Internal Model Controller with automatically correcting gain of the internal model controller

• US 5,537,310 A
• Filed: 12/20/1994
• Issued: 07/16/1996
• Est. Priority Date: 12/27/1993
• Status: Expired due to Term

• Current Assignee: Yamatake Corporation
• Original Assignee: Yamatake-Honeywell Company Limited
• Inventors: Tanake, Masato, Mibuchi, Hiroyuki
• Primary Examiner(s): RUGGIERO, JOSEPH F
• Application Number:
  US08/359,630
• Time in Patent Office: 574 Days
• Field of Search: 364/148, 364/149-151, 364/153, 364/157, 364/158, 364/159, 364/164, 364/165, 364/166, 364/176, 364/177, 364/183, 364/152, 318/561
• US Class Current: 700/31
• CPC Class Codes: G05B 13/042 in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

An internal model control (IMC) automatically corrects the gain of the internal model control when the settings in the internal model are improper. The controller calculates and outputs a manipulated variable according to an IMC control algorithm, and an adaptive tuning section adaptively tunes the model gain of an internal model such that change rates of the controlled variable and the reference controlled variable used in said IMC section become the same value during a control response period in a transient state. When a transient state of control is detected, a transient state detection signal is output and a response start region is detected. The change rates of the controlled variable and the reference controlled variable in this response start region are determined. The corrected gain of the internal model is calculated from the change rates of the controlled variable and the reference controlled variable in the response start region. The corrected gain is then output to an internal model storage section, thereby correcting the gain of the internal model.

Claims

1. A controller having an IMC structure which calculates a manipulated variable output to a controlled system process from a command of control, calculates a reference controlled variable corresponding to a controlled variable of the controlled system process as a control result by using an internal model representing the controlled system process, and feeds back a difference between the controlled variable and the reference controlled variable, thereby adaptively stabilizing control, comprising:

an IMC section for calculating and outputting a manipulated variable according to an IMC control algorithm; and

an adaptive tuning section for adaptively tuning a model gain of an internal model as a constituent element of the IMC control algorithm such that change rates of the controlled variable and the reference controlled variable as variables used in said IMC section become the same value during a control response period in a transient state, said IMC section comprising;

a command filter section for outputting an input command of control with a transfer function having a delay characteristic,

a first subtraction processing section for subtracting a feedback amount from the output from said command filter section,

a manipulated variable calculating section including a command/disturbance filter section for outputting the output from said first subtraction processing section with a transfer function having a delay characteristic, and a manipulating section for calculating a manipulated variable on the basis of the output from said command/disturbance filter section on the basis of parameters of said internal model, and outputting the calculated manipulated variable,

an internal model storage section for storing parameters of said internal model,

an internal model output calculating section for calculating a reference controlled variable from the manipulated variable on the basis of parameters of said internal model, and

a second subtraction processing section for subtracting the reference controlled variable output from said internal model output calculating section from the controlled variable of said controlled system process, and outputting the feedback amount.

2. A controller having an IMC structure which calculates a manipulated variable output to a controlled system process from a command of control, calculates a reference controlled variable corresponding to a controlled variable of the controlled system process as a control result by using an internal model representing the controlled system process, and feeds back a difference between the controlled variable and the reference controlled variable, thereby adaptively stabilizing control, comprising:

an IMC section for calculating and outputting a manipulated variable according to an IMC control algorithm; and

an adaptive tuning section for adaptively tuning a model gain of an internal model as a constituent element of the IMC control algorithm such that change rates of the controlled variable and the reference controlled variable as variables used in said IMC section become the same value during a control response period in a transient state, said IMC section comprising;

a command filter section for outputting an input command of control with a transfer function having a delay characteristic,

a first subtraction processing section for subtracting a feedback amount from the output from said command filter section,

a manipulated variable calculating section including a command/disturbance filter section for outputting the output from said first subtraction processing section with a transfer function having a delay characteristic, and a manipulating section for calculating a manipulated variable on the basis of the output from said command/disturbance filter section on the basis of parameters of said internal model, and outputting the calculated manipulated variable,

an internal model storage section for storing parameters of said internal model,

an internal model output calculating section for calculating a reference controlled variable from the manipulated variable on the basis of parameters of said internal model, and

a second subtraction processing section for subtracting the reference controlled variable output from said internal model output calculating section from the controlled variable of said controlled system process, and outputting the feedback amount;

said adaptive tuning section comprising;

a deviation change amount calculating section for calculating a deviation as a difference between the command and the controlled variable, and calculating a change amount of the deviation,

a sequential change rate calculating section for calculating change rates of the controlled variable and the reference controlled variable,

a first transient state detecting section for outputting a transient state detection signal upon detecting a transient state of control due to a change in command,

a second transient state detecting section for outputting a transient state detection signal upon detecting a transient state of control due to a factor other than the change in command,

a response start region detecting section for detecting a response start region, in which control responses corresponding to a transient state begin to appear in the controlled variable and the reference controlled variable, on the basis of the deviation and the change rates of the controlled variable and the reference controlled variable, when the transient state detection signal is output, and determining the change rates of the controlled variable and the reference controlled variable in the response start region, and

a model gain calculating section for calculating a corrected gain of said internal model from the change rates of the controlled variable and the reference controlled variable in the response start region, and updating a gain as one of the parameters stored in said internal model storage section into the corrected gain.

3. A controller having an IMC structure which calculates a manipulated variable output to a controlled system process from a command of control, calculates a reference controlled variable corresponding to a controlled variable of the controlled system process as a control result by using an internal model representing the controlled system process, and feeds back a difference between the controlled variable and the reference controlled variable, thereby adaptively stabilizing control, comprising:

an IMC section for calculating and outputting a manipulated variable according to an IMC control algorithm; and

an adaptive tuning section for adaptively tuning a model gain of an internal model as a constituent element of the IMC control algorithm such that change rates of the controlled variable and the reference controlled variable as variables used in said IMC section become the same value during a control response period in a transient state, said adaptive tuning section comprising;

a deviation change amount calculating section for calculating a deviation as a difference between the command and the controlled variable, and calculating a change amount of the deviation,

a sequential change rate calculating section for calculating change rates of the controlled variable and the reference controlled variable,

a first transient state detecting section for outputting a transient state detection signal upon detecting a transient state of control due to a change in command,

a second transient state detecting section for outputting a transient state detection signal upon detecting a transient state of control due to a factor other than the change in command,

a response start region detecting section for detecting a response start region, in which control responses corresponding to a transient state begin to appear in the controlled variable and the reference controlled variable, on the basis of the deviation and the change rates of the controlled variable and the reference controlled variable, when the transient state detection signal is output, and determining the change rates of the controlled variable and the reference controlled variable in the response start region, and

a model gain calculating section for calculating a corrected gain of said internal model from the change rates of the controlled variable and the reference controlled variable in the response start region, and updating a gain as one of the parameters stored in said internal model storage section into the corrected gain.

4. A controller according to claim 3, wherein said adaptive tuning section further comprises

a peak controlled variable detecting section for detecting maximum and minimum values of the controlled variable on the basis of the change rate of the controlled variable after the gain of said internal model is corrected by said model gain calculating section, and setting a controlled variable at the time point, at which the maximum and minimum values are detected, as a peak controlled variable,

a damping determining section for evaluating a fluctuated state of the controlled variable on the basis of the peak controlled variable, and outputting a re-execution designation signal for re-executing correction of the gain of said internal model to said response start region detecting section upon determining that damping of fluctuations is insufficient, and

a settlement determining section for evaluating a settled state of control on the basis of the deviation and the deviation change amount output from said deviation change amount calculating section, and outputting initialization signals for resuming detection of a transient state to prepare for a next transient state to said first and second transient state detecting sections upon determining a settled state.

5. A controller according to claim 3, wherein said adaptive tuning section further comprises

an excessive stability determining section for evaluating an excessively stable state of control on the basis of the deviation and the deviation change amount output from said deviation change amount calculating section, and outputting a correction designation signal upon determining the excessively stable state, and

an excessive stability gain correcting section for calculating a corrected gain of said internal model to cancel the excessively stable state in response to the correction designation signal, and updating a gain as one of the parameters stored in said internal model storage section into the corrected gain.

6. A controller according to claim 3, wherein said adaptive tuning section further comprises

a settlement determining section for evaluating a settled state of control on the basis of the deviation and the deviation change amount output from said deviation change amount calculating section, and outputting a correction designation signal upon determining the settled state, and

a quick response characteristic model gain correcting section for calculating a corrected gain of said internal model to restore a quick response characteristic of control in response to the correction designation signal, and updating a gain as one of the parameters stored in said internal model storage section into the corrected gain.

7. A controller according to claim 3, wherein said adaptive tuning section further comprises

a settlement determining section for evaluating a settled state of control on the basis of the deviation and the deviation change amount output from said deviation change amount calculating section, and outputting a correction designation signal upon determining the settled state,

an idle time error estimating section for estimating an idle time error of said internal model on the basis of time information of the response start region which is output from said response start region detecting section, and

a model idle time correcting section for calculating a corrected idle time of said internal model on the basis of the idle time error in response to the correction designation signal, and updating an idle time as one of the parameters stored in said internal model storage section into the corrected idle time.

8. In a system process controller which calculates a manipulated variable output to the system process in response to an input command and calculates a reference control variable corresponding to a control variable of the system process using an internal model control algorithm representing the system process, and feeds back a difference between the control variable and the reference control variable, a method for adaptively modifying a parameter of an internal model control algorithm comprising the steps of:

calculating a sequential change rate of the input command;

calculating a deviation change rate between said sequential change rate of the input command and a change rate of the reference control variable; and

calculating an adjusted gain parameter for said internal model control algorithm as a function of said deviation change rate.

9. A method for adaptively modifying a parameter of an internal model control algorithm as in claim 8, comprising the further steps of:

detecting a transient state of said input command; and

adjusting the gain of the internal model control algorithm such that the change rate of the input command and the change rate of the reference control variable converge during a response period in said transient state.

10. A method for adaptively modifying a parameter of an internal model control algorithm as in claim 9, comprising the further steps of:

detecting maximum and minimum values of the controlled variable on the basis of the change rate of the controlled variable after the gain of said internal model is adjusted;

setting, as a peak controlled variable, the control variable at the point in time at which the maximum and minimum values are detected;

evaluating a fluctuated state of the controlled variable on the basis of the peak controlled variable;

re-executing adjustment of the gain of said internal model upon determining that damping of the fluctuated state is insufficient,

evaluating a settled state of control on the basis of the deviation and the deviation change amount output, and

resuming detection of a transient state to prepare for a next transient state to said first and second transient state detecting sections upon determining a settled state.

11. A method for adaptively modifying a parameter of an internal model control algorithm as in claim 9, comprising the further steps of:

evaluating an excessively stable state of control on the basis of the deviation and the deviation change amount;

outputting an adjustment designation signal upon determining the excessively stable state,

calculating an adjusted gain of said internal model to cancel the excessively stable state in response to the adjustment designation signal, and

updating a gain as one of the parameters stored in said internal model storage section into the adjusted gain.

12. A method for adaptively modifying a parameter of an internal model control algorithm as in claim 9, comprising the further steps of:

evaluating a settled state of control on the basis of the deviation and the deviation change amount output;

outputting an adjustment designation signal upon determining the settled state,

calculating an adjusted gain of said internal model to restore a quick response characteristic of control in response to the adjustment designation signal, and

updating a gain as one of the parameters stored in said internal model storage section into the adjusted gain.

13. A method for adaptively modifying a parameter of an internal model control algorithm as in claim 9, comprising the further steps of:

evaluating a settled state of control on the basis of the deviation and the deviation change amount output from said deviation change amount;

outputting an adjustment designation signal upon determining the settled state,

estimating an idle time error of said internal model;

calculating an adjusted idle time of said internal model on the basis of the idle time error in response to the adjustment designation signal, and

updating an idle time as one of the parameters stored in said internal model storage section into the adjusted idle time.