LOAD IDENTIFICATION FOR A COMPOSITE LAMINATED SHELL USING RADIAL BASE FUNCTION NEURAL NETWORK

Due to the large computation expense of finite element inverse analysis and considering the demerits of the conventional BP neural network such as low convergence speed and local extremum, a new method combining finite element analysis and radial base function neural network is presented to identify concentrated load in a shell structure. The charge outputs of four piezoelectric sensors are calculated by finite element method and used to train the neural network. Furthermore, several samples without training are input into the neural network to predict the location and the magnitude of the load. Finally, an example shows that the present method is effective, feasible and promising.