Vibration load identification based on the neural network model

摘要

The paper utilizes the neural network method to identify the loads of the WZ12–1 offshore platform. Firstly, the finite element model of the WZ12–1 platform is build using the software of ANSYS. By applying loads to the identification points of the finite element model, we can obtain the data for training a neural network model. Then, a three-layer BP network is established and trained until converges using the data obtained previously. Finally, we input the actual response data into the trained network and obtain the corresponding loads on the WZ12–1 platform. The results show that the neural network method could gain a great advantage over the traditional techniques in the load identification of offshore platforms which are large and complex.