Automated Harmonic Signal Removal Technique Using Stochastic Subspace-Based Image Feature Extraction.

Area of Science:

• Structural Dynamics and Signal Processing
• Vibration Analysis and System Identification

Background:

• Stochastic subspace-based algorithms (SSI) are effective for modal parameter estimation but struggle with harmonic excitations from rotating machinery.
• Automating SSI for harmonic removal is challenging due to the need for user-defined parameters, hindering practical application.

Purpose of the Study:

• To develop an automated harmonic removal technique for system identification that requires no user-defined parameters.
• To enhance the accuracy and consistency of modal parameter estimation in the presence of harmonic disturbances.

Main Methods:

• Utilizes image-based feature extraction for clustering and classification of harmonic components and structural poles from stabilization diagrams.
• Applies the developed algorithm to numerical simulations and experimental data featuring harmonic excitation.

Main Results:

• The proposed automated harmonic removal method accurately identifies and discards harmonic influences from output signals.
• Modal parameter estimation demonstrated high accuracy and consistency both before and after harmonic component removal.

Conclusions:

• The developed image-based approach offers a robust and automated solution for harmonic removal in system identification.
• This technique significantly improves the reliability of modal analysis for structures subjected to harmonic excitations.