Natural Frequencies Identification by FEM Applied to a 2-DOF Planar Robot and Its Validation Using MUSIC Algorithm

Area of Science:

• Robotics
• Mechanical Engineering
• Signal Processing

Background:

• Accurate identification of natural frequencies (NFs) is crucial for robot design and control.
• Traditional methods may struggle with high-resolution identification, especially in noisy environments.

Purpose of the Study:

• To present and validate a novel methodology for identifying the natural frequencies (NFs) of a two degrees-of-freedom (2-DOF) planar robot.
• To compare the effectiveness of the Multiple Signal Classification (MUSIC) algorithm against the fast Fourier transform (FFT) method.

Main Methods:

• Finite Element Method (FEM) analysis using ANSYS™ software for robot modeling.
• Experimental validation using two distinct 2-DOF planar robots with varying materials and actuators.
• Vibration signal analysis employing the Multiple Signal Classification (MUSIC) algorithm.

Main Results:

• The proposed methodology successfully identifies NFs with high resolution, even for closely spaced frequencies.
• The MUSIC algorithm demonstrates superior performance compared to FFT, particularly for signals with high noise levels.
• High-frequency resolution was achieved with a short data set, showcasing efficiency.

Conclusions:

• The MUSIC algorithm-based methodology provides accurate and high-resolution natural frequency identification for planar robots.
• This approach is robust in noisy conditions and efficient in terms of data requirements.
• Accurate NF identification aids in path planning and controller gain selection to prevent resonance and ensure stability.