Predictive emissions monitoring using a statistical hybrid model

• US 20090070047A1
• Filed: 08/15/2008
• Published: 03/12/2009
• Est. Priority Date: 03/18/2005
• Status: Active Grant

• Current Assignee: CMC Solutions, LLC
• Original Assignee: CMC Solutions, LLC
• Inventors: Swanson, Brian G.
• Granted Patent: US 8,768,664 B2
• Time in Patent Office: Days
• US Class Current: 702/32
• CPC Class Codes: G05B 17/02 electric

Abstract

A method for generating a model representative of a process. The process includes a result variable representing a product of the process, and a plurality of process variables representing characteristics of the process other than the product of the process. The method includes steps of acquiring a plurality of test values of the result variable; acquiring, for each process variable of the plurality of process variables, a plurality of test values of the process variable, each test value being associated with a result variable value; providing, for each first coefficient of a plurality of first coefficients, a separate test value of the first coefficient associated with each process variable; and, for each second coefficient of a plurality of second coefficients, providing a separate test value of the second coefficient associated with each test value of each process variable. At least one of the method steps is a computer-implemented step.

Claims

1. A method for generating a model representative of a process, the process including a result variable representing a product of the process, and a plurality of process variables representing characteristics of the process other than the product of the process, the method comprising the steps of:

acquiring a plurality of test values of the result variable, each test value of the plurality of result variable test values being measured at a corresponding point in time;

for each process variable of the plurality of process variables, acquiring a plurality of test values of the process variable, each test value of the plurality of process variable test values being associated with a result variable value of the plurality of result variable values, each test value of the plurality of process variable test values being measured at a point in time substantially simultaneous with a point in time at which one of the test values of the plurality of result variable test values is measured;

for each first coefficient of a plurality of first coefficients, providing a separate test value of the first coefficient associated with each process variable of the plurality of process variables, each separate test value of the first coefficient being a function of at least a portion of the test values of the plurality of test values of the associated process variable; and

for each second coefficient of a plurality of second coefficients, providing a separate test value of the second coefficient associated with each test value of each process variable of the plurality of process variables, each separate test value of the second coefficient being a function of at least a portion of the test values of the plurality of test values of the associated process variable, wherein at least one of the above steps is a computer-implemented step.

2. A computer system including computer readable program code means stored in an element thereof, for performing a method in accordance with claim 1.

3. A computing device including computer readable program code means stored in a memory thereof for performing a method in accordance with claim 1.

4. A computer readable medium encoded with computer readable instructions for performing a method in accordance with claim 1.

5. A predictive emissions monitoring system comprising a computer readable medium in accordance with claim 4.

6. A power generation system comprising:

an emissions source; and

a predictive emissions monitoring system in accordance with claim 5 operatively coupled to the emissions source.

7. An article of manufacture comprising a computer-usable medium having computer-readable program code means embodied therein for performing a method in accordance with claim 1.

8. A computer data signal embodied in a transmission medium, the data signal comprising computer-readable source code for performing a method in accordance with claim 1.

9. A method for predicting a quantitative measure of a characteristic of an emission from an emissions source, comprising the steps of:

acquiring a plurality of test values of the characteristic, each test value of the plurality of characteristic test values being measured at a corresponding point in time;

for each process variable of a plurality of process variables, acquiring a plurality of test values of the process variable, each test value of the plurality of process variable test values being associated with a test value of the plurality of first variable test values, each test value of the plurality of process variable test values being measured at a point in time substantially simultaneous with a point in time at which one of the test values of the plurality of first variable test values is measured;

a computer-implemented step of, for each process variable of the plurality of process variables, and using the plurality of test values of the process variable, calculating an associated test correlation coefficient indicating a correlation between the process variable and the characteristic;

a computer-implemented step of, for each process variable of the plurality of process variables, and using the plurality of test values of the process variable, calculating an associated test tolerance value specifying a range of values within which the test value of the process variable is located;

for each test value of the plurality of test values of each process variable of the plurality of process variables, calculating an associated test delta value equal to a difference between the test value and a most-recent prior time-successive test value of the plurality of test values of the process variable;

for each test value of the plurality of test values of each process variable of the plurality of process variables, calculating an associated test indicator variable value indicating a result of a comparison between the tolerance value associated with the process variable value and the delta value associated with the process variable value;

for each process variable of selected ones of the plurality of process variables, acquiring a plurality of comparison values of the process variable;

for each process variable of the selected ones of the plurality of process variables and using the plurality of comparison values of the process variable, calculating an associated comparison correlation coefficient indicating a correlation between the process variable and the characteristic,

for each process variable of the selected ones of the plurality of process variables and using the plurality of comparison values of the process variable, calculating an associated comparison tolerance value describing a range of values within which the comparison value of the process variable is located;

for each comparison value of the plurality of comparison values of each selected one of the plurality of process variables, calculating an associated comparison delta value equal to a difference between the comparison value and a most-recent prior time-successive comparison value of the plurality of comparison values of the process variable;

for each comparison value of the plurality of comparison values of each selected one of the plurality of process variables, calculating an associated comparison indicator value indicating a result of a comparison between the tolerance value associated with the process variable value and the delta value associated with the process variable value;

for each process variable in each predetermined combination of a plurality of predetermined combinations of the selected ones of the process variables,

iteratively comparing each comparison value of the variable with each test value of the variable, the comparison value of the tolerance associated with the process variable with the test value of the tolerance associated with the process variable, the comparison value of the indicator associated with the process variable with the test value of the indicator associated with the process variable, the comparison value of the delta associated with the process variable with the test value of the delta associated with the process variable, and the comparison value of the correlation coefficient associated with the process variable with the test value of the correlation coefficient associated with the process variable;

for each process variable in each predetermined combination of a plurality of predetermined combinations of the selected ones of the process variables,

identifying all test values of the additional variable where the test value of the additional variable differs from a comparison value of the additional variable by an amount equal to or less than a first predetermined amount,

where the test value of the delta associated with the test value of the additional variable differs from the comparison value of the delta associated with the comparison value of the additional variable by an amount equal to or less than a second predetermined amount,

and where the test value of the indicator associated with the test value of the additional variable differs from the comparison value of the indicator associated with the comparison value of the additional variable by an amount equal to or less than a third predetermined amount; and

assigning, as the predicted value of the emissions variable, an average of all of the test values of the emissions variable that are associated with respective test values of each process variable for which the test process variable values differ from the comparison process variable values by the associated predetermined amount, wherein at least one of the above steps is a computer-implemented step.

10. A computer system including computer readable program code means stored in an element thereof, for performing a method in accordance with claim 9.

11. A computer readable medium encoded with computer readable instructions for performing a method in accordance with claim 9.

12. A predictive emissions monitoring system comprising a computer readable medium in accordance with claim 11.

13. A computer data signal embodied in a transmission medium, the data signal comprising computer-readable source code for performing a method in accordance with claim 9.

14. A method for predicting a value of a first variable based on values of a plurality of additional variables, comprising the steps of:

acquiring a plurality of test values of the first variable, each test value of the plurality of first variable test values being measured at a corresponding point in time;

for each additional variable of the plurality of additional variables, acquiring a plurality of test values of the additional variable, each value of the plurality of additional variable values being associated with a value of the plurality of first variable values, each test value of the plurality of additional variable test values being measured at a point in time substantially simultaneous with a point in time at which one of the test values of the plurality of first variable test values is measured;

for each first coefficient of a plurality of first coefficients, providing a separate test value of the first coefficient associated with each additional variable of the plurality of variables, each separate test value of the first coefficient being a function of at least a portion of the test values of the plurality of test values of the associated additional variable;

for each second coefficient of a plurality of second coefficients, providing a separate test value of the second coefficient associated with each test value of each additional variable of the plurality of variables, each separate test value of the second coefficient being a function of at least a portion of the test values of the plurality of test values of the associated additional variable;

for each selected one of a plurality of selected ones of the additional variables of the plurality of additional variables, acquiring a plurality of comparison values of the additional variable;

for each first coefficient of the plurality of first coefficients, providing a separate comparison value of the first coefficient associated with the additional variable, each separate comparison value of the first coefficient being a function of at least a portion of the comparison values of the plurality of comparison values of the associated additional variable;

for each second coefficient of the plurality of second coefficients, providing a separate comparison value of the second coefficient associated with the comparison value of each additional variable, each separate comparison value of the second coefficient being a function of at least a portion of the comparison values of the plurality of comparison values of the associated additional variable;

for each additional variable in each predetermined combination of a plurality of predetermined combinations of the selected ones of the additional variables, iteratively comparing;

each comparison value of the additional variable with each test value of the additional variable; and

the comparison values of selected ones of the second coefficients associated with the comparison value of the variable with the test values of the selected ones of the second coefficients associated with the test value of the variable;

for each additional variable in each predetermined combination of a plurality of predetermined combinations of the selected ones of the additional variables, identifying all test values of the additional variable where the test value differs from a comparison value of the additional variable by an amount equal to or less than an associated predetermined amount, and where, for each of the selected ones of the second coefficients associated with each test value of the additional variable, all test values of the selected ones of the second coefficients differ from the comparison values of the selected ones of the second coefficients by an amount equal to or less than an associated predetermined amount; and

assigning, as the predicted value of the first variable, an average of all of the test values of the first variable that are associated with respective test values of each additional variable for which the test first variable values differ from the comparison additional variable values by the associated predetermined amount, wherein at least one of the above steps is a computer-implemented step.

15. A computer system including computer readable program code means stored in an element thereof, for performing a method in accordance with claim 14.

16. A computer readable medium encoded with computer readable instructions for performing a method in accordance with claim 14.

17. A predictive emissions monitoring system comprising a computer readable medium in accordance with claim 16.

18. A computer data signal embodied in a transmission medium, the data signal comprising computer-readable source code for performing a method in accordance with claim 14.