1. Sébastien Cabaret - October 2007 1 From regulation basics to advanced control

2. Sébastien Cabaret - October 2007 2 Overview Regulation: what is a control loop? What is a PID controller? What is advanced control? Identifying, Modeling …. Tuning Advanced control example: predictive control Schneider tool for Modeling and Tuning available in ITCO Application for GCS: MultiController

3. Sébastien Cabaret - October 2007 3 Regulation: what is a control loop? “I want to see a measured value which corresponds to my request”

4. Sébastien Cabaret - October 2007 4 Regulation: what is a control loop? Reaction: The water temperature reacts on heating power changes Temperature Sensor TE Controller or human! Acts on heating power (4-20mA) (0-500W) Desired Temperature (ex: 150C)

5. Sébastien Cabaret - October 2007 5 Regulation: what is a control loop? Control Loop system Representation Example: Open Loop representation

6. Sébastien Cabaret - October 2007 6 Regulation: what is a control loop? Control Loop system Representation Example: Closed Loop representation

7. Sébastien Cabaret - October 2007 7 Reaction: The water temperature reacts on heating power changes Temperature Sensor (y) TE C(p) Acts on heating power (u, 4-20mA) Desired Temperature (SP, 150C) G(p)

8. Sébastien Cabaret - October 2007 8 What is a PID controller? PID means Proportional, Derivative Integrative. In a classic control loop system, the PID is the controller placed before the process:

9. Sébastien Cabaret - October 2007 9 What is a PID controller? PID Elementary actions Proportional

10. Sébastien Cabaret - October 2007 10 What is a PID controller? PID Elementary actions Integrative Ti is the coefficient given to increase or decrease the integrative action

11. Sébastien Cabaret - October 2007 11 What is a PID controller? PID Elementary actions Derivative Td is the coefficient given to increase or decrease the derivative action

12. Sébastien Cabaret - October 2007 12 What is a PID controller? PID Elementary actions Sum up

13. Sébastien Cabaret - October 2007 13 What is advanced control? Advanced Control - Sébastien Cabaret – 9 Feb. 2006 Basic corrections PID Other strategies Advanced strategies GPC, PFC, RST, IMC… Others Fuzzy, Neuronal network, … Need for process identification System complexity Advanced Control

14. Sébastien Cabaret - October 2007 14 Identifying, Modeling… Tuning Process to tune a controller We should have the knowledge of the system We should give information to the controller for its tuning System information Controller parameters TUNING Ex: P,I and D for PID Data acquisition Identification Modeling Methods

15. Sébastien Cabaret - October 2007 15 Advanced control example: predictive control The predictive control method is an advanced control strategy It is a good compromise between performance and complexity It is based on a model for the prediction of the process output and on a determinate horizon It also uses a reference trajectory to attempt the desire response Several predictive controls exist due to various mathematical approaches of automation people.

16. Sébastien Cabaret - October 2007 16 The predictive control is closed to human driver behavior The controller contains the model of the process to drive The model allows to predict the effect of the action to the system output The driver has built a «mental picture» of its car behaviors He knows the efficiency of the brakes and knows the effect to his car A process model is integrated into the controller SetPoint The controller has the system knowledge and is able to calculate future action to have a desire output behavior Action Process Output

17. Sébastien Cabaret - October 2007 17 The model used by the controller is a dynamic representation of the input/output relationships (ex: mental model of the car vs. the road) The reference trajectory is known by the controller (ex: car trajectory) The horizon definition is specified (ex: 20 seconds) Predictive Control

18. Sébastien Cabaret - October 2007 18 Model Reference trajectory Real trajectory Horizon Set Point

19. Sébastien Cabaret - October 2007 19 Future

20. Sébastien Cabaret - October 2007 20 Schneider tool for Modeling and Tuning available in ITCO Optireg Schneider PLC Some predictive algorithms PID DataStore

21. Sébastien Cabaret - October 2007 21 Application: MultiController object in GCS project

22. Sébastien Cabaret - October 2007 22 Application: MultiController object in GCS project The MultiController is a standard UNICOS object for Schneider PLC and PVSS SCADA system Replace UNICOS PID controller UNICOS compatible (modes, connection, hierarchy) It has a single interface for all regulation algorithms The design allows the addition of new control loop algorithms without changing the object interface It has been design to offer a recipe mechanism. It allows the process expert to keep and reuse pertinent sets of tuning parameters IF1 PID MultiController RST DC3 SF1 PFC generalized Smith Predictor

23. Sébastien Cabaret - October 2007 23 Application: MultiController object in GCS project The MultiController is a standard UNICOS object for Schneider PLC and PVSS SCADA system It has a unique Human Machine Interface with different views It is composed of a synoptic, trend views, navigation buttons It allows a global control of the regulation loop via a centralized object representation in the HMI with different views

24. Sébastien Cabaret - October 2007 24 MultiController operation under PVSS

25. Sébastien Cabaret - October 2007 25 MultiController future application: adaptive control System to control Online Model Identification under PLC u y Model representation: Ex: first order in discrete approach B 1.z -1 H(z)= 1+A 1.z -1 GPC tuning Mechanism in PLC (Predictive strategy) A 1, B 1 GPC Parameters Set Point MultiController

26. Sébastien Cabaret - October 2007 26 Advanced Control Questions