基于能量特征的小波概率神经网络损伤识别方法

Damage identification methods of wavelet probabilistic neural network based on energy features

以小波能量特征向量作为概率神经网络(PNN)的输入向量集,提出了小波概率神经网络(WPNN)的损伤识别方法.为了验证该方法的有效性,对钢框架进行了损伤识别研究,并考虑了随机噪声的影响.识别结果表明:WPNN抗噪声能力强,识别精度高,在结构损伤识别与在线检测方面具有潜力.

By combining wavelet energy feature vectors with probabilistic neural network (PNN) in noisy conditions,a new damage identification method called wavelet probabilistic neural network (WPNN) was proposed.Damage identification of a steel frame was utilized to illustrate the effect of this method,and the noise was also considered.The identification result showed that it has high identification accuracy and noise-resistant,being of potential in on-line structural damage detection.