Data processing: vehicles – navigation – and relative location – Vehicle control – guidance – operation – or indication
Reexamination Certificate
1997-05-07
2001-05-22
Louis-Jacques, Jacques H. (Department: 3661)
Data processing: vehicles, navigation, and relative location
Vehicle control, guidance, operation, or indication
C701S102000, C701S106000, C701S029000, C706S023000, C706S025000
Reexamination Certificate
active
06236908
ABSTRACT:
TECHNICAL FIELD
The present invention relates to virtual vehicle sensors which use neural networks trained using a simulation model to monitor a vehicle parameter.
BACKGROUND ART
Modern engines utilize an electronic engine control module (ECM) to continuously monitor and control engine operation to optimize fuel economy, emissions control, and performance. The ECM uses various physical sensors to collect information reflecting current operating conditions. The information is used to generate output signals for various actuators which control operation of the engine. Using the actuators, the ECM controls the air-fuel ratio, fuel injection, ignition timing, and various other functions to control operation of the engine. Optimal control of the engine over a wide range of engine operating conditions (and ambient conditions) depends on the availability, accuracy, and reliability of data gathered by the engine sensors.
An ideal engine control system would be capable of directly measuring each engine operating parameter which affects any control variable. However, any realizable design is subject to considerations such as the cost, durability, repairability, and/or technological feasibility (including packaging considerations) of appropriate sensors. The deployment of more and more physical sensors results in per-unit cost penalties in development and manufacturing. Replacement and repair costs also rise due to the increased number of sensors and difficulty in diagnosing sensor malfunctions. As such, actual systems typically involve design compromises to accommodate technological difficulties and reduce the cost and complexity of the physical system employed to monitor and control the engine. It is therefore desirable to improve the availability, accuracy, and reliability of data used to effect engine control without significantly impacting the cost, complexity, or repairability of the vehicle.
SUMMARY OF THE INVENTION
A general object of the present invention is to use one or more neural networks within the ECM which act as virtual sensing devices to replace or enhance traditional physical engine sensors. The neural networks are trained using data produced by a simulation model calibrated with actual engine test data. Use of the simulation model reduces the development time required while providing insight into the effect of various design parameters on engine operation.
In carrying out the above object and other objects, features, and advantages of the present invention, a method for controlling a vehicle component using a plurality of physical sensors for sensing first operating parameters and a controller in communication with the plurality of physical sensors includes monitoring signals generated by the plurality of physical sensors to determine values for the first operating parameters and processing the values for the first operating parameters using a neural network embedded in the controller to determine a value for a second operating parameter. The value for the second operating parameter is based on a linear combination of the plurality of values for the first operating parameters such that the neural network functions as a sensor for the second operating parameter. The method also includes controlling the vehicle component based on the value of the second operating parameter.
The neural network is trained using data generated by a simulation model. The simulation model is calibrated using test data gathered during operation of the vehicle component. The trained neural network is embedded in the controller to function as a virtual sensor to sense the second operating parameter to provide improved control of the vehicle component.
Numerous advantages are associated with the present invention. For example, the present invention allows sensing of operating parameters which are currently difficult or cost-prohibitive to measure directly. The present invention utilizes simulation models to generate more comprehensive data representing more operating conditions than would be economically feasible using traditional testing and mapping. The comprehensive data results in more accurate training of the neural networks thereby leading to a more accurate sensor. The sensor may be used to provide monitoring of fundamental physical quantities characterizing operation of the vehicle components which are otherwise unavailable using physical sensors. The present invention is applicable to a wide variety of control systems although particularly suited for control of vehicle engines.
The above advantages, and other advantages, objects, and features of the present invention, will be readily apparent from the following detailed description of the best mode for carrying out the invention when taken in connection with the accompanying drawings.
REFERENCES:
patent: 5130936 (1992-07-01), Sheppard et al.
patent: 5212765 (1993-05-01), Skeirik
patent: 5274714 (1993-12-01), Hutcheson et al.
patent: 5313407 (1994-05-01), Tiernan et al.
patent: 5361213 (1994-11-01), Fujieda et al.
patent: 5386373 (1995-01-01), Keeler et al.
patent: 5539638 (1996-07-01), Keeler et al.
patent: 5548528 (1996-08-01), Keeler et al.
patent: 5559285 (1996-09-01), Bryant et al.
patent: 5583964 (1996-12-01), Wang
patent: 5625750 (1997-04-01), Puskorius et al.
patent: 5745653 (1998-04-01), Jesion et al.
patent: 5781700 (1998-07-01), Puskorius et al.
“Mapping Engines with Analytical Models and Neural Networks”, by Jie Cheng et al, Ford Research Laboratory, Dearborn, MI, pp. 864-867.
SAE Technical Paper No. 790290, “Comparison of Model Calculations and Experimental Measurements of the Bulk Cylinder Flow Processes in a Motored PROCO Engine”, G.C. Davis et al, Feb. 26-Mar. 2, 1979, pp. 1-33.
“The Effect of Inlet Velocity Distribution and Magnitude on In-Cylinder Turbulence Intensity and Burn Rate—Model Versus Experiment”, by C.G. Davis et al, Journal of Engineering for Gas Turbines and Power, Jul. 1988, vol. 110, pp. 509-514.
SAE Technical Paper No. 890679, “The Effects of Load Control with Port Throttling at Idle—Measurements and Analyses”, By C.E. Newman et al, Feb. 27-Mar. 3, 1989, pp. 1-13.
SAE Technical Paper No. 922165, “Monte Carlo Simulation of Cycle By Cycle Variability”, by Diana D. Brehob et al, Oct. 19-22, 1992, pp. 1-13.
SAE Technical Paper No. 932751, “The Effect of Valve Overlap on Idle Operation: Comparison of Model and Experiment”, by H.A. Cikanek, Oct. 18-21, 1993, pp. 1-9.
Cheng Jie
Davis George Carver
LaCrosse Stephanie Mary
Newman, Jr. Charles Edward
Tascillo Anya Lynn
Drouillard Jerome R.
Ford Global Technologies Inc.
Louis-Jacques Jacques H.
LandOfFree
Virtual vehicle sensors based on neural networks trained... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Virtual vehicle sensors based on neural networks trained..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Virtual vehicle sensors based on neural networks trained... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2519767