基于正则化频率变化率与神经网络的石油井架结构损伤识别

阐述结构健康监测中损伤识别的关键问题,介绍基于正则化的频率变化率的损伤定位方法,提出一套基于模态参数和神经网络的损伤识别技术。按相似理论建立了JJ160/41-K型井架实物模型,根据其基本结构和受力特点,以JJ160/41-K井架的实物模型为对象进行损伤研究。结果表明:井架模型在局部结构发生变化时整体的固有频率发生改变;将模态分析理论和神经网络方法相结合,可对已定位损伤的井架进行定量分析;通过实例运算验证了其对结构损伤诊断的有效性。

Damage Identification of Oil Derrick Based on Normalized Change Ratio of Frequency and Neural Network

Several key issues related to damage identification of structural health monitoring are expounded. A method for damage positioning, which is on the basis of the normalized change ratio of frequency, is developed. Then, a set of damage identification techniques based on modal parameter and neural network is put forward. A physical model for JJ160/41-K type oil derrick is built up with similarity theory. According to the basic structure and force characteristic of JJ160/41-K type oil derrick, damage analysis is made on the physical modal of JJ160/41-K derrick. Results show that the holistic inherent frequency of derrick will synchronously change with the change of local structure of derrick model, and using mode analysis theory combined with neural network method can make a qualitative analysis on known-damage-location derrick. Finally, the validity of this method to diagnose structure damage is verified through an actual example.