NOx emission-control system using a virtual sensor

• US 20030216855A1
• Filed: 05/15/2002
• Published: 11/20/2003
• Est. Priority Date: 05/15/2002
• Status: Active Grant

• Current Assignee: Caterpillar Incorporated
• Original Assignee: Caterpillar Incorporated
• Inventors: Liang, Cho Y., Jacobson, Evan E., Srinivasan, Syamaia
• Granted Patent: US 6,882,929 B2
• Time in Patent Office: Days
• US Class Current: 701/114
• CPC Class Codes: F01N 11/00 Monitoring or diagnostic devices for exhaust-gas treatment apparatus , e.g. for catalytic activity safety, indicating or supervising devices for internal combustion engines F02B77/08; testing of machines G01M13/00
• CPC Class Codes: F01N 13/0097 the purifying devices are arranged in a single housing
• CPC Class Codes: F01N 2560/026 for measuring or detecting NOx
• CPC Class Codes: F01N 2610/02 the substance being ammonia or urea
• CPC Class Codes: F01N 2900/14 said parameters being related to the exhaust gas
• CPC Class Codes: F01N 2900/1402 Exhaust gas composition
• CPC Class Codes: F01N 3/208 Control of selective catalytic reduction [SCR], e.g. dosing of reducing agent
• CPC Class Codes: F01N 9/005 using models instead of sensors to determine operating characteristics of exhaust systems, e.g. calculating catalyst temperature instead of measuring it directly
• CPC Class Codes: F02D 2041/1423 Identification of model or controller parameters
• CPC Class Codes: F02D 2041/1433 using a model or simulation of the system
• CPC Class Codes: F02D 41/0235 in relation with the state of the exhaust gas treating apparatus control of exhaust gas treating apparatus per se F01N
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: F02D 41/1462 with determination means using an estimation
• CPC Class Codes: G05B 13/042 in which a parameter or coefficient is automatically adjusted to optimise the performance
• CPC Class Codes: Y02C 20/10 of nitrous oxide (N2O)
• CPC Class Codes: Y02T 10/12 Improving ICE efficiencies
• CPC Class Codes: Y02T 10/40 Engine management systems

Abstract

A method and system may be provided to perform a process for controlling NO_x emissions of an target engine. In one embodiment of the invention, the process may include determining predicted NO_x values based on a model reflecting a predetermined relationship between control parameters and NO_x emissions, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed associated with the engine. Further, the process may include adjusting the model based on a determination of whether the predicted NO_x values meet a predetermined criteria associated with actual NO_x values. The adjusted model may be stored in a memory associated with the engine whereby NO_x emissions exhausted from the engine may be reduced based on virtual NO_x emission values determined from the adjusted model.

Claims

1. A method for controlling NO_x emissions of an engine, comprising:

determining predicted NO_x values based on a model reflecting a predetermined relationship between control parameters and NO_x emissions, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed associated with the engine;

adjusting the model based on a determination of whether the predicted NO_x values meet a predetermined criteria associated with actual NO_x values;

storing the adjusted model in a memory associated with the engine; and

reducing NO_x emissions exhausted from the engine based on virtual NO_x emission values determined from the adjusted model.

2. The method of claim 1, wherein reducing NO_x emissions includes:

operating the engine; and

determining the virtual NO_x emission values based on values for each of the control parameters collected during the operation of the engine and the stored adjusted model.

3. The method of claim 1, wherein the actual NO_x values are obtained from a physical NO_x sensor that measures actual NO_x emission values based on measured values of the control parameters obtained during a previous operation of the engine.

4. The method of claim 1, wherein reducing NO_x emissions includes:

minimizing NO_x emissions exhausted from the engine based on the virtual NO_x x emission values.

5. The method of claim 1, wherein reducing NO_x emissions includes:

adjusting an operating parameter of the engine.

6. The method of claim 1, where reducing NO_x emissions includes:

adjusting a urea solution dosing rate applied to the engine exhaust.

7. The method of claim 1, wherein the model is a neural network, the control parameters being provided as inputs to the neural network and the predicted NO_x emission values being produced as outputs of the neural network.

8. The method of claim 7, wherein adjusting the model includes:

adjusting mathematical weights associated with nodes included in the neural network until the predicted NO_x values meet the predetermined criteria.

9. The method of claim 1, wherein the predetermined criteria is associated with a value reflecting an acceptable difference between the predicted NO_x values and the actual NO_x values.

10. The method of claim 1, wherein determining virtual NO_x emission values includes:

applying the values for each of the control parameters collected during the operation of the engine to the stored adjusted model to produce the virtual NO_x emission values.

11. The method of claim 1, wherein prior to determining predicted NO_x values based on a model, the method includes:

collecting test values of each of the control parameters from each of a plurality of sample engines during respective operations;

collecting actual test NO_x emission values from each of the sample engines based on the collected test values of the control parameters; and

creating the model based on a relationship between the actual test NO_x emission values and the test values for the control parameters.

12. A method for controlling NO_x emissions of an engine that includes a physical NO_x sensor that measures actual NO_x emission values produced by the engine during operation and a controller including a model that produces predicted values of NO_x emissions based on collected values of control parameters corresponding to the engine during the operation, the method comprising:

collecting actual NO_x emission values from the physical NO_x sensor and corresponding values for each of the control parameters during operations of the engine;

applying the collected control parameter values to the model to produce predicted NO_x emission values;

training the model during the operations based on an analysis of the predicted NO_x emission values and the actual collected NO_x emission values;

determining whether the physical NO_x sensor has failed;

setting the trained model based on a training that occurred immediately prior to the physical NO_x sensor failure; and

controlling NO_x emissions exhausted from the engine during subsequent operations based on predicted NO_x emission values produced from the trained model in response to collected values associated with each of the control parameters during the subsequent operations.

13. The method of claim 12, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed associated with the engine.

14. The method of claim 12, wherein the model is a neural network that receives the control parameters as inputs and generates the predicted NO_x emission values as an output.

15. The method of claim 12, wherein setting the trained model includes:

storing the trained model in a non-volatile memory device within the controller.

16. A method for adjusting the performance of an emission-control system associated with a target engine, wherein the emission-control system includes a controller, the method comprising:

creating a model reflecting a relationship between NO_x emission values produced by at least one test engine and control parameters collected during test operations of the at least one test engine;

adjusting the model to a first trained state;

storing the adjusted model in the controller;

determining characteristics associated with the target engine;

readjusting the model stored in the controller to a second trained state based on the determined characteristics; and

controlling NO_x emissions exhausted from the target engine during operations based on predicted NO_x emission values produced by the readjusted model.

17. The method of claim 16, wherein adjusting the model to a first trained state includes:

applying test values of each control parameter to the model to produce predicted NO_x emission values; and

adjusting the model to the first trained state based on a determination whether the predicted NO_x values meet a predetermined criteria associated with actual NO_x values obtained from a physical NO_x sensor that measures NO_x emission values exhausted from the target engine.

18. The method of claim 16, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed.

19. The method of claim 16, wherein the characteristics associated with the target engine include at least one of a specific operating criteria and a desired work application for the target engine.

20. The method of claim 16, wherein readjusting the model includes:

readjusting the model stored in the controller based on a comparison of NO_x emission values produced by the model and actual NO_x emission values collected from previous operations of a plurality of field engines that each have the same determined characteristics as the target engine.

21. The method of claim 20, wherein each field engine is included in a host system and the determined characteristics of the field engines include at least a work environment that each host system was previously operated.

22. The method of claim 19, wherein the characteristics of the field engines include performance specifications associated with each field engine.

23. The method of claim 16, wherein readjusting the model includes:

collecting historical operating data associated with one or more engines that have the same determined characteristics as the target engine, wherein the historical operating data includes measured values for each of the control parameters and corresponding measured NO_x emission values associated with each of the one or more engines during previous operations;

applying the measured values of the control parameters to the stored model to produce predicted NO_x emission values; and

readjusting the stored model based on an analysis associated with the predicted NO_x emission values and the measured NO_x emission values associated with the one or more engines.

24. A system for controlling NO_x emissions of an engine, the system comprising:

a controller configured to;

receive measured values of each of a plurality of control parameters during operations of the engine, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed,

predict NO_x emission values of the engine based on values associated with each of the plurality of control parameters, and

produce control signals as a function of the predicted NO_x emission values; and

a system coupled with the controller to receive the control signals and configured to control NO_x emissions exhausted from the engine as a function of the control signals.

25. The system of claim 24, wherein the controller includes a memory including program code configured to, when executed by a processor, predict NO_x emission values of the engine as a function of each of the plurality of control parameters.

26. The system of claim 24, wherein the controller includes an interface coupled with the engine and operable to receive the measured values of each of the plurality of control parameters during operations of the engine.

27. The system of claim 26, wherein the interface is coupled with the engine through one or more sensors that collect the measured values of each of the plurality of control parameters during operations of the engine.

28. The system of claim 24, wherein the controller includes a processor coupled with the system configured to execute program code to produce the predicted NO_x emission values and the control signals as a function of the predicted NO_x emission values.

29. The system of claim 25, wherein the program code represents a neural network that uses the control parameters as inputs and generates the predicted NO_x emission values as outputs when executed by the processor.

30. The system of claim 24, wherein the system configured to control NO_x emissions is a urea-based selective catalytic reduction system.

31. The system of claim 24, wherein the system configured to reduce NO_x emissions includes an engine control unit coupled with the controller to receive the control signals and configured to adjust one or more operating parameters of the engine.

32. A system for controlling NO_x emissions of an engine, the system comprising:

a physical NO_x sensor that measures actual NO_x emission values produced by the engine during operation; and

a controller including software that performs emission control processes, when executed by a processor, the emission control processes including;

a process for providing predicted values of NO_x emissions based on collected values of control parameters corresponding to the engine during the operation;

a process for collecting actual NO_x emission values from the physical NO_x sensor and corresponding values for each of the control parameters during the operation of the engine;

a process for training the process for providing predicted NO_x emission values during the operation based on an analysis of the predicted NO_x emission values and the actual NO_x emission values;

a process for determining whether the physical NO_x sensor has failed; and

a process for controlling NO_x emissions exhausted from the engine during subsequent operation based on predicted NO_x emission values produced from the trained process for predicting in response to collected values associated with each of the control parameters during the subsequent operation.

33. An apparatus for controlling NO_x emissions from an engine, comprising:

a controller including;

a memory including program code configured to predict NO_x emission values of the engine based on values associated with each of the plurality of control parameters, wherein the program code represents a neural network that uses the control parameters as inputs and generates the predicted NO_x emission values as outputs,

an interface coupled with the engine and configured to receive measured values of each of a plurality of control parameters during operations of the engine, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed, and

a processor coupled with the memory and the interface and configured to execute the program code to produce predicted NO_x emission values as a function of the received measured values and to produce control signals as a function of the predicted NO_x emission values; and

a system coupled with the processor to receive the control signals and configured to control NO_x emissions exhausted from the engine as a function of the control signals.

34. An apparatus for controlling NO_x emissions from an engine, comprising:

an engine;

an exhaust system coupled with the engine to process emissions produced by the engine during operation;

a physical NO_x sensor coupled with the exhaust system and configured to measure actual NO_x emission values produced by the engine during operation; and

a controller coupled with the physical NO_x sensor and the engine and configured to;

collect actual NO_x emission values from the physical NO_x sensor and corresponding values for each of a plurality of engine control parameters from the engine during operation, wherein the control parameters include ambient humidity, manifold pressure, manifold temperature, fuel rate, and engine speed associated with the engine during operation;

apply the collected control parameter values to a neural network to produce predicted NO_x emission values;

train the neural network during operation of the engine based on an analysis of the predicted NO_x emission values and the actual collected NO_x emission values;

determine whether the physical NO_x sensor has failed;

set the trained neural network based on a training that occurred immediately prior to the physical NO_x sensor failure; and

provide control signals to the engine, wherein the control signals control NO_x emissions exhausted from the engine during subsequent operations based on predicted NO_x emission values produced from the trained neural network in response to collected values associated with each of the control parameters during the subsequent operations.