基于模态应变能比与神经网络的复合材料结构损伤辨识

从结构动力学特性入手，以模态应变能比作为表征结构损伤的标识量，对含损伤的复合材料机翼结构进行损伤辨识仿真，通过神经网络建立起损伤标识量和损伤状态之间的映射模型。仿真结果表明，模态应变能比对结构损伤位置和损伤程度都比较敏感，是一种有效的损伤标识量。神经网络可准确地识别出结构的损伤位置和损伤程度，应用于损伤识别是有效的。

COMPOSITE STRUCTURAL DAMAGE MONITORING BASED ON RATIO OF MODAL STRAIN ENERGY AND NEURAL NETWORK TECHNIQUE

Starts out with the dynamic characteristics of structures and uses modal strain energy ratio as the damage sign parameters (DSP) to supervise the healthy status of composite wing. By use of artificial neural network (ANN) to seek the essential relations of the damage sign parameters and the damage/healthy status. The results of emulation indicate that damage features formed by mode strain energy ratio are feasible to identify the structure damage localization and damage severity. The back propagation(BP) networks are precise enough to identify the damage of structures, so they could be applied in structural health monitoring for composite structures.