Machine vibration analysis based on experimental modal analysis with radial basis functions

Machinery vibration analysis is important to ensure quality of the machined part and to avoid unexpected downtime before the scheduled maintenance. In this paper, we propose an approach using experimental modal analysis (EMA) based on frequency response functions (FRF) for machine vibration analysis. The proposed method utilizes a set of vibration sensors to gather vibration measurements at selected locations on the machine, from which the relative sensitivities are computed to yield a 3D sensitivity map. Radial basis function (RBF) is adopted as function approximator to model the overall vibration sensitivities over an extended space and frequencies of the machine structure. The model can be used for vibration analysis in real-time to determine effective tool and workpiece location or to determine optimal sensor placement sites in condition monitoring. The strengths of the approach are evaluated against results from finite element analysis (FEA) carried out with the machine model.