Ajuste de modelo de máquinas rotativas utilizando otimização por enxame de partículas (PSO)

Abstract

This study addresses the numerical model updating of rotating machinery using a Hybrid Particle Swarm Optimization (PSO-H) algorithm, validated through experimental bench tests. The research focuses on calibrating critical parameters such as bearing stiffness and damping coefficients to minimize discrepancies between simulated and experimental responses. A one-dimensional Finite Element Model (FEM) was developed, incorporating gyroscopic effects, unbalance forces, and rotor-bearing interactions based on Lalanne and Ferraris' formulations (1998). Experimental data were acquired through impact tests with Frequency Response Function (FRF) measurements. The PSO-H algorithm demonstrated efficient error minimization between models and experiments, achieving stable convergence within 20 iterations. Results showed average errors below 1% for natural frequencies and Cross-Modal Amplitude Correlation (CMAC) values above 0.99 for the first two vibration modes. The methodology proves robust for Structural Health Monitoring (SHM) applications, with potential extensions to nonlinear systems.