Parameter optimization of single sample virtually expanded method

Abstract EN

Aiming at single sample experiment of sample number n=1, the relationship of parameters virtually expanded from n=1 to n=13 is derived in this paper，and the big-samples data are gained by Bootstrap method.

Instead of existing methods, a developing particle swarm optimization based on Minimax is put forward.

With the application of this method in the parameter optimization, the lower confidence limit approaches the lower confidence limit of the Semiempirical Evaluation Method with more rapid speed and higher precision.

In this way, the most suitable augmented parameters virtually expanded from n=1 to n=13 are gained, which provides a better virtual augment method for the sample augment from n=1 to n=13.