【摘要】 概率神经网络 (下称PNN)以贝叶斯概率方法描述测量数据 ,因而PNN可以在有噪声情况下进行结构损伤检测·提出了运用传统PNN和自适应PNN进行结构损伤检测的方法与基本原理 ,并分别用两种PNN模型进行了悬索桥的损伤定位研究 ,还讨论了测量噪声对识别精度(IA)的影响·研究发现 ,运用自适应PNN进行损伤定位效果极大地优于传统PNN ,且随着噪声程度的增大 ,IA减少更多还原

【Abstract】 The methods and principles of a traditional probabilistic neural network (PNN) and an adaptive PNN model were given respectively. The damage localization for suspension bridge was studied using the traditional and adaptive PNN, and the effect of noise level on identification accuracy (IA) was discussed. The IA for damage localization by using the adaptive PNN is better than that by using the traditional PNN. Furthmore, the IA values decrease as the noise level increse.更多还原