Discriminant Analysis of SES XR-1D Broach Data

摘要

Data from test runs of the Surface Effects Ship XR-1D test craft were collected and sorted into four performance classes with respect to the broaching (air ingestion) phenomenon. These classes included broach, no-broach and two intermediate conditions. Each data point was characterized by a twenty parameter measurement vector and the resulting 20-dimensional, four class data base was analyzed by the On Line Pattern Analysis and Recognition System (OLPARS) to determine the degree of separability of broach from no-broach as well as the measurements most meaningful in achieving this separability. The results show that broach is separable from no-broach with low error and that no measurement parameter other than draft had any clearly greater effect on results than any other parameter. (Author)