Local Entropy Optimization-Adaptive Demodulation Reassignment Transform for Advanced Analysis of Non-Stationary...

Area of Science:

• Signal Processing
• Mechanical Engineering
• Vibration Analysis

Background:

• Analyzing complex, non-stationary mechanical vibration signals is challenging due to multiple instantaneous frequencies and closely spaced frequency ridges.
• Existing time-frequency analysis methods often struggle with signals exhibiting non-proportionality and cross-frequency interference.

Purpose of the Study:

• To introduce a novel time-frequency analysis method, the Local Entropy Optimization-Adaptive Demodulation Reassignment Transform (LEOADRT).
• To enhance the analysis of complex non-stationary mechanical vibration signals with improved time-frequency resolution.

Main Methods:

• LEOADRT employs a demodulation term to convert signal components into stationary signals, optimizing parameters using Rényi entropy's local optimal theory.
• Energy redistribution is performed using a maximum local energy criterion to refine time-frequency map ridges.
• The method is designed for signals with multiple instantaneous frequencies and closely spaced frequency intervals.

Main Results:

• LEOADRT demonstrates superior performance compared to existing methods like SBCT, EMCT, VSLCT, and GLCT.
• The algorithm effectively processes complex non-stationary signals, including those with non-proportionality and closely spaced frequencies.
• Significant improvements in time-frequency resolution were achieved.

Conclusions:

• LEOADRT offers a powerful tool for the real-time analysis of multi-component and cross-frequency mechanical vibration signals.
• The method provides strong support for mechanical fault diagnosis, condition monitoring, and predictive maintenance.
• This advancement is particularly valuable for complex industrial applications requiring high-resolution signal analysis.