Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad...

18
Chapter 9 Chapter 9 Wireless Model Predictive Control

Transcript of Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad...

Page 1: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Chapter 9Chapter 9

Wireless Model Predictive Control

Page 2: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

MPC Simulation of Measurement Value on MPC Simulation of Measurement Value on Detection of Bad Status Detection Detection of Bad Status Detection MPC Simulation of Measurement Value on MPC Simulation of Measurement Value on Detection of Bad Status Detection Detection of Bad Status Detection

In many recent MPC designs a similar mechanism is used to facilitate MPC operation over a predefined period of time using a simulated measurement when a wired measurement failure is indicated by Bad Status

Page 3: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

MPC Simulation of Measurement Value on MPC Simulation of Measurement Value on Detection of Constant StatusDetection of Constant StatusMPC Simulation of Measurement Value on MPC Simulation of Measurement Value on Detection of Constant StatusDetection of Constant Status

The same principle of MPC using a simulated measurement is applied as well for lab measurements that are available at irregular periods of time with a much slower update rate than the MPC scan rate

Page 4: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Setting MPCPro Action on Detection of Bad Setting MPCPro Action on Detection of Bad or Constant Status or Constant Status Setting MPCPro Action on Detection of Bad Setting MPCPro Action on Detection of Bad or Constant Status or Constant Status

MPCPro operation is managed by the measurement status.

The status of an Analog Input (AI) measurement used in MPC configuration as a controlled or constrained variable (CV) defines whether MPC uses an AI measurement or a simulated process value.

The maximum time for using a simulated process value and the type of MPC Fail mode if the AI output status is Bad are defined during the MPC configuration process (Figure 9-3), where the selected Fail mode type is Local.

Page 5: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

MPCPro Operator Screen Showing How Much MPCPro Operator Screen Showing How Much Time Is Left to Operate Using Simulated AI ValueTime Is Left to Operate Using Simulated AI ValueMPCPro Operator Screen Showing How Much MPCPro Operator Screen Showing How Much Time Is Left to Operate Using Simulated AI ValueTime Is Left to Operate Using Simulated AI Value

AI Bad or Constant status is indicated on the MPCPro operator screen by the timer symbol and an indication of the time left for operation in Auto mode

Page 6: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Principles of Managing AI Status for Principles of Managing AI Status for Wireless MPC OperationWireless MPC OperationPrinciples of Managing AI Status for Principles of Managing AI Status for Wireless MPC OperationWireless MPC Operation

For enabling wireless MPCPro operation it is important that the AI develop an appropriate status depending whether a new measurement value has been communicated or the last communicated value is being held

AI status should be Good over a period of somewhat more than one MPC scan when a new communicated value is detected; otherwise, AI status should be Constant.

Page 7: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

An An Example of Code for AI Status Example of Code for AI Status Generation for Use in MPCGeneration for Use in MPCAn An Example of Code for AI Status Example of Code for AI Status Generation for Use in MPCGeneration for Use in MPC

MPCPro will work with wireless measurements, provided the wireless measurements develop an AI status that triggers simulation. An example of the custom code added to AI measurement processing before the measurement value and status are access by MPCPro is shown in

Page 8: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Use of Simulated Measurement in Slower Use of Simulated Measurement in Slower SubmodelSubmodelUse of Simulated Measurement in Slower Use of Simulated Measurement in Slower SubmodelSubmodel

The wireless MPC concept may be applied as well to the implementation of multi-rate MPC control. When the fastest scan coincides with the slower scan the real measurements are used to update the models that are used to predict simulated values.

Page 9: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Bottom Temperature Step Response – Wireless Bottom Temperature Step Response – Wireless MPC with 8 Second Measurement UpdateMPC with 8 Second Measurement UpdateBottom Temperature Step Response – Wireless Bottom Temperature Step Response – Wireless MPC with 8 Second Measurement UpdateMPC with 8 Second Measurement Update

Testing was conducted using a simplified Divided Wall Column process model. The response using a wireless transmitter with 8 second update rate is shown below to a step change in the bottoms temperature setpoint

Page 10: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Bottom Temperature Step Response – Wireless Bottom Temperature Step Response – Wireless MPC with a 16 Second Measurement UpdateMPC with a 16 Second Measurement UpdateBottom Temperature Step Response – Wireless Bottom Temperature Step Response – Wireless MPC with a 16 Second Measurement UpdateMPC with a 16 Second Measurement Update

The step response trend of wireless MPC does show small “bumps” when a new measurement value is transmitted and the process model is corrected. The “bump” size depends on the model accuracy and how unmeasured disturbances affected the trended process output.

Page 11: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Exercise: Wireless Model Predictive ControlExercise: Wireless Model Predictive ControlExercise: Wireless Model Predictive ControlExercise: Wireless Model Predictive ControlThis workshop provides several exercises that are used to explore wireless MPC operation. A simplified process model of a divided wall column DWC) is used to demonstrate how wireless MPC performance differs from wired MPC performance.

Step 1: Set MPC for wired operation and open the PredictPro Operate application to view the divided wall column MPC function block

Step 2: Reset the control performance calculation and then make a 10% setpoint change for Top Temperature and observe the response trend.

Step 3: Record IAE and number of communications recorded for this test.

Step 4: Using the COM_SEL parameter in the test module, enable Window wireless measurement update with a period of 16 sec, default period of 32 sec and 1 percent deadband.

Step 5: Perform steps 2-3 for wireless MPC operation – compare respective performance of the wired and wireless operation.

Page 12: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Process: Wireless Model Predictive ControlProcess: Wireless Model Predictive ControlProcess: Wireless Model Predictive ControlProcess: Wireless Model Predictive Control

A simulation of a divided wall column (DWC) is used in this workshop to demonstrate how wireless measurement inputs are applied in model predictive control (MPC).

Page 13: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Enabling Wireless SimulationEnabling Wireless SimulationEnabling Wireless SimulationEnabling Wireless Simulation For the workshop simulation of wireless

communication, the COL_SEL parameter is used to enable and disable wireless communication

Page 14: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Model Predictive Control Operation PrincipleModel Predictive Control Operation PrincipleModel Predictive Control Operation PrincipleModel Predictive Control Operation Principle An MPC controller is shown below for a process with two inputs and one

output, in a form that allows one to see the analogy with a typical feedback control loop. The process has a manipulated variable (MV) and a disturbance variable (DV) on the input and a controlled variable (CV) on the output.

Page 15: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Illustration of MPC Controller OperationIllustration of MPC Controller OperationIllustration of MPC Controller OperationIllustration of MPC Controller Operation The process model computes a predicted trajectory of the controlled variable

(CV) that is the process output. After this trajectory is corrected for any mismatch between the predicted value and an actual measured value of the controlled variable, the predicted trajectory is subtracted from the future trajectory of the setpoint to form an error vector as shown.

Page 16: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Multivariable MPC Controlled Generic Process Multivariable MPC Controlled Generic Process Multivariable MPC Controlled Generic Process Multivariable MPC Controlled Generic Process The advantages of MPC are most evident when it is used as a

multivariable controller. A generic multivariable process controlled by MPC is presented as a black box.

Page 17: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

MPC Modifications for Wireless OperationMPC Modifications for Wireless OperationMPC Modifications for Wireless OperationMPC Modifications for Wireless OperationThe process model that is the basis for Model Predictive Control can also be used in a simple way for implementing wireless MPC. For wireless operation, MPC must be modified in the following way :

1.The MPC internal model should be applied for developing simulated measurement values.2.The model prediction is not corrected until a new measurement value is available.3.Process disturbance inputs (DV) should use the last measurement value until a new wireless measurement value is available.

Page 18: Chapter 9 Wireless Model Predictive Control. MPC Simulation of Measurement Value on Detection of Bad Status Detection  In many recent MPC designs a similar.

Process Modeling in MPC OperationProcess Modeling in MPC OperationProcess Modeling in MPC OperationProcess Modeling in MPC OperationAt any time instance k, wired MPC updates the process output prediction in three steps

1. The prediction made at the time k-1 (the bottom dotted curve) is shifted one scan to the left.

2. A step response, scaled by the current change in the process input, is added to the output prediction.

3. The prediction curve is moved to the point to match the current measured process output.