Model predictive controller

• US 6,056,781 A
• Filed: 02/06/1998
• Issued: 05/02/2000
• Est. Priority Date: 10/13/1992
• Status: Expired due to Term

• Current Assignee: Dow Global Technologies LLC (Dow, Inc.)
• Original Assignee: The Dow Chemical Company (Dow, Inc.)
• Inventors: McCroskey, Patrick S., Steckler, David K., Wassick, John M., McDonough, John J.
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): Sergent, Douglas W.
• Application Number:
  US09/020,141
• Time in Patent Office: 816 Days
• Field of Search: 395/500.23, 395/500.27, 395/500.33, 364/149, 364/151, 364/153, 364/165, 703/2, 703/6, 703/12, 700/28, 700/29, 700/30, 700/31, 700/38
• US Class Current: 703/12
• CPC Class Codes: G05B 13/048 using a predictor

Abstract

A model predictive controller for a process control system which includes a real-time executive sequencer and an interactive modeler. The interactive modeler includes both a process model and an independent disturbance model. The process model represents the dynamic behavior of the physical process, while the disturbance model represents current and future deviations from the process model. The interactive modeler estimates current process states from the process model and input data received from the executive sequencer. The executive sequencer then projects a set of future process parameter values, which are sought to be controlled, over a predetermined control horizon. The interactive modeler then solves a set of equations as to how the physical process will react to control changes in order to determine an optimized set of control changes. As a result, the process control system will be able to accurately track a predetermined set-point profile in the most effective and cost efficient manner.

Claims

1. A method of executing real-time control of an apparatus, comprising the steps of:

receiving input signals which represent measured parameter values from a plurality of sensors in said apparatus;

periodically transmitting a time-recorded set of said measured parameter values to an interactive modeler which includes

a nonlinear process model having an on-line version for use in said real-time execution of control wherein said on-line version is derived from dynamic behavior in said process,

an independent disturbance model having an on-line version for use in said real-time execution of control wherein said on-line version provides an error model for predicting future model error,

an integrated model building function for use in building a future version of said process model and a future version of said disturbance model, and

an integrated adaptive function for selectively enabling change of the process model on-line version and, alternatively, the disturbance model on-line version into a respective future version and for enabling said respective future version to replace the respective on-line version at a user-specified moment and thereby effect a new respective said on-line version wherein said change is enabled while said interactive modeler is in on-line use in said real-time execution of control;

estimating the current state of at least one unmeasured process parameter in said interactive modeler from said process model;

projecting a desired set of desired directly controlled parameter values and at least one desired indirectly controlled parameter value over a predetermined control horizon;

periodically estimating how said apparatus will react to proposed changes to the value of at least one manipulated parameter over said control horizon, and determining a set of current and future manipulated parameter values which will minimize deviations from said desired set; and

causing a process control device in said apparatus to implement said manipulated parameter values.

2. The method according to claim 1, further comprising the steps of

testing the validity of said input signals, said current state of each estimated unmeasured process parameter, and each current manipulated parameter value;

and

causing said process control device to implement only manipulated parameter values which are determined to be valid.

3. The method according to claim 1, wherein said step of periodically estimating and determining determines an optimized set of current and future manipulated parameter values.

4. The method according to claim 1, wherein said disturbance model is a step response variable disturbance model for predicting future disturbances and wherein said disturbance model is independent of said process model.

5. The method according to claim 2, wherein said step of periodically estimating and determining determines an optimized set of current and future manipulated parameter values.

6. The method according to claim 2, wherein said disturbance model is a step response variable disturbance model for predicting future disturbances and wherein said disturbance model is independent of said process model.

7. The method according to any of claims 1, 2, 3, 4, 5 or 6 wherein said step of periodically estimating and determining solves a set of first order differential equations to derive how the process will react to said proposed changes, and determines an optimized set of current and future manipulated parameter values.

8. The method according to claim 7, wherein said set of current and future manipulated parameter values includes at least one setpoint which enables tracking of a predetermined set-point profile.

9. The method according to claim 7, wherein said optimized set of current and future manipulated parameter values is determined at a first interval rate and the current state of each unmeasured process variable is estimated at a second interval rate which is greater than the first interval rate.

10. The method according to claim 7, further comprising the steps of providing a real-time database for physical data received from said input signals and storing said physical data in said database in the form of records.

11. The method according to claim 10, wherein said records are stored as a set of linked lists.

12. The method according to claim 11, wherein said real-time database includes a list of set-point corners for projecting future setpoints, with each of said set-point corners including a floating point value and a step in time where said floating point value applies.