Optimization method for the identification of unbalance distribution in non linear rotating machinery
Résumé
Traditionally, the influence coefficients method is used very frequently and successfully in the industrial context. This method is efficient and well adapted when the relationship between the unbalance and the displacements is linear. However, if some nonlinearity appears in the rotating machinery, this technique does not work properly and the results obtained regarding the correction weights and their corresponding angular positions are not satisfactory. The choice of the number and the repartition of the correction plans depends on the possible accessibility that varies for each machine. In a previous work, a model based methodology for the balancing of rotating machinery with localized non linearity was developed and validated experimentally. Based on this approach, a new method dedicated to the identification of the unbalance distribution in linear and non linear rotating machinery was realized through pseudorandom optimization methods and the system modeling was performed by using the well-known finite element method. In this paper, the system studied is composed of a horizontal flexible shaft with two rigid discs. The shaft is supported by bearings located at its ends, as follows: a roller bearing at one end and two ball bearings at the other end. The roller bearing is located in a squirrel cage attached to the framework by three identical flexible steel beams. The Electro-Magnetic Actuator (EMA) located on the external cage constitutes a smart active bearing and provides nonlinearity in the dynamics of the system. The method developed is assessed numerically and validated experimentally