Method and apparatus for process control

• US 5,150,289 A
• Filed: 07/30/1990
• Issued: 09/22/1992
• Est. Priority Date: 07/30/1990
• Status: Expired due to Term

• Current Assignee: Invensys Systems Incorporated (Invensys PLC)
• Original Assignee: The Foxboro Company (Schneider Electric SE)
• Inventors: Badavas, Paul C.
• Primary Examiner(s): Smith, Jerry
• Assistant Examiner(s): GORDON, PAUL P
• Application Number:
  US07/559,645
• Time in Patent Office: 785 Days
• Field of Search: 364/152, 364/153, 364/154, 364/155, 364/157, 364/185, 364/582, 364/148
• US Class Current: 700/34
• CPC Class Codes: G05B 21/02 electric

Abstract

A statistical process control system provides asymmetrical nonlinear automatic closed-loop feedback control. The system has application, for example, in the control of equipment that responds to a controlled variable signal to vary a measurable characteristic of a process. The system uses an accumulated deviation of a measured subgroup means minus a target value, divided by the subgroup standard deviation, to modify the manipulated variable each time the accumulated value exceeds a decision interval above or below the target. The system permits the use of independent slack variables subtracted from the accumulated deviation to more closely model the underlying process. Likewise, the system permits the use of independent alarm values and variable gains to permit greater control of the process.

Claims

1. An apparatus for closed-loop control of a process performed with equipment that responds to a controlled variable signal to vary a first characteristic of that process, said apparatus having the improvement comprising

A. batch monitoring means for monitoring values of a first process characteristic of the process being performed and for generating a signal representative of an arithmetic mean of those monitored values and for generating a signal representative of a standard deviation of those monitored values,

B. normalization means coupled to said batch monitoring means for responding to said mean value-representative signal and to said standard deviation-representative signal to generate an error signal representative of a normalized deviation of said monitored values,

C. drift detection means coupled to said normalization means for responding to said error signal to detect a selected change in said first process characteristic, said drift detection means comprising

i) first accumulation means for generating a high-value accumulation signal representative of a summation of time-successive values of a difference between said error signal and a first slack value, and

ii) second accumulation means for generating a low-value accumulation signal representative of a summation of time-successive values of a difference between an additive inverse of said error signal and a second slack value, which may be independent of said first slack value,

D. process control means coupled to said drift detection means for selectively generating a manipulated variable signal representative of a sum of a prior value of said manipulated variable signal and a scaled one of at least one of said high-value accumulation signal and said low-value accumulation signal, and

E. closed-loop control means for applying said manipulated variable signal to the process-performing equipment to drive the controlled variable signal to a target value.

2. An apparatus according to claim 1, wherein said batch monitoring means comprises

A. measuring means for measuring a value of said process first characteristic,

B. sampling means coupled to said measuring means for forming a subgroup of values by sampling said first characteristic value (n) times and for generating a signal representative thereof, where (n) is an integer greater than ONE, and

C. means for responding to said (n) sampled first characteristic values to generate said mean-representative signal and said standard deviation-representative signal.

3. An apparatus according to claim 2, wherein said sampling means comprises consecutive sample means for forming a subgroup by sampling said (n) first characteristic value-representative signals beginning every (n) values.

4. An apparatus according to claim 2, wherein said sampling means comprises skip-m sample means for forming a subgroup by sampling said (n) first characteristic value-representative signals beginning every (n+m) values, where (m) is an integer greater than or equal to ONE.

5. An apparatus according to claim 2, wherein said sampling means comprises moving sample means for forming a subgroup of previous ones of said sampled first characteristic values and combining those values with at least one newly sampled value.

6. An apparatus according to claim 1, wherein said normalization means comprises

A. means for generating a signal representative of a numerical value of a difference between said mean-representative signal and a signal representative of said target value, and

B. means for generating said error signal as representative of a numerical value of said difference-representative signal divided by said standard deviation-representative signal.

7. An apparatus according to claim 6, wherein said normalization means comprises ZERO divide-prevention means responsive to said standard deviation-representative signal having a value less than a selected minimum value for setting said standard deviation-representative signal equal to that minimum value.

8. An apparatus according to claim 6, including means for inputting said target value-representative signal from an operator.

9. An apparatus according to claim 1, wherein said drift detection means comprises

A. first input means for receiving an operator-selectable first slack value, and

B. second input means for receiving an operator-selectable second slack value.

10. An apparatus according to claim 1, wherein said first accumulation means comprises

A. means for generating said high-value accumulation signal as a sum of a value of a prior said high-value accumulation signal and a difference between said error signal and a first slack value, and

B. means for responding to said high-value accumulation signal having a value less than ZERO for setting that signal to represent a value of ZERO.

11. An apparatus according to claim 1, wherein said second accumulation means comprises

A. means for generating said low-value accumulation signal as a sum of a value of a prior said low-value accumulation signal with a difference between an additive inverse of said error signal and a second slack value, and

B. means for responding to said low-value accumulation signal having a value less than ZERO for setting that signal to represent a value of ZERO.

12. An apparatus according to claim 1, wherein said process control means comprises

A. first alarm condition means for responding to said high-value accumulation signal having a value greater than a high-alarm value for generating a high-alarm signal, and

B. second alarm condition means for responding to said low-value accumulation signal having a value greater than a low-alarm value for generating a low-alarm signal, which may be independent of said high-alarm value.

13. An apparatus according to claim 12, including means for inputting at least one of said high-alarm signal and said low-alarm signal from an operator.

14. An apparatus according to claim 12, wherein said process control means includes means responsive to said high-alarm signal for generating said manipulated variable signal as a sum of said prior value of said manipulated variable signal and a multiplicative product of said high-value accumulation signal and a high alarm controller gain value.

15. An apparatus according to claim 14, including means for inputting said high-alarm controller gain value from an operator.

16. An apparatus according to claim 12, wherein said process control means includes means responsive to said low-alarm signal for generating said manipulated variable as a sum of said prior value of said manipulated variable and a multiplicative product of said low-value accumulation signal and a low alarm controller gain value.

17. An apparatus according to claim 16, including means for inputting said low-alarm controller gain value from an operator.

18. An apparatus according to claim 12, wherein said process control means includes means responsive to at least one of said high-alarm signal and said low-alarm signal for generating said manipulated variable signal as a sum of a prior value of said manipulated variable signal and at least one of

i) a multiplicative product of said high-value accumulation signal and a high alarm controller gain value, and

ii) a multiplicative product of said low-value accumulation signal and a low alarm controller gain value.

19. An apparatus according to claim 18, including means for inputting at least one of said high alarm controller gain value and said low alarm controller gain value from an operator.

20. A method for closed-loop control of a process performed on equipment that responds to a controlled variable signal to vary a first characteristic of that process, said method having the improvement comprising the steps of

A. monitoring values of a first process characteristic of the process being performed for generating a signal representative of an arithmetic mean of those monitored values and for generating a signal representative of a standard deviation of those monitored values,

B. normalizing said standard deviation-representative signal to generate an error signal,

C. responding to said error signal to detect a selected change in said plural time-successive values of said first process characteristic, said error-responding step including the steps of

i) generating a high-value accumulation signal representative of a summation of time-successive values of differences between said error signal and a first slack value, and

ii) generating a low-value accumulation signal representative of a summation of time-successive values of differences between an additive inverse of said error signal and a second slack value which may be independent of said first slack value,

D. selectively generating a manipulated variable signal representative of a sum of a prior value of said manipulated variable signal and a scaled one of at least one of said high-value accumulation signal and said low-value accumulation signal, and

E. applying said manipulated variable signal to the process performing equipment to drive the controlled variable signal toward a target value.

21. A method according to claim 20, wherein said monitoring step comprises

A. measuring a value of said first process characteristic,

B. forming a subgroup of values by sampling said first characteristic value (n) times and for generating a signal representative thereof, where (n) is an integer greater than ONE, and

C. responding to said (n) sampled first characteristic values to generate said mean value-representative signal and said standard deviation-representative signal.

22. A method according to claim 21, further comprising the step of forming said subgroup by sampling said (n) first characteristic value-representative signals beginning every (n) values.

23. A method according to claim 21, further comprising the step of forming said subgroup by sampling said (n) first characteristic value-representative signals beginning every (n+m) values, where (m) is an integer greater than or equal to ONE.

24. A method according to claim 21, further comprising forming said subgroup from previous ones of said sampled first characteristic values and combining those values with at least one newly sampled value.

25. A method according to claim 20, wherein said normalizing step comprises the steps of

A. generating a signal representative of a numerical value of a difference between said mean value-representative signal and a signal representative of said target value of said first characteristic, and

B. generating said error signal as representative of a numerical value of said difference-representative signal divided by said standard deviation-representative signal.

26. A method according to claim 25, wherein said normalizing step comprises the step of responding to said standard deviation-representative signal having a value less then a selected minimum value for setting said standard deviation-representative signal equal to that minimum value.

27. A method according to claim 25, further comprising the step of receiving said target value-representative signal from an operator input device.

28. A method according to claim 20, wherein said error-responding step comprises the step of receiving each of said first and second slack values from an operator input device.

29. A method according to claim 20, wherein said high-value accumulation signal generating step comprises the steps of

A. generating said high-value accumulation signal as a sum of a value of a prior said high-value accumulation signal and a difference between said error signal and a first slack value, and

B. responding to said high-value accumulation signal having a value less than ZERO for setting that signal to have a value of ZERO.

30. A method according to claim 29, further comprising the step of receiving said first slack value from an operator input device.

31. A method according to claim 20, wherein said low-value accumulation signal generating step comprises the steps of

A. generating said low-value accumulation signal as a sum of a value of a prior said low-value accumulation signal and a difference between an additive inverse of said error signal and a second slack value, and

B. responding to said low-value accumulation signal having a value less than ZERO for setting that signal to have a value of ZERO.

32. A method according to claim 31, further comprising the step of receiving said second slack value from an operator input device.

33. A method according to claim 20, wherein said process control signal generating step comprises the steps of

A. responding to said high-accumulation signal having a value greater than a high-alarm value for generating a high-alarm signal, and

B. responding to said low-accumulation signal having a value greater than a low-alarm value for generating a low-alarm signal, said low-alarm value which may be independent of said high-alarm value.

34. A method according to claim 33, wherein said process control signal generating step comprises the step of generating said process control signal as a sum of said prior value of said process control signal and a multiplicative product of said high-accumulation signal and a high-alarm controller gain.

35. A method according to claim 34 comprising the further step of receiving said high alarm controller gain from an operator input device.

36. A method according to claim 33, wherein said process control signal generating step comprises the step of generating said process control signal as a sum of said prior value of said process control signal and a multiplicative product of said low-accumulation signal and a low-alarm controller gain.

37. A method according to claim 36 comprising the further step of receiving said low alarm controller gain from an operator input device.

38. A method according to claim 33, wherein said process control signal generating step comprises at least one of the steps of

A. generating said process control signal as a sum of said prior value of said process control signal and a multiplicative product of said high-accumulation signal and a high-alarm controller gain, and

B. generating said process control signal as a sum of said prior value of said process control signal and a multiplicative product of said low-accumulation signal and a low-alarm controller gain.

39. A method according to claim 38, comprising the step receiving at least one of said high-alarm controller gain and said low-alarm controller gain from an operator input device.