基于声发射和神经网络的风机叶片裂纹识别研究

摘要

提出一种对风力机叶片裂纹声发射信号进行模式识别的方法.该方法以叶片无裂纹、萌生裂纹、扩展裂纹和断裂四个阶段为声发射源的四个模式,基于声发射信号含有丰富的发射源信息的特点,通过大量采样获得叶片裂纹声发射信号参数,并依照叶片裂纹声发射参数分析的数值特点确定BP神经网络,用选定的网络对叶片裂纹阶段进行模式识别,以判断裂纹的危害程度.仿真结果表明,利用BP神经网络可以对声发射信号进行有效识别,识别准确率达到90％以上.%It presents a method for pattern recognition of fan blade,which takes four phases of no crack, initial crack,expansion crack and fracture initiation as four modes of acoustic emission source. Based on the characteristics of the acoustic emission source with rich information,acoustic emission signal parameters have been obtained by sampling largely.Then BP neural network is determined according to numerical value characteristics for analyzing acoustic emission parameters,which is applied to recognize pattern in blade cracking phase in order to determine the degree of damage of the crack.Simulation results show that the neural networks can recognize acoustic emission signal effectively with an accuracy rate achieving 90 percent more.