Estimating Each Element of Joint Stiffness Matrix Using Neural Network-Estimation from the Rotation Vector at End of Jointed Part

The new method, which can quickly and accurately estimate the stiffness of the jointed part of the automobile body structure by using neural network, is proposed. The non-linear relation between the rotation vectors of the end section of the jointed part and the stiffness of the jointed part is constructed by using neural network. Once the relation of both is constructed, the values of joint stiffness can be estimated by only inputting the some elements of rotation vectors into the trained neural network. And the ability of this method was verified by using some examples.