采用声发射法对复合材料飞轮试件高速旋转时的损伤信号进行采集。借助体视显微镜、光学显微镜和着色法等辅助检测方法，使损伤类型与损伤信号一一对应。用经过训练的人工神经网络对飞轮试件的损伤模式进行了有效识别。试验还研究了飞轮试件的最终失效原因是由于环向裂纹的贯穿造成的。

The damage signal of high speed composite flywheels was collected by acoustic emission method. The relation between damage type and damage signal was discovered by supplementary detection methods such as stereology microscopy, optical microscopy and coloring penetrating testing. The damage type of fly wheels was distinguished by the trained artificial neural network. Research on the failure cause of high speed composite flywheels showed that the failure was caused by the through circular crack.