考虑不确定性因素的结构损伤检测方法

Methods of Structural Damage Detection with regared to Uncertain Factors

结构损伤检测是当前国内外学术与工程界研究比较活跃的领域 ,如何将不确定性的因素与确定性的损伤检测方法相结合 ,并应用于实际的复杂工程结构中是当前亟需解决的课题 .在贝叶斯概率处理不确定性信息的基础上 ,提出了运用概率神经网络 (PNN)进行复杂结构的损伤检测方法 ,并分别用传统PNN和自适应PNN对悬索桥的桥面板进行了损伤检测 ,以验证该方法的有效性 .研究表明 ,运用PNN进行损伤检测是可行、有效的 ,自适应PNN极大地优于传统PNN ,且随着噪声程度的增大 。

Structural damage detection is currently a very active area of both academic and engineering application.It is an urgent task to integrate the uncertain factors with certain damage detection methods and apply it to the real complex structures.In this paper,based on the Bayesian probability dealing with uncertain meassage,the probabilistic neural network(PNN)method for complex structures damage detection is presented.The traditional PNN and adaptive PNN are respectively utilized to detect the deck of a suspension bridge to validate the presented method.It shows that effective and the adaptive PNN is superior to the traditional PNN.Moreover,the superiority of adaptive PNN is obvious with the increase of the noise level.