Modified Hierarchical Time-Delayed Distribution Entropy: an Effective Method for Signal Complexity Measurement

Guancen Lin,Aijing Lin,Yujia Mi
DOI: https://doi.org/10.1007/s11071-024-10297-4
IF: 5.741
2024-01-01
Nonlinear Dynamics
Abstract:Complex systems in the real world are characterized by multiple interactions among various components and often exhibit complexity, nonlinearity, or chaotic behavior. Signal complexity constitutes a pivotal attribute of nonlinear complex systems. Entropy-based methods for analyzing complexity offer a promising avenue for elucidating the dynamic behaviors inherent in such intricate systems. Diverging from multiscale approaches that primarily target low-frequency signal information, this study introduces Modified Hierarchical Time-delayed Distribution Entropy (MHTDE), a method that amalgamates the benefits of hierarchical analysis, time delay, and Distribution Entropy. This amalgamation facilitates the capture of complexity information spanning both high and low frequencies. Results from simulation experiments demonstrate the precision of MHTDE in quantifying signal complexity, showcasing its robustness to parameter variations and data length alterations. Furthermore, we employ MHTDE to extract complexity features from four real-life datasets and employ the K-Nearest Neighbor algorithm for classification. The classification performance of MHTDE outperforming baseline models. Consequently, the proposed MHTDE method underscores its competence in efficacious feature extraction.
What problem does this paper attempt to address?