An optimization-based identification study of cylindrical floating ring journal bearing system in automotive turbochargers

This paper presents an optimization-based methodology to predict the bearing coefficients from the measured dynamic responses in a rotor supported on floating ring bearings. The speed-dependent unknown direct and cross-coupled bearing stiffness and damping coefficients of the ideal rotor system are identified by predicting the deviations in frequency response amplitudes of ideal and actual rotor systems. The objective function is defined in terms of the total error in radial displacement amplitudes at the two bearing nodes during each operating speed. The modified particle swarm optimization (MPSO) scheme is employed to minimize the objective function subject to the variable limits. The approach also considers an equivalent spring-damper bearing system for simulations along with experimental work. The methodology is illustrated with three case studies namely (1) both the actual and ideal systems considered linear, (2) both actual and ideal systems are nonlinear, and (3) parameter estimation from experimental data. It is observed that the reliable stiffness and damping parameters are predicted with relatively less computational time. Furthermore, the accuracy of predicted parameters is verified by adding the varied amounts of Gaussian white noise signals to the measured responses and identified parameters are found to be within 3% accuracy. Further, the first four critical speeds of the rotor system are identified as 2200 rpm, 2424 rpm, 6120 rpm, and 6330 rpm respectively. The approach is relatively simple and can be applied to any kind of bearings.