Interval model validation for rotor support system using Kmeans Bayesian method

Abstract

To identify the uncertain stiffness and damping of the rotor support system, and obtain the most credible interval parameters, an interval model validation for rotor support system using Kmeans Bayesian method is proposed. Interval model updating is transformed into interval lower bound and interval diameter determination. The minimization of discrepancies between predicted and experimental data is converted into minimizing the area between the cumulative distribution function (CDF) of the errors and CDF of 0. To improve the computational efficiency, the long short-term memory (LSTM) network is constructed to denote the mapping relationship between the unbalance response and support parameters. To obtain the most credible interval, the Kmeans clustering algorithm is combined with the Bayesian method to select the interval bounds with maximum credibility. Kmeans clustering algorithm is used to reduce the number of reconstructed errors while utilizing all the information on reconstruction errors. Bayesian method is used to calculate the credibility of the interval bounds. The performance of the proposed method is validated by a gas-generator rotor and a dual-disks rotor. Results show that the proposed method is effective for model validation of the rotor support parameters.