Pattern recognition adaptive controller

• US 5,355,305 A
• Filed: 10/29/1992
• Issued: 10/11/1994
• Est. Priority Date: 10/29/1992
• Status: Expired due to Term

• Current Assignee: Johnson Controls Technology Company (Johnson Controls International plc)
• Original Assignee: Johnson Service Company
• Inventors: Seem, John E., Haugstad, Howard J.
• Primary Examiner(s): GORDON, PAUL P
• Application Number:
  US07/968,583
• Time in Patent Office: 712 Days
• Field of Search: 364/152, 364/153, 364/157, 364/161, 364/178, 364/572, 364/574
• US Class Current: 700/41
• CPC Class Codes: G05B 11/42 for obtaining a characteristic which is both proportional and time-dependent, e.g. P. I., P. I. D.
• CPC Class Codes: G05B 13/024 in which a parameter or coefficient is automatically adjusted to optimise the performance

Abstract

A pattern recognition adaptive controller configured to dynamically adjust proportional gain and integral time control parameters to minimize integrated absolute errors between a setpoint and a monitored controlled variable. The pattern recognition adaptive controller receives a sampled signal representative of the controlled variable, and determines a smoothed signal based on the sampled signal. The controller characterizes a disturbance in the smoothed signal by a damping factor and a closed loop response time. When a significant load disturbance or setpoint change occurs, the controller estimates an optimal gain based on the damping factor, and an optimal integral time based on the response time. The estimated optimal gain and estimated optimal integral time are used to determining a new gain and a new integral time values, to which the control parameters of the controller are then set.

Claims

1. A method of dynamically adjusting the control parameters of a proportional gain and integral time controller disposed to control an actuator affecting a process, comprising the steps of:

sampling a feedback signal representative of a controlled variable of the process to generate a sampled signal;

generating a smoothed signal based on the sampled signal;

determining an estimated noise level of the sampled signal;

determining a tune noise band based on the estimated noise level and the smoothed signal;

adjusting the gain and integral time values used by the controller if either the difference between a previous setpoint value and a current setpoint value, or the difference between the current setpoint value and the smoothed signal, falls outside the tune noise band; and

utilizing the adjusted control parameters to control the actuator, thereby causing the controller to affect the process.

2. The method of claim 1 wherein the step of generating a smooth signal includes determining a smoothed sample value based upon fitting a quadratic function through a first plurality of evenly spaced points, and estimating the slope of the sampled signal based upon fitting a quadratic function through a second plurality of evenly spaced points.

3. The method of claim 1 wherein the step of determining an estimated noise level includes determining a current noise level based on the difference between the sampled signal and the smoothed signal, and determining a long-term average noise level based on an exponentially weighted moving average of the current noise level and previously determined noise levels.

4. The method of claim 1 wherein the step of determining a tune noise band includes determining a long term average noise level, and wherein the tune noise band is determined based upon the long-term average noise level and the current setpoint value.

5. An apparatus for dynamically adjusting the control parameters of a proportional gain and integral time controller disposed to control an actuator affecting a process, comprising:

means for sampling a feedback signal representative of a controlled variable of the process to generate a sampled signal;

means for generating a smoothed signal based on the sampled signal;

means for determining an estimated noise level of the sampled signal;

means for determining a tune noise band based on the estimated noise level and the smoothed signal;

means for adjusting the gain and integral time values used by the controller if either the difference between a previous setpoint value and a current setpoint value, or the difference between the current setpoint value and the smoothed signal, falls outside the tune noise band; and

means for utilizing the adjusted control parameters to control the actuator, thereby causing the controller to affect the process.

6. The apparatus of claim 5 wherein the means for generating a smooth signal includes means for determining a smoothed sample value based upon fitting a quadratic function through a first plurality of evenly spaced points, and means for estimating the slope of the sampled signal based upon fitting a quadratic function through a second plurality of evenly spaced points.

7. The apparatus of claim 5 wherein the means for determining an estimated noise level includes means for determining a current noise level based on the difference between the sampled signal and the smoothed signal, and means for determining a long-term average noise level based on an exponentially weighted moving average of the current noise level and previously determined noise levels.

8. The apparatus of claim 5 wherein the means for determining a tune noise band includes means for determining a long-term average noise level and the current setpoint value.

9. A method of dynamically adjusting the control parameters of a proportional gain and integral time controller disposed to control an actuator affecting a process, comprising the steps of:

sampling a feedback signal representative of a controlled variable of the process to generate a sampled signal;

generating a smoothed signal based on the sampled signal;

determining an estimated noise level of the sampled signal;

determining a tune noise band based on the estimated noise level and the smoothed signal;

determining whether a significant setpoint change has occurred by comparing the tune noise band to the difference between a previous setpoint value and a current setpoint value;

determining whether a significant load disturbance has occurred by comparing the tune noise band to the difference between the current setpoint value and the smoothed signal;

determining a damping factor based on the slope of the smoothed signal;

determining a closed loop response time based on the height of the smoothed signal;

determining an average disturbance size;

estimating an optimal gain based on the damping factor;

estimating an optimal integral time based on the response time;

determining a new gain and a new integral time based on the estimated optimal gain, the estimated optimal integral time, the current gain and integral time values used in the controller, a signal-to-noise ratio of the sampled signal, and the size of a current load disturbance or setpoint change relative to the average disturbance size;

setting the gain and integral time of the controller to the new gain and the new integral time; and

utilizing the adjusted control parameters to control the actuator, thereby causing the controller to affect the process.

10. The method of claim 9 further comprising the step of determining whether the actuator is saturated.

11. The method of claim 9 wherein the step of generating a smooth signal includes determining a smoothed sample value based upon fitting a quadratic function through a first plurality of evenly spaced points, and estimating the slope of the sampled signal based upon fitting a quadratic function through a second plurality of evenly spaced points.

12. The method of claim 9 wherein the step of determining an estimated noise level includes determining a current noise level based on the difference between the sampled signal and the smoothed signal, and determining a long-term average noise level based on an exponentially weighted moving average of the current noise level and previously determined noise levels.

13. The method of claim 12 wherein the tune noise band is determined based upon the long-term average noise level and the current setpoint value.

14. The method of claim 9 wherein the damping factor is determined, when the smoothed signal is underdamped, according to the formula:

##EQU14## where, for setpoint changes, S₁ is the maximum of the absolute value of the slope of the smoothed signal between the time the smoothed signal falls outside a tune noise range and the time of a first extremum, and S₂ is the maximum of the absolute value of the slope of the smoothed signal between the time of the first extremum and a second extremum, and for load disturbances, S₁ is the maximum of the absolute value of the slope of the smoothed signal between the time of the first extremum and the second extremum, and S₂ is the maximum of the absolute value of the slope of the smoothed signal between the time of the second extremum and a third extremum.

15. The method of claim 9 wherein the closed loop response time is determined, when the smoothed signal is underdamped, according to the formula:

##EQU15## where θ

is the closed loop response time, T is the sampling interval, h is the height of the response, and for setpoint changes, S₁ is the maximum of the absolute value of the slope of the smoothed signal between the time the smoothed signal falls outside a tune noise range and the time of a first extremum, and for load disturbances, S₁ is the maximum of the absolute value of the slope of the smoothed signal between the time of the first extremum and a second extremum.

16. The method of claim 9 wherein the average disturbance size is determined by calculating an exponentially weighted moving average of the difference between the previous setpoint and a first extremum.

17. The method of claim 9 wherein the step of estimating an optimal gain based on the damping factor includes determining the ratio of the estimated optimal gain to the present gain is according to the formula:

##EQU16## where K_opt is the estimated optimal gain, K is the present value of the gain used in the controller D_slope, and the A_n coefficients are determined by linear regression.

18. The method of claim 9 wherein the step of estimating an optimal integral time based on the response time includes determining the ratio of the estimated optimal integral time to the actual sampling interval according to the formula:

##EQU17## where T_i,opt is the estimated optimal integral time, T is the sampling interval, θ

is the closed loop response time, and the B_n coefficients are determined by linear regression.

19. A apparatus for dynamically adjusting the control parameters of a proportional gain and integral time controller disposed to control an actuator affecting a process, comprising:

means for sensing the variable and generating a feedback signal representative of the variable;

means for determining a damping factor based on the slope of the feedback signal;

means for determining a closed loop response time based on the height of the feedback signal;

means for estimating an optimal gain based on the damping factor;

means for estimating an optimal integral time based on the response time;

means for determining a new gain and a new integral time based on the estimated optimal gain, the estimated optimal integral time, and the current gain and integral time values used in the controller;

means for setting the gain and integral time of the controller to the new gain and the new integral time; and

utilizing the adjusted control parameters to control the actuator, thereby causing the controller to affect the process.

20. The apparatus of claim 19 further comprising:

means for sampling the feedback signal; and

means for smoothing the feedback signal based on the sampled values.

21. The apparatus of claim 19 further comprising:

means for determining an estimated noise level of the feedback signal;

means for determining a tune noise band based on the estimated noise level and the feedback signal;

means for determining whether a significant setpoint change has occurred by comparing the tune noise band to the difference between a previous setpoint value and a current setpoint value;

means for determining whether a significant load disturbance has occurred by comparing the tune noise band to the difference between the current setpoint value and the feedback signal; and

wherein the gain and integral time of the controller are set to the new gain and the new integral time when either a significant setpoint change or a significant load disturbance occurs.

22. The apparatus of claim 21 further comprising means for determining an average disturbance size, wherein the new gain and new integral time are further based upon a signal-to-noise ratio of the feedback signal and the size of a current load disturbance or setpoint change relative to the average disturbance size.