抄録
This paper proposes a method to determine model parameters of a numerical vibration model for railway vehicle carbodies by updating the model parameters using genetic algorithm (GA) together with measurement results. To analyze the three-dimensional elastic vibrations of recent light-weighted and structurally-simplified railway vehicle carbodies, a box-type analytical model was proposed. In this model, the carbodies are modeled as a box-type structure consisting of flexible plates and beams. These elements are connected each other by means of artificial springs. The model parameters such as the mass and stiffness properties of the beams and plates, and spring coefficients of the artificial springs must be specified to carry out a numerical calculation by the box-type model. The mass properties are relatively easy to determine, however, the choice of the stiffness parameters have not been well established, and the trial and error method is often adopted. Therefore, a reasonable and efficient methodology is strongly required. In this paper, an approach is proposed to determine the model parameters automatically using GA. In this method, the frequency response functions (FRFs) at the several points on the carbody are used to evaluate fitness between calculation and measurement results. By considering the FRFs at the points on the roof together with those for the floor, appropriate model data can be obtained both for FRFs and modal characteristics. Two different cases, a Shinkansen and commuter type vehicle, are employed as examples to evaluate the method. Good agreements between measurement and calculation results are observed for both cases and the validity of the proposed method is confirmed.
