Computer-implemented system and method for analyzing mixtures of gases

• US 20060155486A1
• Filed: 10/04/2005
• Published: 07/13/2006
• Est. Priority Date: 10/07/2004
• Status: Active Grant

• Current Assignee: E. I. du Pont de Nemours and Company (Corteva, Inc.)
• Original Assignee: E. I. du Pont de Nemours and Company (Corteva, Inc.)
• Inventors: Miller, Charles E., Morris, Patricia A., Piovoso, Michael Joseph, Walsh, Alicia Marie, Dahl, Kenneth S.
• Granted Patent: US 7,460,958 B2
• Time in Patent Office: Days
• US Class Current: 702/32
• CPC Class Codes: G01N 33/0031 comprising two or more sensors, e.g. a sensor array electrochemical electrode arrays G01N27/27
• CPC Class Codes: G01N 33/0034 comprising neural networks or related mathematical techniques
• CPC Class Codes: G01N 33/0037 for NOx
• CPC Class Codes: G01N 33/004 for CO, CO2
• CPC Class Codes: G01N 33/0047 for organic compounds
• CPC Class Codes: Y02A 50/20 Air quality improvement or preservation, e.g. vehicle emission control or emission reduction by using catalytic converters

Abstract

A computer-implemented system and method for converting the resistances of an array of metal oxide sensors into constituents and concentrations of a multi-component gas provided around the sensor array. The method includes preprocessing the resistances of the sensor array; selecting a model that calculates the constituents and concentrations of the multi-component gas based on the preprocessed resistances; and post-processing the constituents and concentrations of the multi-component gas generated by the selected model to provide the actual constituents and concentrations of the multi-component gas.

Claims

1. A method for converting the responses of an array of metal oxide sensors into constituents and concentrations of a multi-component gas provided around the sensor array, the method comprising:

providing a self-diagnostic module to calculate the quality factors associated with corresponding prediction models and determine if the responses of the sensor array are within a preferred range;

providing a self-correction module to compensate for changes in the responses over time so that quality factors can be obtained which are within the preferred limits or to stop the responses and report an error; and

providing a concentration calculation module to calculate the gas concentration outputs from each prediction model, corresponding tolerance limits, and gas concentration outputs determined by weighting the outputs from the prediction models by corresponding quality factors and determining the weighted tolerance limits, so that the gas concentration outputs with the lowest tolerance limits can be reported as the actual constituents and concentrations of the multi-component gas.

2. A method as recited in claim 1, wherein the self-diagnostic module performs one of a linear transformation or a nonlinear transformation on one of the sensor responses, sensor temperature values, or any values of any auxiliary sensor data.

3. A method as recited in claim 2, wherein the self-diagnostic module performs a real-time quality factor calculation for the sensor array.

4. A method as recited in claim 3, wherein the real-time quality factor calculation utilizes a Principal Component Analysis model of one of transformed sensor responses, sensor temperature values, or auxiliary data obtained during calibration experiments.

5. A method as recited in claim 4, wherein the Principal Component Analysis model is compressed by reducing the model scores matrix to a sum of squares scores vector, thus reducing the number of model parameters that must be stored.

6. A method as recited in claim 3, wherein the real-time quality factor calculation evaluates one of the sensor responses, sensor temperature values, or auxiliary data obtained during calibration experiments to determine whether the sensor array is operating within predetermined limits.

7. A method as recited in claim 3, wherein the real-time quality factor calculation determines the pattern of one of the sensor responses, sensor temperature values, or auxiliary data obtained during calibration experiments.

8. A method as recited in claim 3, wherein the real-time quality factor calculation determines the covariance of one of the sensor responses, sensor temperature values, or auxiliary data obtained during calibration experiments.

9. A method as recited in claim 3, wherein the real-time quality factor calculation determines that the sensor array is not operating within the preferred range for any of the prediction models and that the self-correction module must be used.

10. A method as recited in claim 3, wherein the real-time quality factor calculation determines that the sensor array is operating within the preferred range for one or more of the prediction models and outputs from all prediction models are transferred to the concentration calculation module.

11. A method as recited in claim 1, wherein the self-correction module provides a self-referencing process for the sensor array in real-time.

12. A method as recited in claim 11, wherein the self-referencing process comprises increasing the temperature of the sensor array, measuring lambda around the sensor array, and scaling the sensor responses to responses at a predetermined reference state.

13. A method as recited in claim 1, wherein the self-correction module provides a temperature correction process for the sensor array in real-time.

14. A method as recited in claim 13, wherein the temperature correction process comprises removing variations in the sensor responses due to temperature using a fixed relationship between temperature and resistance.

15. A method as recited in claim 13, wherein the temperature correction process comprises removing variations in the sensor responses due to temperature using a recursive algorithm to provide a time-varying relationship between temperature and resistance.

16. A method as recited in claim 1, wherein the self-correction module provides a drift correction process for the sensor array in real-time.

17. A method as recited in claim 16, wherein the drift correction process comprises adjusting the sensor responses for both offset and span, and determining the offset and span correction factors at any given time using real-time sensor responses, sensor temperature, and available auxiliary engine data.

18. A method as recited in claim 1, wherein the self-correction module determines that the self-correction module cannot compensate for the changes in the sensor responses, stops the sensor array from processing data for a predetermined time, reports an error to the sensor array, and begins processing data again after the predetermined time.

19. A method as recited in claim 1, wherein the self-correction module determines that the self-correction module cannot compensate for the changes in the sensor responses, stops the sensor array from processing data for a predetermined time, reports and error to the sensor array, and does not begin processing data again if the number of sequential errors is greater than a predetermined number.

20. A method as recited in claim 1, wherein the concentration calculation module calculates the gas concentration outputs and tolerance limits for each prediction model.

21. A method as recited in claim 20, wherein the gas concentration outputs and tolerance limits for each prediction model are calculated using at least one of a Projection to Latent Structures (PLS) model or another linear model.

22. A method as recited in claim 20, wherein the gas concentration outputs and tolerance limits for each prediction model are calculated using at least one of a neural network model or another nonlinear model.

23. A method as recited in claim 20, wherein the gas concentration outputs and tolerance limits for each prediction model are calculated using at least one of a Projection to Latent Structures (PLS) model, a neural network model, or other linear and nonlinear models.

24. A method as recited in claim 3, wherein the concentration calculation module calculates the gas concentration outputs and tolerance limits weighted by the real-time quality factor calculation.

25. A method as recited in claim 1, wherein the concentration calculation module determines the most accurate gas concentration outputs using the tolerance limits.

26. A method for converting the responses of an array of metal oxide sensors into constituents and concentrations of a multi-component gas provided around the sensor array, the method comprising:

providing a self-diagnostic module to calculate the quality factors associated with corresponding prediction models and determine if the responses of the sensor array are within a preferred range; and

providing a concentration calculation module to calculate the gas concentration outputs from each prediction model, corresponding tolerance limits, and gas concentration outputs determined by weighting the outputs from the prediction models by corresponding quality factors and determining the weighted tolerance limits, so that the gas concentration outputs with the lowest tolerance limits can be reported as the actual constituents and concentrations of the multi-component gas.

27. A method as recited in claim 26, further comprising a step of providing a self-correction module to compensate for changes in the responses over time so that quality factors can be obtained which are within the preferred limits or to stop the responses and report an error.

28. A method as recited in claim 27, wherein the self-correction module provides a self-referencing process for the sensor array in real-time.

29. A method as recited in claim 28, wherein the self-referencing process comprises increasing the temperature of the sensor array, measuring lambda around the sensor array, and scaling the sensor responses to responses at a predetermined reference state.

30. A method as recited in claim 27, wherein the self-correction module provides a temperature correction process for the sensor array in real-time.

31. A method as recited in claim 30, wherein the temperature correction process comprises removing variations in the sensor responses due to temperature using a fixed relationship between temperature and resistance.

32. A method as recited in claim 30, wherein the temperature correction process comprises removing variations in the sensor responses due to temperature using a recursive algorithm to provide a time-varying relationship between temperature and resistance.

33. A method as recited in claim 27, wherein the self-correction module provides a drift correction process for the sensor array in real-time.

34. A method as recited in claim 33, wherein the drift correction process comprises adjusting the sensor responses for both offset and span, and determining the offset and span correction factors at any given time using real-time sensor responses, sensor temperature, and available auxiliary engine data.

35. A method as recited in claim 27, wherein the self-correction module determines that the self-correction module cannot compensate for the changes in the sensor responses, stops the sensor array from processing data for a predetermined time, reports an error to the sensor array, and begins processing data again after the predetermined time.

36. A method as recited in claim 27, wherein the self-correction module determines that the self-correction module cannot compensate for the changes in the lo sensor responses, stops the sensor array from processing data for a predetermined time, reports and error to the sensor array, and does not begin processing data again if the number of sequential errors is greater than a predetermined number.