【摘要】 为了提高分布式光纤传感器监测结构动态响应的精度并提高损伤诊断的准确性，运用长短期记忆神经网络，建立布里渊增益谱-频率数据关系模型，确定布里渊光时域分析技术中振幅传输模式扫描的固定频率。基于分布式光纤振动传感信息沿光纤密集分布特点，通过计算应变传递率梯度，提出了基于分布式动应变传递率梯度的损伤指标。不同损伤程度的悬臂梁分布式光纤振动传感实验表明，该方法能够实现结构损伤位置的诊断，且对毫米级裂纹仍有较强的敏感性。更多还原

【Abstract】 To enhance the accuracy of distributed optical fiber sensors in monitoring the dynamic response of structures and elevate the precision of damage diagnosis, a model depicting the relationship between Brillouin gain spectrum(BGS) and frequency has been developed using the long short-term memory network(LSTM). This model facilitates the identification of the fixed frequency required for scanning in Amplitude TransferBrillouin Optical Time Domain Analysis(AT-BOTDA). Based on the dense distribution characteristics of distributed optical fiber vibration sensing information along the optical fiber, a damage index based on the distributed dynamic strain transmissibility gradient is proposed by calculating the strain transmissibility gradient. Distributed fiber optic vibration sensing experiments conducted on cantilever beams with varying degrees of damage show that this method can accurately diagnose the location of structural damage and remains highly sensitive to millimeter-scale cracks.更多还原