Virtual continuous emission monitoring system with sensor validation

• US 5,386,373 A
• Filed: 08/05/1993
• Issued: 01/31/1995
• Est. Priority Date: 08/05/1993
• Status: Expired due to Term

• Current Assignee: Rockwell Automation Technologies Incorporated (Allen-Bradley Co., Inc.)
• Original Assignee: Pavilion Technologies Incorporated (Rockwell Automation, Inc.)
• Inventors: Keeler, James D., Havener, John P., Ferguson, Ralph B., Godbole, Devendra
• Primary Examiner(s): NOLAND, THOMAS
• Application Number:
  US08/102,405
• Time in Patent Office: 544 Days
• Field of Search: 73/1 G, 73/1 R, 73/23.21, 73/31.02, 73/31.01, 73/31.03, 395/20, 395/21, 395/22, 395/23, 395/904, 395/906, 395/907, 364/500, 364/551.01, 364/150, 364/151, 364/152, 364/571.01-571.08, 364/577, 364/578, 364/579, 364/492, 364/495, 364/474.15, 364/474.16 , 364/474.19, 364/222.1, 364/226.7, 364/226.8, 364/226.9, 364/221-221.9
• US Class Current: 700/266
• CPC Class Codes: F02D 2041/1433 using a model or simulation of the system
• CPC Class Codes: F02D 41/1401 characterised by the control or regulation method F02D41/1473, F02D41/1477 take precedence
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: G01N 33/0034 comprising neural networks or related mathematical techniques
• CPC Class Codes: G01N 33/0075 for multiple spatially distributed sensors, e.g. for environmental monitoring transmission systems for measured values G08C
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks

Abstract

A continuous emission monitoring system for a manufacturing plant (10) includes a control system (16) which has associated therewith a virtual sensor network (18). The network (18) is a predictive network that receives as inputs both control values to the plant (10) and also sensor values. The network (18) is then operable to map the inputs through a stored representation of the plant (10) to output a predicted pollutant sensor level. This predicted pollutant sensor level is essentially the prediction of an actual pollutant sensor level that can be measured by a pollutant sensor (14). The network (18) therefore is a substitute for the pollutant sensor (14), thus providing a virtual sensor. The sensor values from the plant (10) are first processed through a sensor validation system (22). The sensor validation system (22) is operable to determine which of the sensors falls outside of acceptable limits and then substitutes therefor a predicted output, such that the inputs to the virtual sensor network (18) are within acceptable limits.

Claims

1. A method for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising the steps of:

measuring on a periodic basis the level of pollutants emitted by the plant;

measuring the control values to the plant and the sensor values;

storing a representation of the plant in a virtual sensor predictive network having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant comprising inputs to the virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values, the virtual sensor predictive network operable to be trained upon the measured level of pollutants, the measured control values and the sensor values to generate the stored representation;

generating an error when the difference between the measured level of pollutants and the predicted pollutant level exceeds a predetermined threshold;

retraining the virtual sensor predictive network upon the generation of an error; and

controlling the plant utilizing the predicted pollutant level in accordance with a predetermined control scheme to vary the control values.

2. The method of claim 1, wherein the step of periodically measuring the level of pollutants comprises periodically disposing an external emissions sensor in a predetermined location within the manufacturing plant to measure the level of pollutants emitted by the plant and, after the level of pollutants is measured, removing the external emissions sensor.

3. The method of claim 1, wherein the step of measuring the control values to the plant and the sensor values comprises measuring only select ones of the control values to the plant and select ones of the sensor values to comprise inputs to the virtual sensor predictive network such that only the ones of the control values and the ones of the sensor values that substantially affect the predicted pollutant level output of the virtual sensor predictive network are selected.

4. The method of claim 1, wherein the virtual sensor predictive network comprises a neural network, wherein the step of storing the representation of the plant comprises:

providing an input layer and inputting to the input layer the control values and the sensor values;

providing an output layer that is operable to output at least the predicted pollutant level; and

providing at least one hidden layer and storing the representation of the plant in the hidden layer and mapping the input layer to the output layer through the stored representation in the at least one hidden layer.

5. The method of claim 4, and further comprising, storing the measured level of pollutants, the measured control values and the sensor values in a training database and training the neural network on the training database using the measured level of pollutants as a target output.

6. The method of claim 1, wherein the step of storing the representation of the plant in a virtual sensor predictive network comprises storing the representation of the plant in a non-linear virtual sensor predictive network.

7. The method of claim 1, wherein the step of controlling the plant utilizing a predicted pollutant level comprises controlling the pollutant level output by the plant utilizing a predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

8. The method of claim 1, and further comprising:

determining if any of the sensor values are outside of acceptable limits that are defined for each of the associated sensors, in accordance with predetermined criteria; and

substituting an associated known value for each of the sensor values determined to be outside of the acceptable limits for the associated sensor.

9. The method of claim 8, wherein the step of substituting comprises predicting from a past history of the sensor values associated with the sensor determined to have a sensor value outside of the associated acceptable limits, a predicted value, the predicted value comprising the known value for that sensor.

10. The method of claim 9, wherein the past history is a function of the sensor values for the other sensors.

11. The method of claim 8, wherein the step of determining comprises:

providing a sensor validation predictive network having as an input the actual sensor values of the plant, the sensor validation predictive network having associated therewith a stored representation of each of the actual sensor values as a function of the other of the actual sensor values to provide on the output thereof a predicted sensor value for each of the actual sensor values input thereto;

predicting with the predictive network the predicted sensor values; and

comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

12. The method of claim 11, herein the step of substituting comprises replacing each of the actual sensor values that are input to the sensor validation predictive network with the predicted sensor values of the ones of the predicted sensor values determined to be outside of the predetermined limits.

13. A method for monitoring emissions in a manufacturing plant that is operable to generate pollutants, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising the steps of:

storing a representation of the plant in a virtual sensor predictive network having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant comprising inputs to the virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values;

determining if the sensor values for any of the sensors are outside of acceptable limits in accordance with predetermined criteria, which predetermined criteria is defined for each of the sensors; and

substituting a known value for the sensor values of the ones of the sensors having sensor values determined to be outside of the acceptable limits by predicting from a past history of the sensor value to be substituted, a predicted sensor value for that sensor value, the predicted sensor value comprising the known value.

14. The method of claim 13, wherein the past history is a function of the sensor values associated with the other sensors.

15. The method of claim 13, wherein the step of determining comprises:

providing a sensor validation predictive network having as an input the actual sensor values output by the plant, the sensor validation predictive network having associated therewith a stored representation of each of the actual sensor values for each of the sensors as a function of the actual sensor values of the other of the sensors to provide on the output thereof the predicted sensor value for each of the actual sensor values input thereto;

predicting with the sensor validation predictive network the predicted sensor values; and

comparing the difference of the input sensor values with the predicted sensor values with predetermined limits for that difference.

16. The method of claim 15, wherein the step of substituting comprises replacing each of the actual sensor values that are input to the sensor validation predictive network with the predicted sensor values of the ones of the predicted sensor values determined to be outside of the predetermined limits.

17. A system for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising:

a pollutant sensor for measuring on a periodic basis the level of pollutants emitted by the plant;

a plant information system for measuring the control values to the plant and the sensor values of the plant;

a virtual sensor predictive network for storing a representation of the plant and having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant from said plant information system comprising inputs to said virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values, said virtual sensor predictive network operable to be trained upon the measured level of pollutants, the measured control values and the sensor values to generate the stored representation;

said pollutant sensor operable to compare the measured level of pollutants with the predicted pollutant level output by said virtual sensor predictive network and generate an error when the difference between the measured level of pollutants and the predicted pollutant level exceeds a predetermined threshold;

a training system for retraining said virtual sensor network upon the generation of an error; and

a controller for controlling the plant utilizing the predicted pollutant level output by said virtual sensor predictive network in accordance with a predetermined control scheme to vary the control values.

18. The system of claim 17, wherein said plant information system is operable to measure only select ones of the control values to the plant and select ones of the sensor values to comprise inputs to said virtual sensor predictive network such that only the ones of the control values and the ones of the sensor values that substantially affect the predicted pollutant level output of said virtual sensor predictive network are selected.

19. The system of claim 17, wherein said virtual sensor predictive network comprises a neural network, said neural network comprising:

an input layer for receiving as inputs the control values and the sensor values;

an output layer that is operable to output at least the predicted pollutant level; and

at least one hidden layer for storing the representation of the plant therein and mapping said input layer to said output layer through the stored representation in said at least one hidden layer.

20. The system of claim 19, and further comprising, a training database for storing the measured level of pollutants, the measured control values and the sensor values and a training system for training said neural network on said training database using the measured level of pollutants as a target output.

21. The system of claim 17, wherein said virtual sensor predictive network comprises a non-linear virtual sensor predictive network.

22. The system of claim 17, wherein said controller is operable to control the pollutant level output by the plant utilizing said predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

23. The system of claim 17, and further comprising:

a sensor validation system for determining as invalid the sensor values if the sensor values are outside of acceptable limits that are defined for each of the associated sensors, in accordance with predetermined criteria; and

a substitute sensor for substituting a known value for each of the sensor values determined to be outside of the acceptable limits, each sensor having a known value associated therewith.

24. The system of claim 23, wherein said substitute sensor comprises a predictive network for generating a predicted sensor value for the invalid one of the sensor values from a past history of the sensor values to be substituted, said predicted sensor value comprising said known value.

25. The system of claim 24, wherein said past history is a function of the sensor values for the other sensors.

26. The system of claim 23, wherein said sensor validation system comprises:

a sensor validation predictive network having as an input select ones of the actual sensor values of the plant, said sensor validation predictive network having associated therewith a stored representation of each of the select actual sensor values as a function of the other of the input actual sensor values to provide on the output thereof a predicted value for each of the actual sensor input thereto; and

a comparator for comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

27. The system of claim 26, wherein said substitute sensor comprises a switching device for replacing each of the select actual sensor values that are input to said sensor validation predictive network with the predicted values of the ones of the predicted values determined to be outside of the predetermined limits.

28. A system for monitoring emissions in a manufacturing plant that is operable to generate pollutants, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising:

a virtual sensor predictive network for storing a representation of the plant and having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values to the plant and the sensor values of the plant comprising inputs to said virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values;

a sensor validation system for determining if select ones of the sensor values associated with select ones of the sensors are outside of acceptable limits in accordance with predetermined criteria; and

a substitute sensor for substituting a known value for the select sensor values when it is determined that they are outside of the acceptable limits, said substitute sensor comprising a predictive network for generating a predicted value for the one of the select sensor values determined to be outside of the acceptable limits, said predictive network generating said predicted value from a past history of the select sensor value to be substituted, said predicted value comprising said known value.

29. The system of claim 28, wherein the past history is a function of the select sensor values from other of the select sensors.

30. The system of claim 28, wherein said sensor validation system comprises:

a sensor validation predictive network having as an input select ones of the actual sensor values of the plant, said sensor validation predictive network having associated therewith a stored representation of each of the select actual sensor values as a function of the other of the input actual sensor values to provide on the output thereof a predicted value for each of the actual sensor values input thereto; and

a comparator for comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for sensor value inputs.

31. The system of claim 30, wherein said substitute sensor comprises a switching device for replacing each of the select actual sensor values that are input to said sensor validation predictive network with the predicted values of the ones of the predicted values determined to be outside of the predetermined limits.

32. A method for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising the steps of:

measuring the level of pollutants emitted by the plant as a function of time;

measuring the control values to the plant and the sensor values as a function of time, the measured level of pollutants and the measured control values and sensor values providing a training dataset;

storing a representation of the plant in a virtual sensor predictive network having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant comprising inputs to the virtual sensor predictive network, the stored representation learned by training the virtual sensor predictive network on the training dataset; and

controlling the plant utilizing the predicted pollutant level in accordance with a predetermined control scheme to vary the control values.

33. The method of claim 32, wherein the step of measuring the level of the pollutants emitted by the plant comprises measuring the level of pollutants emitted by the plant at least one time in order to learn and store the representation of the plant in the virtual sensor predictive network.

34. The method of claim 33, wherein the step of measuring the level of pollutants at least one time comprises:

measuring the level of pollutants on a periodic basis and comparing it with the predicted pollutant level and generating an error when the difference between the measured level of pollutants and the predicted pollutant level exceeds a predetermined threshold; and

retraining the network upon the generation of an error.

35. The method of claim 34, wherein the step of periodically measuring the level of pollutants comprises periodically disposing an external emissions sensor in a predetermined location within the manufacturing plant to measure the level of pollutants emitted by the plant and, after the level of pollutants is measured, removing the external emissions sensor.

36. The method of claim 32, wherein the step of measuring the control values to the plant and the sensor values comprises measuring only select ones of the control values to the plant and select ones of the sensor values to comprise inputs to the virtual sensor predictive network such that only the ones of the control values and the ones of the sensor values that substantially affect the predicted pollutant level output of the virtual sensor predictive network are selected.

37. The method of claim 32, wherein the virtual sensor predictive network comprises a neural network, wherein the step of storing the representation of the plant comprises:

providing an input layer and inputting to the input layer the control values and the sensor values;

providing an output layer that is operable to output at least the predicted pollutant level; and

providing at least one hidden layer and storing the representation of the plant in the hidden layer and mapping the input layer to the output layer through the stored representation in the at least one hidden layer.

38. The method of claim 32, wherein the step of storing the representation of the plant in the virtual sensor predictive network comprises storing the representation of the plant in a non-linear virtual sensor predictive network.

39. The method of claim 32, wherein the step of controlling the plant utilizing a predicted pollutant level comprises controlling the pollutant level output by the plant utilizing a predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

40. The method of claim 32, and further comprising:

determining if any of the sensor values or any of the control values are outside of acceptable limits that are defined for each of the sensor values and control values in accordance with predetermined criteria; and

substituting an associated known value for the each of the sensor values or each of the control values determined to be outside of the acceptable limits.

41. The method of claim 40, wherein the step of substituting comprises predicting from a past history of the sensor value or control value to be substituted, a predicted value, the predicted value comprising the known value.

42. The method of claim 41, wherein the past history is a function of the other sensor values or control values.

43. The method of claim 40 wherein the step of determining comprises:

providing a sensor validation predictive network having as an input the actual sensor values of the plant, the sensor validation predictive network having associated therewith a stored representation of each of the actual sensor values as a function of the other of the actual sensor values to provide on the output thereof a predicted sensor value for each of the actual sensor values input thereto;

predicting with the sensor validation predictive network the predicted sensor values; and

comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

44. The method of claim 43, wherein the step of substituting comprises replacing each of the actual sensor values that are input to the sensor validation predictive network with the predicted sensor values of the ones of the predicted sensor values determined to be outside of the predetermined limits.

45. A method for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising the steps of:

measuring the level of pollutants emitted by the plant;

measuring the control values to the plant and the sensor values;

storing a representation of the plant in a virtual sensor neural network having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant by the steps of;

providing an input layer and inputting to the input layer the control values and the sensor values,

providing an output layer that is operable to output at least the predicted pollutant level, and

providing at least one hidden layer and storing the representation of the plant in the hidden layer and mapping the input layer to the output layer through the stored representation in the at least one hidden layer;

the control values of the plant and the sensor values of the plant comprising inputs to the input layer of the virtual sensor neural network, the stored representation learned from the measured pollutant level, the control values and the sensor values; and

controlling the plant utilizing the predicted pollutant level in accordance with a predetermined control scheme to vary the control values.

46. The method of claim 45, wherein the step of measuring the level of the pollutants emitted by the plant comprises measuring the level of pollutants emitted by the plant at least one time in order to learn and store the representation of the plant in the virtual sensor neural network.

47. The method of claim 46, wherein the step of measuring the level of pollutants at least one time comprises:

measuring the level of pollutants on a periodic basis and comparing it with the predicted pollutant level and generating an error when the predicted pollutant level exceeds a predetermined threshold;

the virtual sensor neural network operable to be trained upon the measured level of pollutants, the measured control values and the sensor values to generate the stored representation; and

retraining the virtual sensor neural network upon the generation of an error.

48. The method of claim 47, wherein the step of periodically measuring the level of pollutants comprises periodically disposing an external emissions sensor in a predetermined location within the manufacturing plant to measure the level of pollutants emitted by the plant and, after the level of pollutants is measured, removing the external emissions sensor.

49. The method of claim 45, wherein the step of measuring the level of pollutants emitted by the plant comprises measuring the level of pollutants emitted by the plant as a function of time and wherein the step of measuring the control values to the plant and the sensor values are performed as a function of time, the measured level of pollutants and the measured control values and sensor values providing a training dataset wherein the step of storing a representation of the plant in the virtual sensor neural network comprises training the virtual sensor neural network on the training dataset to generate the stored representation.

50. The method of claim 45, wherein the step of measuring the control values to the plant and the sensor values comprises measuring only select ones of the control values to the plants and select ones of the sensor values to comprise inputs to the virtual sensor neural network such that only the ones of the control values and the ones of the sensor values that substantially affect the predicted pollutant level output of the virtual sensor neural network are selected.

51. The method of claim 45, and further comprising, storing the measured level of pollutants, the measured control values and the sensor values in a training database and training the virtual sensor neural network on the training database using the measured level of pollutants as a target output.

52. The method of claim 45, wherein the step of controlling the plant utilizing a predicted pollutant level comprises controlling the pollutant level output by the plant utilizing a predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

53. The method of claim 45, and further comprising:

determining if any of the sensor values or any of the control values are outside of acceptable limits that are defined for each of the sensor values and control values in accordance with predetermined criteria; and

substituting an associated known value for the each of the sensor values or each of the control values determined to be outside of the acceptable limits.

54. The method of claim 53, wherein the step of substituting comprises predicting from a past history of the sensor value or control value to be substituted, a predicted value, the predicted value comprising the known value.

55. The method of claim 54, wherein the past history is a function of the other sensor values or control values.

56. The method of claim 53, wherein the step of determining comprises:

providing a sensor validation predictive network having as an input the actual sensor values of the plant, the sensor validation predictive network having associated therewith a stored representation of each of the actual sensor values as a function of the other of the actual sensor values to provide on the output thereof a predicted sensor value for each of the actual sensor values input thereto;

predicting with the sensor validation predictive network the predicted sensor values; and

comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

57. The method of claim 56, wherein the step of substituting comprises replacing each of the actual sensor values that are input to the sensor validation predictive network with the predicted sensor values of the ones of the predicted sensor values determined to be outside of the predetermined limits.

58. A method for monitoring emissions in a manufacturing plant that is operable to generate pollutants, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising the steps of:

storing a representation of the plant in a virtual sensor predictive network having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant comprising inputs to the virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values;

providing a sensor validation predictive network having as an input the actual sensor values of the plant, the sensor validation predictive network having associated therewith a stored representation of each of the actual sensor values as a function of the other of the actual sensor values to provide on the output thereof a predicted sensor value for each of the actual sensor values input thereto;

predicting with the sensor validation predictive network the predicted sensor values;

determining the difference between the input sensor values for each of the sensors and the associated predicted sensor values;

comparing the difference determined for each of the sensors with predetermined limits associated with each of the sensors for that difference; and

substituting a known value for the sensor values of each of the sensors wherein the determined difference exceeds the predetermined limits.

59. The method of claim 58, wherein the step of substituting comprises predicting from a past history of the sensor value to be substituted, a predicted sensor value, the predicted sensor value comprising the known value.

60. The method of claim 59, wherein the past history is a function of the other sensor values.

61. The method of claim 58, wherein the step of substituting comprises replacing each of the actual sensor values that are input to the sensor validation predictive network with the predicted sensor values of the ones of the sensors having predicted sensor values determined to be outside of the predetermined limits.

62. A system for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls operating in response to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising:

a pollutant sensor for measuring the level of pollutants emitted by the plant;

a plant information system for measuring only select ones of the control values to the plant and select ones of the sensor values of the plant;

a virtual sensor predictive network for storing a representation of the plant and having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the select ones of the control values to the plant and the select ones of the sensor values of the plant from said plant information system comprising inputs to said virtual sensor predictive network, wherein only the ones of the control values to the plant and the ones of the sensor values from the plant that substantially affect the predicted pollutant level output of said virtual sensor predictive network are selected by said plant information system, the stored representation learned from the measured pollutant level, the control values and the sensor values; and

a controller for controlling the plant utilizing the predicted pollutant level output by said virtual sensor predictive network in accordance with a predetermined control scheme to vary the control values.

63. The system of claim 62, wherein said pollutant sensor is operable to measure the level of pollutants emitted by the plant at least one time in order to learn and store the representation of the plant in the said virtual sensor predictive network.

64. The system of claim 63, wherein:

said pollutant sensor is operable to measure the level of pollutants on a periodic basis and compare the measured level of pollutants with the predicted pollutant level output by said virtual sensor predictive network and generate an error when the difference between the predicted pollutant level and the measured level of pollutants exceeds a predetermined threshold;

said virtual sensor predictive network operable to be trained upon the measured level of pollutants, the control values and the sensor values to generate the stored representation; and

further comprising;

a training system for retraining said virtual sensor predictive network upon the generation of an error.

65. The system of claim 62, wherein said virtual sensor predictive network comprises a neural network, said neural network comprising:

an input layer for receiving as inputs the control values and the sensor values;

an output layer that is operable to output at least the predicted pollutant level; and

at least one hidden layer for storing the representation of the plant therein and mapping said input layer to said output layer through the stored representation in said at least one hidden layer.

66. The system of claim 65, and further comprising, a training database for storing the measured level of pollutants, the control values and the sensor values and a training system for training said neural network on said training database using the measured level of pollutants as a target output.

67. The system of claim 62, wherein said virtual sensor predictive network comprises a non-linear virtual sensor predictive network.

68. The system of claim 62, wherein said controller is operable to control the pollutant level output by the plant utilizing a predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

69. The system of claim 62, and further comprising:

a sensor validation system for determining as invalid the sensor values if the sensor values are outside of acceptable limits that are defined for each of the associated sensors, in accordance with predetermined criteria; and

a substitute sensor for substituting a known value for each of the sensor values determined to be outside of the acceptable limits, each sensor having a known value associated therewith.

70. The system of claim 69, wherein said substitute sensor comprises a predictive network for generating a predicted sensor value for the invalid one of the sensor values from a past history of the sensor values to be substituted, said predicted sensor value comprising said known value.

71. The system of claim 70, wherein said past history is a function of the sensor values associated with the other sensors.

72. The system of claim 69, wherein said sensor validation system comprises:

a sensor validation predictive network having as an input select ones of the actual sensor values of the plant, said sensor validation predictive network having associated therewith a stored representation of each of the select actual sensor values as a function of the other of the input actual sensor values to provide on the output thereof a predicted value for each of the actual sensor values input thereto; and

a comparator for comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

73. The system of claim 69, wherein said substitute sensor comprises a switching device for replacing each of the select actual sensor that are input to said sensor validation predictive network with the predicted values of the ones of the predicted values determined to be outside of the predetermined limits.

74. A system for monitoring emissions in a manufacturing plant that are generated as a by-product of the operation of the plant, the manufacturing plant having controls responsive to received control values to alter the operation of the plant and sensors to measure the operating parameters of the plant and output associated sensor values, comprising:

a pollutant sensor for measuring the level of pollutants emitted by the plant;

a plant information system for measuring the control values to the plant and the sensor values of the plant;

a virtual sensor neural network for storing a representation of the plant and having;

an input layer for receiving as inputs the control values and the sensor values,

an output layer that is operable to output at least a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the stored representation learned from the measured pollutant level, the control values and the sensor values, and

at least one hidden layer for storing the representation of the plant therein and mapping said input layer to said output layer through the stored representation in said at least one hidden layer; and

a controller for controlling the plant utilizing the predicted pollutant level output by said virtual sensor neural network in accordance with a predetermined control scheme to vary the control values.

75. The system of claim 74, wherein said pollutant sensor is operable to measure the level of pollutants emitted by the plant at least one time in order to learn and store the representation of the plant in the said virtual sensor neural network.

76. The system of claim 75, wherein:

said pollutant sensor is operable to measure the level of pollutants on a periodic basis and compare the measured level of pollutants with the predicted pollutant level output by said virtual sensor neural network and generate an error when the difference between the predicted pollutant level and the measured level of pollutants exceeds a predetermined threshold;

said virtual sensor neural network operable to be trained upon the measured level of pollutants, the control values and the sensor values to generate the stored representation; and

further comprising;

a training system for retraining the virtual sensor neural network upon the generation of an error.

77. The system of claim 74, wherein said plant information system is operable to measure only select ones of the control values to the plant and select ones of the sensor values to comprise inputs to said virtual sensor neural network such that only the ones of the control values and the ones of the sensor values that substantially affect the predicted pollutant level output of said virtual sensor neural network are selected.

78. The system of claim 74, and further comprising, a training database for storing the measured level of pollutants, the control values and the sensor values and a training system for training said virtual sensor neural network on said training database using the measured level of pollutants as a target output.

79. The system of claim 74, wherein said controller is operable to control the pollutant level output by the plant utilizing the predicted pollutant level to maintain the predicted pollutant level within a predetermined range.

80. The system of claim 74, and further comprising:

a sensor validation system for determining as invalid the sensor values if the sensor values are outside of acceptable limits that are defined for each of the associated sensors, in accordance with predetermined criteria; and

a substitute sensor for substituting a known value for each of the sensor values determined to be outside of the acceptable limits, each sensor having a known value associated therewith.

81. The system of claim 80, wherein said substitute sensor comprises a predictive network for generating a predicted sensor value for the invalid one of the sensor values from a past history of the sensor value to be substituted, said predicted sensor value comprising said known value.

82. The system of claim 81, wherein said past history is a function of the sensor values associated with the other sensors.

83. The system of claim 80, wherein said sensor validation system comprises:

a sensor validation predictive network having as an input select ones of the actual sensor values of the plant, said sensor validation predictive network having associated therewith a stored representation of each of the select actual sensor values as a function of the other of the input actual sensor values to provide on the output thereof a predicted value for each of the actual sensor values input thereto; and

a comparator for comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences.

84. The system of claim 83, wherein said substitute sensor comprises a switching device for replacing each of the select actual sensor values that are input to said sensor validation predictive network with the predicted values of the ones of the predicted values determined to be outside of the predetermined limits.

85. A system for monitoring emissions in a manufacturing plant that is operable to generate pollutants, the manufacturing plant having controls responsive to control values to alter the operation of the plant and sensors outputting sensor values as a measure the operating parameters of the plant, comprising:

a virtual sensor predictive network for storing a representation of the plant and having as an output a predicted pollutant level that corresponds to the actual pollutant level output by the manufacturing plant, the control values of the plant and the sensor values of the plant comprising inputs to said virtual sensor predictive network, the stored representation learned from the measured pollutant level, the control values and the sensor values;

a sensor validation predictive network having as an input select ones of the actual sensor values of the plant and actual control values of the plant, said sensor validation predictive network having associated therewith a stored representation of each of the select actual sensor values and actual control values as a function of the other of the input actual sensor and control values to provide on the output thereof a predicted value for each of the actual sensor and control values input thereto; and

a comparator for comparing each of the differences of the input actual sensor values and the predicted sensor values with predetermined limits for those differences to determine if select ones of the sensor and control values are outside of those predetermined limits; and

a substitute sensor for substituting a known value for the select sensor or control values when it is determined that they are outside of the predetermined limits.

86. The system of claim 85, wherein said substitute sensor comprises a switching device for replacing each of the select actual sensor values that are input to said sensor validation predictive network with the predicted values of the ones of the predicted values determined to be outside of the predetermined limits.