Development of Vibration and Stress Estimation Technology for Mechanical Structures by Virtual Sensing

Abstract

In this paper, we propose a virtual sensing technology combined with model order reduction. The virtual sensing is a simulation technology that integrates a numerical model and sensor data with a Kalman filter. In addition, the virtual sensing enables to perform various analysis using the estimated states. However, since the virtual sensing is computed in the time domain, the use of finite element models with large degrees of freedom increases the computational cost. This issue can be solved by the model order reduction which is a technique to reduce the computational cost by reducing the degree of freedom of the finite element model while maintaining the analytical accuracy. The proposed methodology is validated experimentally by which the states are estimated by the virtual sensing using the finite element model reduced by the model order reduction and the actual vibration measured by a small number of sensors when a motor is driven. The virtual sensing estimates 3-axis acceleration accurately. Then, the Mises stress map in the piping is calculated by using estimated vibrational displacement.