IDENTIFICATION OF BEARING DYNAMIC COEFFICIENTS FROM UNBALANCE RESPONSES MEASUREMENTS AND NONLINEAR OPTIMIZATION TECHNIQUES

ALCEMIR MILIAVACCA¹; CARLOS ALBERTO BAVASTRI¹; RODRIGO BUBNIAK SILVERIO¹

¹ UFPR

doi:10.20906/CPS/COB-2015-1959

Resumo

Bearing dynamic features are important parameters that governs the vibration characteristics in rotating machines. However, these coefficients are usually difficult to determine and this fact can produce inaccurate overall results. This work proposes a methodology to determine, simultaneously, the bearing dynamic coefficients and unbalance distribution by structural modifications and non-linear optimization techniques. The method adjusts the unbalance response simulated by a finite element model to measured curves at run-downs procedure. As the objective function is multimodal, a hybrid method including genetic algorithm and a direct Nelder-Mead search was used for global optimization. A base formed by eigenvectors of an invariant primary system (real rotor and the initial constant bearing coefficients) was used to obtain the numerical unbalance response in the optimization environment. This base allows find the response of the real rotor (compound system), when bearing coefficients are changing in the optimization environment, in the equivalent way to structural modification techniques. This methodology allows solve the eigenvalue problem only once, reducing dramatically the computational time cost. A numerical simulation was performed and the results are presented and discussed. An analysis on the GA optimization parameters, as well as the value of the initial bearing parameters in the eigenvalue problem was also performed. The methodology shows to be efficient and reliable to identified, simultaneously, the bearing parameters and unbalance excitations.

Palavras-chave: BEARING; ROTORDYNAMIC; UNBALANCE RESPONSE; OPTIMIZATION; STRUCTURAL MODIFICATION