Data processing: artificial intelligence – Neural network – Learning task
Reexamination Certificate
1997-12-23
2001-07-03
Davis, George B. (Department: 2122)
Data processing: artificial intelligence
Neural network
Learning task
C706S020000, C706S013000, C706S015000, C706S016000, C706S027000
Reexamination Certificate
active
06256619
ABSTRACT:
TECHNICAL FIELD
The invention relates to the processing of analog signal data for the diagnosis, evaluation and classification of the performance of a machine in which the processing involves a neural network, and in which the processing system parameters undergo an optimization evolution as the data is being processed.
BACKGROUND ART
Vibration Data Neural network signal analysis, with some signal processing modification, has been employed in the art to identify and to diagnose problems in machinery and in products being manufactured.
As an example, in U.S. Pat. No. 5,361,628 diagnostic testing and classification of automobile engines is described, in which, in connection with a neural network, subsampling and filtration for reduction of a vibration signal band is used in order not to overload the neural network.
In general, heretofore in the art problem identification and diagnostic systems involving a neural network are set up and optimized under static conditions using known parameters. There are however, conditions that may evolve in operation, such as changes in the material being processed, changes in one or more of the sensed parameters such as rotation rate, and progressive changes such as the increasing effect of bearing deterioration, that can cause an overall system to perform less than optimally with an attendant reduction in accuracy and responsiveness.
DISCLOSURE OF THE INVENTION
A machine performance analog data processing system involving a neural network is provided in which there is a self optimization capability that varies the signal processing factors in the system of the invention in response to a detected contrast in the classification patterns produced by the system.
The system responds to contrasting conditions and provides feedback type guidance in varying such processing factors as sampling rate, frame length, signal transformation, neural network vigilance and architecture, each in a direction that will maximize the overall performance.
The system of the invention can be applied to machine performance signal classification tasks associated with the analysis and control of mechanical systems, such as vibration data processing and sound field processing.
REFERENCES:
patent: 5058180 (1991-10-01), Khan
patent: 5249259 (1993-09-01), Harvey
patent: 5546505 (1996-08-01), Austvold et al.
patent: 5761381 (1998-06-01), Arci et al.
patent: 5781698 (1998-07-01), Teller et al.
patent: 5854993 (1998-12-01), Grichnik
Chapter 1, “A Gentle Introduction to Genetic Algorithms,” David E. Goldberg, Reprinted w/corrections Jan., 1989.
Caterpillar Inc.
Davis George B.
Kibby Steven G.
Riddles Alvin J.
LandOfFree
Self optimizing neural network analog data processing system does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Self optimizing neural network analog data processing system, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Self optimizing neural network analog data processing system will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2471591