Assistant Professor, University of Maryland, College Park, MD
(Received 7 September 1990; accepted 4 September 1991)
An efficient algorithm to characterize experimental stress-strain data is preacated. The algorithm segments the curve into regions by locating regions with a maximum correlation coefficient. Thus the algorithm determines the initial region of the curve to be considered in determining the initial slope or, in the case of a stress versus strain curve, the initial modulus. The algorithm is of particular use in analyzing the elastic behavior of composites where nonlinear stress-strain, knees, damage, and other abnormalities may occur. The results of the algorithm are compared to linear regions determined from calculations of correlation coefficients involving all possible regions within a data file. In general, the algorithm provides the same linear regions as are determiend from the complete calulation of correlation coefficients and is at least one order of magnitude faster.
Paper ID: JTE11910J