Adaptive, neural-based signal processor
First Claim
Patent Images
1. A neural-like processing method for analyzing acoustic emission signals that emanate from a medium under test, comprising the steps of:
- (a) obtaining acoustic emission signals emanating from a medium under test to which a known source signal has been applied;
(b) encoding said known source signal with a descriptor characteristic of said known source signal, such as its location with respect to the medium, its strength, and its time function;
(c) concatenating said obtained acoustic emission signals to form a series chain of acoustic emission signals defining a pattern vector representative of said medium under test;
(d) appending said known source descriptor to said series chain;
(e) introducing said pattern vector to a storage medium;
(f) repeating steps (a) through (e) a number of times in order to establish a data base characterizing the medium and its response to known source signals; and
(g) synthesizing from said data base in said storage medium, characteristics of an unknown source based upon simulated acoustic emissions, or characteristics of acoustic emissions based upon a simulated unknown source.
1 Assignment
0 Petitions
Accused Products
Abstract
A method and system for analyzing emission signals emanating from a test medium for the purpose of determining characteristics of the test medium. The system and method utilize adaptive neural processing to prognosticate futur
This invention was made in part with government support under grant No. MSM 8405466, awarded by the National Institute of Health; and under grant No. N0014-85-K-0595, awarded by the Office of Naval Research. The government has certain rights in the invention.
84 Citations
34 Claims
-
1. A neural-like processing method for analyzing acoustic emission signals that emanate from a medium under test, comprising the steps of:
-
(a) obtaining acoustic emission signals emanating from a medium under test to which a known source signal has been applied; (b) encoding said known source signal with a descriptor characteristic of said known source signal, such as its location with respect to the medium, its strength, and its time function; (c) concatenating said obtained acoustic emission signals to form a series chain of acoustic emission signals defining a pattern vector representative of said medium under test; (d) appending said known source descriptor to said series chain; (e) introducing said pattern vector to a storage medium; (f) repeating steps (a) through (e) a number of times in order to establish a data base characterizing the medium and its response to known source signals; and (g) synthesizing from said data base in said storage medium, characteristics of an unknown source based upon simulated acoustic emissions, or characteristics of acoustic emissions based upon a simulated unknown source. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
-
-
9. A neural-like processing method for analyzing emission signals that emanate from a medium under test, comprising the tests of:
-
(a) obtaining emission signals emanating from a medium under test to which a known source signal has been applied; (b) encoding said known source signal with a descriptor characteristic of said known source signal; (c) concatenating said obtained emission signals to form a series chain of emission signals defining a pattern vector representative of said medium under test; (d) appending said known source descriptor to said series chain; (e) introducing said pattern vector to a storage medium; (f) repeating steps (a) through (e) a number of times in order to establish a data base characterizing the medium and its response to known source signals; and (g) synthesizing from said data base in said storage medium characteristics of an unknown source based upon simulated emissions, or characteristics of emissions based upon a simulated unknown source. - View Dependent Claims (10, 11, 12, 13, 14, 15, 16)
-
-
17. A neural-like processing method for analyzing emission signals that emanate from a medium under test, comprising the steps of:
-
(a) concatenating obtained emission signals to form a series chain of emission signals defining a pattern vector representative of a medium under test; (b) introducing said pattern vector to a storage medium; and (c) synthesizing from said storage medium, characteristics of an unknown source whose signal is applied to said medium under test based upon simulated emissions, or characteristics of emissions based upon a simulated unknown source whose signal is applied to said medium under test. - View Dependent Claims (18, 19, 20, 21, 22, 23, 24)
-
-
25. A neural-type system for analyzing emission signals from a medium under test, comprising:
-
an array of sensors located about a medium under test, for generating a plurality of emission signals; a signal conditioner connected to said array of sensors for receiving said emission signals and utilizing said emission signals to generate a pattern vector, said generated pattern vector comprising a concatenation of said emission signals and an appended descriptor characterizing a source for stimulating said array of sensors to generate said plurality of emission signals; a neural network module connected to said signal conditioner for weighting said pattern vector, and generating an output comprising an adaptively changed pattern vector; and a data output device connected to said neural network module for receiving and storing an output of said neural network module. - View Dependent Claims (26, 27, 28, 29, 30, 31, 32, 33, 34)
-
Specification