Browse Prior Art Database

Reducing Data from Physical Sensors

IP.com Disclosure Number: IPCOM000076316D
Original Publication Date: 1972-Feb-01
Included in the Prior Art Database: 2005-Feb-24
Document File: 3 page(s) / 22K

Publishing Venue

IBM

Related People

Herd, HH: AUTHOR

Abstract

A method is provided for reducing measured data taken by physical sensors to represent coefficient values, and for testing each of the coefficients for statistical significance relative to a specified objective. The method comprises a combination of mathematical techniques related to known physical laws and permits the testing of each underlying determinist function in order to determine the statistical significance of each coefficient so that it may be retained or discarded.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 3

Reducing Data from Physical Sensors

A method is provided for reducing measured data taken by physical sensors to represent coefficient values, and for testing each of the coefficients for statistical significance relative to a specified objective. The method comprises a combination of mathematical techniques related to known physical laws and permits the testing of each underlying determinist function in order to determine the statistical significance of each coefficient so that it may be retained or discarded.

The method also incorporates a procedure for determining a "cut-off" point above which the terms are primarily stochastic in content, and below which the terms are the best attainable representation of the presumed deterministic component which can be achieved from the measured data.

A value "MAX" is calculated in a manner fully described in U.S. patent 3,626,384 issued December 7, 1971 to J.B. Davis et al and assigned to the International Business Machines Corporation. The value "MAX" also called "Cut" is the point at which the higher coefficients are eliminated, in order to provide a number of coefficients (the coefficients of order lower than MAX) which represent the dependent variable with minimum error.

The coefficients of higher order than MAX are used to calculate V(1) which is an estimate of the variance based on the degrees of freedom N-1-k(max).

As this procedure alone will identify the highest order term which is significant, it can be used to establish the cutoff term (k(MAX) in the referenced patent) without use of the procedure described in the referenced patent, and in addition, if there are any lower order coefficients which have a lower probability than is acceptable to the experimenter, they can be identified and deleted. This is the additional power of this procedure.

The procedures...