A method for damage detection applicable to large slender steel structures such as towers of large-scale wind turbines, long-span bridges, and high-rise buildings is presented. This method is based on continuous strain data obtained by distributed fiber optic sensor (FOS) and neural network (NN) analysis. An analytical model for cracked beam based on an energy balance approach was used to train a NN. The continuous static strains and the natural frequencies obtained from the distributed FOSs were used as the input to the trained NN to estimate the crack depths and locations. An experimental study was carried out on a cracked cantilever beam to verify the present method for damage identification. The cracks were inflicted on the beam, and static and free vibration tests were performed for the intact case and the damage cases. The distributed FOSs were used to measure the continuous strains. The damage estimation was carried out for the 5 damage cases using the NN technique. It has been found that the identified crack depths and locations agree reasonably well with the inflicted cracks on the structure.

中文翻译：

---

利用分布式光纤传感器静，动态应变数据的大型结构损伤检测方法

提出了一种适用于大型细长钢结构的损伤检测方法，例如大型风力涡轮机的塔架，大跨度桥梁和高层建筑物。该方法基于通过分布式光纤传感器（FOS）和神经网络（NN）分析获得的连续应变数据。基于能量平衡方法的裂纹梁解析模型用于训练NN。从分布式FOS获得的连续静态应变和固有频率被用作经过训练的NN的输入，以估计裂纹的深度和位置。在破裂的悬臂梁上进行了实验研究，以验证本发明的损伤识别方法。对梁施加裂纹，并对完好无损的情况和损坏的情况进行了静态和自由振动测试。分布式FOS用于测量连续应变。使用NN技术对5个破坏案例进行了破坏估算。已经发现，确定的裂纹深度和位置与在结构上造成的裂纹相当吻合。