Process stabilizing process controller

• US 5,486,998 A
• Filed: 06/14/1993
• Issued: 01/23/1996
• Est. Priority Date: 06/14/1993
• Status: Expired due to Term

• Current Assignee: Alpha Coal West, Inc.
• Original Assignee: Amax Coal West, Inc. (RAG International Mining GmbH)
• Inventors: Corso, Ronald
• Primary Examiner(s): Gordon, Paul P.
• Application Number:
  US08/077,244
• Time in Patent Office: 953 Days
• Field of Search: 364/148, 364/152, 364/157, 364/138, 364/578, 364/503, 364/572, 395/11, 395/20, 395/23, 395/21, 395/906
• US Class Current: 700/32
• CPC Class Codes: G05B 13/028 using expert systems only
• CPC Class Codes: G05B 23/027 Alarm generation, e.g. communication protocol; Forms of alarm

Abstract

A process controller is disclosed for controlling a process. The process controller includes a process supervisor for accessing process control data clusters from a library. Each data cluster provides control data, e.g., setpoints, deadbands, execution frequencies and alarm ranges for process actuator controllers. Each actuator controller maintains a related process attribute within a deadband by controlling at least one process actuator. Data clusters for the library are provided in substantially real-time by a learning unit. Data continuously output from process related sensors are transformed and supplied to the learning unit as training data for learning at least one relationship between process control data and data for at least one substantially uncontrollable process influencing factor (e.g., processing site environmental factors such as temperature and barometric pressure). Prior to inserting a newly provided data cluster into the library, the data cluster is evaluated for predicted successful performance when accessed by the supervisor. Data clusters are also evaluated after each use by the supervisor, with non-performing clusters becoming inaccessible for further use. The supervisor can also fine tune or adjust control data to more closely fit the present behavior of the process. The process controller is embodied in controlling a coal drying process using a combustion gas heat transfer technique.

Claims

1. An automated control processing method comprising the steps of:

conducting a process which is controlled as said process is performed;

establishing, within a learning means, at least one learned relationship between data for at least a first process control parameter for controlling said process, and data for at least a first process influencing factor, said learning means receiving process training data output in said step of conducting;

generating at least a first data cluster, said first data cluster generated from an output of said learning means, said first data cluster including first control data related to said first process control parameter and including first applicability data related to said first process influencing factor for determining when said control data is to be used in controlling said process;

entering said first data cluster into a plurality of stored data clusters, each said data cluster having control data related to said first process control parameter and applicability data related to said first process influencing factor for determining when said control data is to be used in controlling said process;

controlling said process using said first data cluster as an accessed data cluster obtained from said plurality of stored data clusters;

post-filtering said plurality of stored data clusters to retain said first data cluster when used as said accessed data cluster and said process has acceptable performance.

2. A method as claimed in claim 1, wherein said process includes a drying of coal.

3. A method as claimed in claim 2, wherein the control data of said first data cluster includes one of a setpoint, a deadband, an execution frequency and alarm data related to an attribute controller for controlling one of:

free oxygen within a heat transport gas, a drying chamber temperature, a deck pressure differential and an exhaust fan temperature.

4. A method as claimed in claim 2, wherein said step of conducting includes deluging said process with water when the process enters an emergency state.

5. A method as claimed in claim 1, wherein said step of conducting includes triggering an automated telephone call to at least one predetermined person upon detecting a critical alarm situation while said process is in an unattended state.

6. A method as claimed in claim 1, wherein said step of conducting includes at least one of filtering and transforming data output from said process into process training data.

7. A method as claimed in claim 1, wherein said first data cluster includes values for one or more of the following:

a deadband, a setpoint, an execution frequency and an alarm range.

8. A method as claimed in claim 1, wherein said first data cluster includes one or more of the following:

a data acquisition switch, a lag time, a delay time, trending time attributes, an in-service indicator and use delay attributes.

9. A method as claimed in claim 1, wherein said first process influencing factor has an uncontrolled variation that influences the process sufficiently so that the process is controlled to accommodate the uncontrolled variation.

10. A method as claimed in claim 9, wherein, whenever the process includes drying of coal, said first process influencing factor is one of:

ambient temperature, windspeed, wind direction, humidity, barometric pressure, moisture content of coal to be dried, temperature of coal to be dried and particle size of coal to be dried.

11. A method as claimed in claim 9, wherein said step of controlling includes obtaining process measurement data for at least one process sensor related to said first process influencing factor.

12. A method as claimed in claim 1, wherein said step of controlling includes periodically calculating an acceptability index for determining at least one of whether a process measurement is outside a predetermined deadband and whether a measurement of a characteristic for a product produced by the process is outside a predetermined range.

13. A method as claimed in claim 1, wherein the said first data cluster is related to predetermined performance target data for comparing with product characteristic data obtained from at least one measurement of a product produced by the process, thereby to determine at least one of the quantity and quality of said product.

14. A method as claimed in claim 13, wherein whenever the process includes a drying of coal, said predetermined performance target data includes one or more values of:

moisture content of coal dried, amount of coal dried, percentage of free oxygen in a heat transport gas, a drying chamber temperature and an exhaust fan temperature.

15. A method as claimed in claim 1, wherein said step of entering includes associating said first data cluster with a process commencing data cluster contained in said plurality of stored data clusters, said process commencing data cluster used for commencing processing to obtain a product with predetermined product characteristics from said process, said process commencing data cluster having predetermined first target data related to the predetermined product characteristics, said first data cluster having predetermined second target data related to the predetermined product characteristics, said first target data and said second target data having only differences that are inconsequential to said step of controlling.

16. A method as claimed in claim 15, wherein whenever said accessed data cluster is said first data cluster, then a data cluster used as said accessed data cluster immediately previously to said first data cluster is one of:

said process commencing data cluster and a second data cluster associated to said process commencing data cluster by said step of associating.

17. A method as claimed in claim 15, wherein said first data cluster includes a submode and said process commencing data cluster includes a mode.

18. A method as claimed in claim 1, wherein said step of controlling includes adjusting a first control value of said first control data of said first data cluster such that said first control value is closer to a related process measurement.

19. A method as claimed in claim 18, wherein said first control value is one of a setpoint and a deadband value.

20. A method as claimed in claim 18, wherein, whenever the process includes a drying of coal, said process measurement includes one of free oxygen gas content of a heat transport gas, a drying chamber temperature, a deck pressure differential and an exhaust fan temperature.

21. A method as claimed in claim 1, wherein said first process influencing factor is controlled sufficiently so that uncontrolled variations in said first process influencing factor do not influence the process sufficiently to warrant controlling the process to accommodate the uncontrolled variation.

22. A method as claimed in claim 21, wherein whenever the process includes drying of coal, said first process influencing factor is one of:

drying chamber temperature, and free oxygen content of a heat transport gas.

23. A method as claimed in claim 1, wherein said step of controlling includes changing an execution frequency of an attribute controller to increase process stability.

24. A method, as claimed in claim 1, wherein said step of controlling includes selecting said first data cluster as said accessed data cluster by comparing said first applicability data with data related to measurement(s) of said first process influencing factor.

25. A method as claimed in claim 1, wherein said step of establishing includes supplying said learning means with training data including one or more of:

process sensor related data, product characteristic related data for a product output from the process, and process control related data.

26. A method as claimed in claim 1, wherein said step of post-filtering includes determining a process performance measurement of said first data cluster, said process performance measurement being a function of at least one of:

a number of times a process attribute is detected as deviating from a related setpoint, a size of each such deviation, and a deadband for the process attribute.

27. A method as claimed in claim 1, wherein said step of entering includes the step of pre-filtering said first data cluster prior to using said first data cluster as said accessed data cluster.

28. A method as claimed in claim 27, wherein said step of pre-filtering includes determining a confidence prediction value for said first data cluster, said confidence prediction value indicative of an expected performance of said first data cluster.

29. A method as claimed in claim 28, wherein said step of determining includes deriving said confidence prediction value by determining a measurement of a difference between said first control data and predetermined composite control data derived from said process training data.

30. A method as claimed in claim 27, wherein said step of pre-filtering includes determining if said first data cluster is sufficiently distinct from every data cluster included in said plurality of stored data clusters.

31. A coal drying control processing method comprising the steps of:

controlling a coal drying process in real time using control data from a first data cluster derived from an output of a learning unit;

obtaining a second data cluster from a plurality of stored data clusters for controlling the coal drying process, said second data cluster also having control data to be used in said controlling step, each data cluster of said plurality of data clusters being derived from an output of said learning unit;

said step of obtaining including comparing a first value obtained from control data in said first data cluster with a second value obtained from control data in said second data cluster;

supplying said second data cluster to said controlling step.

32. A method as claimed in claim 31, wherein said step of controlling includes adjusting a setpoint closer to a measured value of a related process attribute to increase process stability.

33. A method as claimed in claim 31, wherein said step of controlling includes adjusting an execution frequency of a process attribute controller to increase process stability.

34. A method as claimed in claim 33, wherein said step of adjusting includes comparing a time interval between activations of said process attribute controller with a predetermined time interval threshold.

35. A method as claimed in claim 31, wherein said step of controlling includes selecting an attribute controller for adjusting one of a related:

setpoint, deadband and execution frequency.

36. A method as claimed in claim 35, wherein said step of selecting includes determining a number of errors an attribute controller yields, each error being a function of process measurement data related to the attribute controller, a setpoint and deadband related to the attribute controller.

37. A method as claimed in claim 31, wherein said step of controlling includes remedying, without requiring process operator intervention, an alarm activating condition of the coal drying process for each alarm activated wherein said alarm requires timely action affecting said alarm activating condition.

38. A method as claimed in claim 31, wherein said step of controlling includes collecting coal drying process related data for training of said learning unit.

39. A method as claimed in claim 31, wherein said step of obtaining includes selecting said second data cluster such that a value of performance target data related to said second data cluster is within a predetermined range of a corresponding value of performance target data related to said first data cluster by a predetermined function.

40. A method, as claimed in claim 31, wherein:

said step of controlling includes monitoring an input by a process operator for determining whether the input effectively addresses an alarm condition.

41. A coal drying control processing method comprising the steps of:

controlling a coal drying process in real time;

supplying a data cluster to said controlling step, said data cluster derived from an output of a learning unit;

wherein said step of supplying includes a first filtering of said data cluster prior to becoming accessible for use in said step of controlling and a second filtering of said data cluster after termination use in said controlling step, wherein said first filtering includes applying heuristic rules to said data cluster.

42. A coal drying control processing method comprising the steps of;

controlling a coal drying process in real time;

supplying a data cluster to said controlling step, said data cluster derived from an output of a learning unit;

wherein said step of supplying includes a first filtering of said data cluster prior to becoming accessible for use in said step of controlling and a second filtering of said data cluster after termination of use in said controlling step, wherein said second filtering includes preparing an air stability measurement, related to said data cluster, with a predetermined threshold, the comparison used in determining whether said data cluster is to become inaccessible for further use in said step of controlling.

43. A method as claimed in claim 42, wherein the error stability measurement is a function of a deviation of at least one process attribute from a related setpoint.

44. A coal drying control processing method comprising the steps of:

controlling a coal drying process in real time;

supplying a data cluster to said controlling step, said data cluster derived from an output of a learning unit;

wherein said step of supplying includes a first filtering of said data cluster prior to becoming accessible for use in said step of controlling and a second filtering of said data cluster after termination of use in said controlling step.

45. A coal drying control processing method comprising the steps of:

controlling a coal drying process in real time;

supplying a data cluster to said controlling step, said data cluster derived from an output of a learning unit;

wherein said step of supplying includes a first filtering of said data cluster prior to becoming accessible for use in said stem of controlling and a second filtering of said data cluster after termination of use in said controlling step, wherein said first filtering includes determining confidence prediction data for said data cluster, said confidence prediction data a composite of data obtained from said cold drying process.

46. A coal drying control processing method comprising the steps of:

controlling a coal drying process in real time;

supplying a data cluster to said controlling step, the data cluster derived from an output of a learning unit, the data cluster obtained from a plurality of data clusters by selecting the data cluster such that performance target data related to the data cluster is a function of a previous performance target data being used in said step of controlling;

wherein said performance target data includes at least one of;

moisture content of coal to be dried, a dried coal tonnage rate, the percentage of free oxygen in a heat transport gas, a drying chamber temperature and an exhaust fan temperature.

47. An automated control processing method for controlling a process in real time comprising:

supplying control data to a plurality of attribute controllers, each attribute controller used in controlling a related process attribute, each related process attribute having a current operating value;

stabilizing the process being controlled so that a measurement of process stability is within a predetermined acceptable range;

adjusting said control data supplied to a first attribute controller of said plurality of attribute controllers for controlling a first process attribute, said control data including a deadband and a setpoint;

wherein said step of adjusting includes;

(A1) obtaining a direction for shifting said setpoint toward a current operating value for said first process attribute;

(A2) performing at least one of;

shifting said setpoint in said direction and shifting said deadband in said direction.

48. A method, as claimed in claim 47, wherein:

said first process attribute controller controls one of;

free oxygen within a heat transport gas, a drying chamber temperature, a deck pressure differential and an exhaust fan temperature.

49. A method as claimed in claim 47, wherein said measurement of process stability is a function of at least one of a measurement of a process attribute and a measurement of a characteristic of a product produced by the process.

50. A method as claimed in claim 47, wherein said step of adjusting includes selecting said first attribute controller to have its related control data adjusted, wherein an ordering of one or more attribute controller errors is utilized.

51. A method as claimed in claim 50, wherein each error of said one or more attribute controller errors is a function of the deviation of a setpoint from a measurement of a process attribute related to the setpoint.

52. A control processing method comprising the steps of:

providing access to a stored plurality of data clusters, each data cluster for providing process control data to a process manager for controlling a process, each data cluster related to;

(A1) performance target data for specifying target values for one or more processing characteristics of the process;

(A2) applicability data for specifying when the data cluster is applicable for use by the process manager, said applicability data including data related to a measurement of one or more process influencing factors having an uncontrolled variation that influences the process sufficiently so that the process is controlled to accommodate the uncontrolled variation;

entering a first data cluster, related to first performance target data, into said stored plurality of data clusters, said step of entering including associating said first data cluster with a process commencing data cluster contained in said stored plurality of data clusters, said process commencing data cluster used for commencing the process, and said process commencing data cluster being related to a second performance target data, wherein said first and second performance target data have only differences that are inconsequential to a controlling of the process;

accessing said first data cluster from said stored plurality of data clusters for controlling the process, said step of accessing including determining whether a data cluster for obtaining process control data immediately previously to said first data cluster is said process commencing data cluster.

53. An apparatus for controlling a coal drying process comprising:

supervisor means for determining process control data used in controlling the coal drying process;

learning means for;

(a) learning at least a first process relationship using;

i) first data, output by said supervisor means, for one or more process control parameters, and

ii) second data related to at least a first process influencing factor, said first process influencing factor has an uncontrolled variation that influences the coal drying process sufficiently so that the coal drying process is controlled to accommodate the uncontrolled variation; and

(b) outputting learned data relating to said first process relationship;

library means for providing access to a plurality of data clusters having process control data, a first data cluster in said plurality of data clusters is;

(a) generated using said learned data, and

(b) accessed by said supervisor means for obtaining process control data;

filter means for accessing said first data cluster by said supervisor means, said filter means using at least a first measurement for said first process influencing factor.

54. An apparatus, as claimed in claim 53, wherein:

said first data cluster includes process control parameter data for controlling the coal drying process and applicability data for determining when said process control parameter data is applicable for controlling the coal drying process, said applicability data having a value for said first substantially uncontrollable process influencing factor.

55. An apparatus, as claimed in claim 53, wherein:

said supervisor means includes tuning means for modifying control data used by a process attribute controller in controlling a process attribute, said tuning means provides for at least one of;

(a) shifting a setpoint for the process attribute toward a current measurement of the process attribute;

(b) changing a time interval relating to a frequency of activation of the process attribute controller.

56. An apparatus, as claimed in claim 55, wherein:

the process attribute is one of;

free oxygen content of a heat transport gas, a drying chamber temperature, a deck pressure differential and an exhaust fan temperature.

57. An apparatus, as claimed in claim 53, wherein:

said supervisor means further includes an alarm processing means for detecting at least one of;

(a) a process measurement outside of a predetermined bound, and (b) a trend in process control related data different from a predetermined expected trend.

58. An apparatus, as claimed in claim 53, wherein:

said supervisor means further includes an alarm processing means for alleviating at least one alarm condition without requiring manual intervention.

59. An apparatus, as claimed in claim 53, wherein:

said supervisor means includes an object-oriented knowledge base for storing at least one of;

(a) process data related to a process attribute measurement, and (b) alarm data.

60. An apparatus, as claimed in claim 53, wherein:

said library means includes an access means for supplying a data cluster from said plurality of data clusters to said supervisor means, said access means determining when to provide said supervisor means with a data cluster by monitoring process related data output by said supervisor means.

61. An apparatus, as claimed in claim 53, wherein:

said learning means includes a neural net for learning said at least one relationship between said first and second data.

62. An apparatus, as claimed in claim 53, wherein:

said filter means includes means for determining a performance related measurement for said first data cluster, said performance related measurement related to at least one of the following for a process attribute;

(a) a number of times the process attribute is detected as deviating from a related setpoint by a predetermined amount;

(b) a size of a process attribute deviation from a related setpoint;

(c) a deadband.

63. An apparatus, as claimed in claim 62, wherein said first data cluster is accessed by said supervisor means only when said performance related measurement is within a predetermined range.

64. An apparatus, as claimed in claim 53, wherein:

said filter means includes means for comparing applicability data, related to said first data cluster, for said first process influencing factor with data ranges determined using measurements of said first process influencing factor.

65. An apparatus, as claimed in claim 53, further including a generating/filtering means for receiving said learned data and providing data clusters as input to said library means, said generating/filtering means comparing said learned data with predetermined data related to previous coal drying processing.

66. An apparatus for controlling a process comprising:

manager means for using process control data in controlling the process;

learning means for;

(A1) learning at least a first process relationship between;

(i) first data, derived from output by the process related to one or more process control parameters, and

(ii) second data related to at least a first process influencing factor, said first process influencing factor has an uncontrolled variation that influences the process sufficiently so that the process is controlled to accommodate the uncontrolled variation; and

(A2) outputting learned data related to said first process relationship;

access means for supplying a first data cluster having process control data to said manager means, said first data cluster generated from the output learned data, said access means including filter means for determining access by said manager means to said first data cluster, said filter means using at least a measurement of said first process influencing factor in determining access to said first data cluster.

67. An apparatus as claimed in claim 66, further including repository means for storing said first data cluster as one of a plurality of data clusters, each data cluster having process control data for said manager means when supplied by said access means.

68. An apparatus as claimed in claim 66, wherein said manager means includes an expert system means for processing one of:

process alarm rules and process adjusting rules.

69. An automated control processing method for controlling a process in real time comprising:

activating each attribute controller of a plurality of attribute controllers, each said attribute controller used in controlling the process, each said attribute controller having an activation frequency or execution frequency for determining a time interval between successive activations of the attribute controller;

stabilizing the process being controlled so that a measurement of process stability for the process is within a predetermined acceptable range;

adjusting a first execution frequency for a first attribute controller of said plurality of controllers wherein said step of adjusting includes;

(A1) determining an execution frequency increment for adjusting said first execution frequency;

(A2) adding said execution frequency increment to said first execution frequency.

70. A method, as claimed in claim 69, wherein:

said first attribute controller controls one of;

free oxygen within a heat transport gas, a drying chamber temperature, a deck pressure differential and an exhaust fan temperature.

71. An automated control processing method comprising the steps of:

conducting a process which is controlled as the process is performed;

providing a stored plurality of data clusters wherein each data cluster of said stored plurality of data clusters is used in controlling the process;

accessing a first data cluster of said stored plurality of data clusters by a process manager for controlling the process;

replacing access of said first data cluster by the process manager with access to a second data cluster of said stored plurality of data clusters for controlling the process;

post-filtering said stored plurality of data clusters for retaining said first data cluster in said stored plurality of data clusters for a subsequent step of accessing by the process manager, said post-filtering including determining whether the process manager provides acceptable performance in controlling the process during said step of accessing.