Analysis of the synergistic complementarity between bubble entropy and dispersion entropy in the application of feature extraction

Abstract

Most of the existing studies on the improvement of entropy are based on the theory of single entropy, ignoring the relationship between one entropy and another. Inspired by the synergistic relationship between bubble entropy (BE) and permutation entropy (PE), which has been pointed out by previous authors, this paper aims to explore the relationship between bubble entropy and dispersion entropy. Since dispersion entropy outperforms permutation entropy in many aspects, it provides better stability and enhances the computational efficiency of permutation entropy. We also speculate that there should be potential synergy between dispersion entropy and bubble entropy. Through experiments, we demonstrated the synergistic complementarity between BE and DE and proposed a double feature extraction method based on BE and DE. For the single feature extraction experiment, dispersion entropy and bubble entropy have better recognition performance for sea state signals and bearing signals, respectively; in double feature extraction, the combination of bubble entropy and dispersion entropy makes the recognition rate of sea state signals increase by 10.5% and the recognition rate of bearing signals reach 99.5%.