APPARATUS AND METHOD FOR REAL-TIME SEQUENTIAL QUADRATIC PROGRAMMING IN INDUSTRIAL PROCESS CONTROL SYSTEMS

• US 20130282146A1
• Filed: 04/24/2012
• Published: 10/24/2013
• Est. Priority Date: 04/24/2012
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Lu, Joseph Z.
• Granted Patent: US 9,122,261 B2
• Time in Patent Office: Days
• US Class Current: 700/29
• CPC Class Codes: G05B 19/042 using digital processors G05B19/05 takes precedence

Abstract

A method includes identifying a nonlinear model used by an optimizer to perform optimization operations associated with an industrial process to be controlled. The method also includes generating a Hessian matrix associated with the nonlinear model. The method further includes providing the Hessian matrix to an advanced process controller that uses the Hessian matrix to perform both process control and optimization operations. The Hessian matrix can approximate a nonlinear objective function. The method can also include providing one or more approximated nonlinear constraints, a solution of a quadratic sub-problem that approximates the nonlinear model, or a combination of multiple solutions of the quadratic sub-problem to the advanced process controller. The Hessian matrix can be updated and provided to the advanced process controller during each of multiple iterations. During each iteration, the industrial process can be adjusted to move an operating point of the industrial process towards an optimal ending position.

Claims

1. A method comprising:

identifying a nonlinear model used by an optimizer to perform optimization operations associated with an industrial process to be controlled;

generating a Hessian matrix associated with the nonlinear model; and

providing the Hessian matrix to an advanced process controller that uses the Hessian matrix to perform both process control and optimization operations.

2. The method of claim 1, wherein the Hessian matrix approximates a nonlinear objective function.

3. The method of claim 2, further comprising:

generating an approximation of one or more nonlinear constraints associated with the nonlinear model; and

providing the one or more approximated nonlinear constraints to the advanced process controller.

4. The method of claim 2, further comprising:

generating a quadratic sub-problem that approximates the nonlinear model;

obtaining an optimization solution to the sub-quadratic sub-problem from the optimizer; and

providing the optimization solution to the advanced process controller.

5. The method of claim 2, further comprising:

generating a quadratic sub-problem that approximates the nonlinear model;

generating a first solution to the quadratic sub-problem using sequential quadratic programming;

obtaining a second solution to the quadratic sub-problem from the optimizer;

combining the first and second solutions to generate a combined solution; and

providing the combined solution to the advanced process controller.

6. The method of claim 5, wherein the combined solution comprises a conjugate vector of the first and second solutions.

7. The method of claim 1, further comprising:

updating the Hessian matrix and providing the updated Hessian matrix to the advanced process controller during each of multiple iterations; and

during each iteration, adjusting the industrial process to move an operating point of the industrial process towards an optimal ending position.

8. The method of claim 1, wherein generating the Hessian matrix comprises performing a Quasi-Newton method.

9. The method of claim 1, further comprising:

using an optimization solution collar to resolve dynamic conflicts between the optimizer and the APC controller.

10. The method of claim 1, wherein the advanced process controller performs frequency-zero quadratic programming (QP) operations.

11. An apparatus comprising:

at least one memory configured to store a nonlinear model used by an optimizer to perform optimization operations associated with an industrial process to be controlled; and

at least one processing device configured to;

generate a Hessian matrix associated with the nonlinear model; and

provide the Hessian matrix to an advanced process controller that uses the Hessian matrix to perform both process control and optimization operations.

12. The apparatus of claim 11, wherein the Hessian matrix approximates a nonlinear objective function.

13. The apparatus of claim 12, wherein the at least one processing device is further configured to:

generate an approximation of one or more nonlinear constraints associated with the nonlinear model; and

provide the one or more approximated nonlinear constraints to the advanced process controller.

14. The apparatus of claim 12, wherein the at least one processing device is further configured to:

generate a quadratic sub-problem that approximates the nonlinear model;

obtain an optimization solution to the quadratic sub-problem from the optimizer; and

provide the optimization solution to the advanced process controller.

15. The apparatus of claim 12, wherein the at least one processing device is further configured to:

generate a quadratic sub-problem that approximates the nonlinear model;

generate a first solution to the quadratic sub-problem using sequential quadratic programming;

obtain a second solution to the quadratic sub-problem from the optimizer;

combine the first and second solutions to generate a combined solution; and

provide the combined solution to the advanced process controller.

16. The apparatus of claim 11, wherein the at least one processing device is further configured to update the Hessian matrix and provide the updated Hessian matrix to the advanced process controller during each of multiple iterations.

17. A computer readable medium embodying a computer program, the computer program comprising computer readable program code for:

identifying a nonlinear model used by an optimizer to perform optimization operations associated with an industrial process to be controlled;

generating a Hessian matrix associated with the nonlinear model; and

providing the Hessian matrix to an advanced process controller that uses the Hessian matrix to perform both process control and optimization operations.

18. The computer readable medium of claim 17, wherein the Hessian matrix approximates a nonlinear objective function, and further comprising computer readable program code for:

generating an approximation of one or more nonlinear constraints associated with the nonlinear model; and

providing the one or more approximated nonlinear constraints to the advanced process controller.

19. The computer readable medium of claim 17, wherein the Hessian matrix approximates a nonlinear objective function, and further comprising computer readable program code for:

generating a quadratic sub-problem that approximates the nonlinear model;

obtaining an optimization solution to the quadratic sub-problem from the optimizer; and

providing the optimization solution to the advanced process controller.

20. The computer readable medium of claim 17, wherein the Hessian matrix approximates a nonlinear objective function, and further comprising computer readable program code for:

generating a quadratic sub-problem that approximates the nonlinear model;

generating a first solution to the quadratic sub-problem using sequential quadratic programming;

obtaining a second solution to the quadratic sub-problem from the optimizer;

combining the first and second solutions to generate a combined solution; and

providing the combined solution to the advanced process controller.