
the slow, non-periodic
update of the process
measurement
Comparing Performance - Test Conditions
A first order plus deadtime process was used to compare the performance of PIDPlus
to PID with observers modified for wireless control.
 Process Gain = 1
 Process Time Constant = 6 sec
 Process Deadtime = 2 sec
The PI control was tuned for a lambda factor of 1.
 GAIN = 1/Process Gain
 RESET = Process Time Constant + Process Deadtime
The process input and output were scaled 0-100%. The simulated wireless transmitter
is configured for 1% change and 10 sec default period using Windowed
communications. The module execution rate was set at 0.5 sec.
Test Environment – Wireless Control Using
Kalman Filter vs wired PID
Modified Kalman vs. Wired PID
Test Environment – Wireless Control Using
Smith Predictor vs Wired PID
Modified Smith Predictor vs Wired PID
Test Environment – Wireless Control Using
PIDPlus vs Wired PID
PIDPlus vs Wired PID
Wireless Control Performance
 Test were performed
under ideal conditions
where the model used in
the Kalman filter and Smith
predictor are based on the
process gain and
response.
 All methods provide good
control. As measured by
IAE, the PIDPlus did
significantly better than the
modified Smith Predictor
and slight worse that the
modified Kalman filter.
 PIDPlus does not require
that the user entry process
model parameters.
Conclusion
 A method for dealing with non-periodic measurement updates
is a requirement when closed loop control is implemented using
wireless transmitters.
 The DeltaV PIDPlus is designed to use non-periodic, slow
measurement updates. Alternative approach based on using
PID with a modifed Kalman filter or a Smith Predictor may also
be used but requires setup of the process model.
 The PIDPlus is preferred for wireless control since it is more
robust for changes in process gain, dynamics, and noise level.
Where To Get More Information
 Workshop 12-4381, Emerson Exchange 2013, Wojsznis and Blevins, “Addressing
Control in the Presence of Process and Measurement Noise” (Using Kalman Filter)
 E. Cheever. “Introduction to Kalman Filter”,
http://www.swarthmore.edu/NatSci/echeeve1/Ref/Kalman/ScalarKalman.html
 Application Exchange, “Kalman Filtering in DeltaV – Control in the Presence of
Noise”, http://www2.emersonprocess.com/en-
US/brands/deltav/interactive/Pages/Interactive.aspx
 S. Han, X. Zhu, K.M. Aloysius, M. Nixon, T. Blevins, D. Chen, “Control over
WirelessHART Network”, 36th Annual Conference of the IEEE Industrial Electronics
Society, 2010
 F. Siebert, and T. Blevins, “WirelessHART Successfully Handles Control”, Chemical
Process, January, 2011
Thank You for Attending!
Enjoy the rest of the conference.

Control using wireless measurements

  • 1.
    Control Using WirelessMeasurements PIDPlus vs an Observer with PID
  • 2.
  • 3.
    Introduction In this workshopwe show how an observer may be modified for use with a wireless measurement and compare control performance to that achieve using DeltaV PIDPlus.  Background on observers and development of PIDPlus  Modification of Kalman filter and Smith Predictor for use with a Wireless Measurement in Control applications  Test Results –PIDPlus, Modified Kalman Filter and Smith Predictor  Summary  Where To Get More Information
  • 4.
    Traditional Approach inControl  Controllers are commonly designed to over-sample the measurement by a factor of 2-10X.  To minimize control variation feedback control is executed 4X to 10X times faster than the process response time i.e. process time constant plus process delay.  However, the underlying assumption in the control design (using z transform, difference equations) and digital implementation of the PID is that the algorithm is executed on a periodic basis.
  • 5.
  • 6.
    Control Using WirelessMeasurements To minimize the power consumed in communicating new measurement values the transmitter may use Window communication according to the following rules:  The transmitter will periodically sample the measurement 4-10x faster than the process response time.  If the magnitude of the difference between the new measurement value and the last communicated measurement value is greater than a specified resolution or if the time since the last communication exceeds a refresh time the new value is communicated. When the measurement is not updated, the calculated reset and derivative action may not be appropriate.
  • 7.
    PIDPLUS for WirelessControl  The PID is restructured to reflect the reset contribution for the expected process response since the last measurement update.  The rate contribution is recomputed and updated only when a new measurement is received - using the elapsed time since the last new measurement.
  • 8.
    Use of Observersin Wireless Control The DeltaV Future architecture team has investigated whether an observer can be used for PID control with a wireless measurement. Two types of observers were considered:  Kalman Filter  Smith Predictor It was necessary to modify these observers to work correctly in control using a PID and wireless measurement
  • 9.
    Kalman Filter -Modificatons for Noise with Non-zero Mean  Composite is available through Application exchange for use with a continuous measurement  See workshop 124381, Emerson Exchange 2013, “Addressing Control in the Presence of Process and Measurement Noise”
  • 10.
    Kalman Filter -Modificatons for Wireless Measurement and Noise with Non-zero Mean  The calculation of the residual must be modified to account for the slow, non- periodic update of the process measurement
  • 11.
    Application of SmithPredictor With PID  Smith Predictor is a standard module in the DeltaV library.  Most commonly used in continuous process control applications that are dominated by large deadtime.
  • 12.
    Smith Predictor –Modification for Wireless Measurement  The calculation of the residual must be modified to account for the slow, non-periodic update of the process measurement
  • 13.
    Comparing Performance -Test Conditions A first order plus deadtime process was used to compare the performance of PIDPlus to PID with observers modified for wireless control.  Process Gain = 1  Process Time Constant = 6 sec  Process Deadtime = 2 sec The PI control was tuned for a lambda factor of 1.  GAIN = 1/Process Gain  RESET = Process Time Constant + Process Deadtime The process input and output were scaled 0-100%. The simulated wireless transmitter is configured for 1% change and 10 sec default period using Windowed communications. The module execution rate was set at 0.5 sec.
  • 14.
    Test Environment –Wireless Control Using Kalman Filter vs wired PID
  • 15.
  • 16.
    Test Environment –Wireless Control Using Smith Predictor vs Wired PID
  • 17.
  • 18.
    Test Environment –Wireless Control Using PIDPlus vs Wired PID
  • 19.
  • 20.
    Wireless Control Performance Test were performed under ideal conditions where the model used in the Kalman filter and Smith predictor are based on the process gain and response.  All methods provide good control. As measured by IAE, the PIDPlus did significantly better than the modified Smith Predictor and slight worse that the modified Kalman filter.  PIDPlus does not require that the user entry process model parameters.
  • 21.
    Conclusion  A methodfor dealing with non-periodic measurement updates is a requirement when closed loop control is implemented using wireless transmitters.  The DeltaV PIDPlus is designed to use non-periodic, slow measurement updates. Alternative approach based on using PID with a modifed Kalman filter or a Smith Predictor may also be used but requires setup of the process model.  The PIDPlus is preferred for wireless control since it is more robust for changes in process gain, dynamics, and noise level.
  • 22.
    Where To GetMore Information  Workshop 12-4381, Emerson Exchange 2013, Wojsznis and Blevins, “Addressing Control in the Presence of Process and Measurement Noise” (Using Kalman Filter)  E. Cheever. “Introduction to Kalman Filter”, http://www.swarthmore.edu/NatSci/echeeve1/Ref/Kalman/ScalarKalman.html  Application Exchange, “Kalman Filtering in DeltaV – Control in the Presence of Noise”, http://www2.emersonprocess.com/en- US/brands/deltav/interactive/Pages/Interactive.aspx  S. Han, X. Zhu, K.M. Aloysius, M. Nixon, T. Blevins, D. Chen, “Control over WirelessHART Network”, 36th Annual Conference of the IEEE Industrial Electronics Society, 2010  F. Siebert, and T. Blevins, “WirelessHART Successfully Handles Control”, Chemical Process, January, 2011
  • 23.
    Thank You forAttending! Enjoy the rest of the conference.