Variable-Step Multiscale Fuzzy Dispersion Entropy: A Novel Metric for Signal Analysis

Area of Science:

• Signal Processing
• Entropy Metrics
• Complexity Analysis

Background:

• Fuzzy dispersion entropy (FuzDE) integrates fuzzy entropy (FE) and dispersion entropy (DE) for signal analysis.
• Traditional FuzDE overlooks time-scale information present in signals.
• Existing multiscale methods can be limited by signal length.

Purpose of the Study:

• Introduce variable-step multiscale fuzzy dispersion entropy (VSMFuzDE) to capture rich scale information.
• Overcome limitations of traditional multiscale processing regarding signal length.
• Enhance signal analysis by considering hidden information across time scales.

Main Methods:

• Developed VSMFuzDE by integrating variable-step multiscale processing with FuzDE.
• Applied VSMFuzDE to analyze simulated and real-world signals.
• Compared VSMFuzDE performance against other entropy metrics.

Main Results:

• VSMFuzDE demonstrates increased robustness and sensitivity to dynamic changes in chirp signals.
• Exhibits superior separability for noise signals compared to existing methods.
• Achieved high classification performance, with recognition rates up to 99.2% for gear signals and 100% for ship-radiated noise.

Conclusions:

• VSMFuzDE effectively characterizes abundant scale information, overcoming limitations of traditional methods.
• The proposed VSMFuzDE offers enhanced performance in signal classification tasks.
• VSMFuzDE provides accurate identification capabilities for diverse signal categories.