This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ve a maximum of one. Finally, the average polynomial coefficient matrix is found from the normalized waviness variation and the average autospectrum is found from the normalized waviness and roughness variation. 4. HYBRID COVARIANCE MODEL EVALUATION To evaluate the modeling capability of the hybrid covariance model, two comparisons are shown; first, with simulated surface variation data and second, with measured surface data. 328 M. R. Tonks, K. W. Chase and C. C. Smith 4.1. Simulated Data A set of surfaces were simulated as if being manufactured using a process that gives surface variation of all three types and for which descriptors of typical variation have been identified. 10,000 surfaces were generated and the covariance was calculated directly from the simulated data. The covariance was also calculated using the hybrid method according to the known variation descriptors. Figure 4 shows four plots comparing the simulated covariance to the covariance generated with the hybrid model, a 3D bar graph of the geometric covariance matrix generated by each m...
View Full Document
- Spring '10
- The Land