Abstract: The quality of silicon carbide coating is the key factor for the utilization of carbon-carbon
composites as thermal structural materials. In order to optimize and control the deposit process, a neural network(NN) model for SiC-CVD
process was developed. The outputs of NN model were found to be best fit with the sample data’s (<0.35%). With the help of the NN model, a series of predictions about the deposit condition-results were made.
Finally, the predication about the influence of Ar on the deposit rate predicted by ANN model was identical well with the experimental
results, which shows that the model established by ANN can mirror the parameters-results relationships and inner regularity in the
SiC-CVD process.

Key words: carbon-carbon composites, SiC coating, artificial neural network, modeling