Abstract:This paper presents the experimental results from the testing of ball-screw inerter with different inertances and different excitation inputs which were carried out using a CNC hydraulic servo exciting test-platform. The experimental results show that the dynamic characteristics of inerter are influenced by the nonlinear factors. In order to master the dynamic performance of inerter, a neural network for mechanical properties prediction was considered to be constructed. To the problem that the BP algorithm is prone to fall into local optimum, the genetic algorithm was used to optimize the training process and improve the generalization ability of the neural network. According to the influencing mechanism of the nonlinear factors on the mechanical properties of inerter, the input variables of neural network were determined to be the inertance and the displacement, velocity and acceleration of inerter under multiple transient time points. 1020 groups of test data were used for network training and prediction and the prediction results are identical to the test results. It is shown that this method can be used as a mechanical properties prediction method for inerter.