Application of Adaptive Learning Network Modeling to Single-Particle Erosion Test Data.

摘要

The methodology of Adaptive Learning Network (ALN) Training has been used to model mass loss ratios resulting from single-particle impacts on carbon-carbon composite materials. The resulting ALN models identify material bulk density, graphitization temperature, dynamic hardness (A), and the stiffness exponent (N) as the parameters that most affect the erosion resistance of Series 300 and Series 400 materials. The development of more detailed models was made difficult by the limited variety in the experimental data base. Keywords: Erosion resistance; Carbon-carbon composites. (edc)