考虑传感器性能退化的结构损伤检测

发布时间：2018-04-22 07:26

本文选题：传感器性能退化 + 高斯分布　；参考：《福州大学》2013年硕士论文

【摘要】：结构健康监测主要通过检测结构的响应及变化来进行健康状态的评价,目前结构损伤检测方法仅考虑环境、荷载对损伤的影响,而没有考虑传感器由于长时间工作与环境的变化导致其性能退化,这严重地影响了损伤诊断率,也增加了检测的难度。因此,在实测的响应中考虑环境、传感器性能退化引起的结构响应变化并从中区分出来,以提高损伤检测精度。针对上述问题,本文展开了系列研究并取得了以下成果及结论：(1)研究提出了传感器性能退化及结构损伤在传感器性能退化检测控制图中的反应模式。通过对传感器异常检测及性能退化分离算法开发,验证结构损伤及传感器性能退化均会引起其响应异常。基于结构损伤对结构响应影响具有全局性,而传感器自身性能退化对结构响应影响具有局部性,可根据隔离传感器前后控制图模式的改变与否判别异常来源,即控制图模式发生变化说明异常来源于传感器性能退化,若控制图模式不发生变化,则异常来源于结构损伤。而且,经结合数值试验与实验室模型试验研究发现,由传感器性能退化引起的控制图模式为非平稳模式,由结构损伤引起的控制图模式为平稳模式。(2)提出改进的传感器异常检测及性能退化分离算法。改进的传感器性能退化检测控制图指标对于传感器性能退化较为敏感,对结构损伤较不敏感,即传感器性能退化会引起控制图中控制指标超限,而结构损伤则不会。由此,可根据改进的传感器性能退检测控制图中控制指标是否超限判断传感器异常来源。该算法简化了传感器异常来源判断方法及步骤,而且具有较强说服力。(3)提出了一种新型的损伤检测策略：基于模态参数与MCPSO(多粒子群协同优化算法)的两阶段损伤检测方法。首先基于模态应变能对损伤位置进行初步定位,经验证损伤定位有效、可行,而且为第二阶段损伤程度优化奠定良好基础；其次,利用阈值过滤一些损伤较小单元(视为无损伤单元),对初步判定损伤的单元进行编码,利用MCPSO通过粒子寻优及反复迭代精确定位单元并识别其损伤程度,经验证损伤识别精度较高且效率高。
[Abstract]:The structural health monitoring mainly evaluates the health state by detecting the response and changes of the structure. At present, the damage detection method only considers the environment and the effect of load on the damage. The sensor's performance degradation is not considered because of the long working hours and the change of the environment, which seriously affects the damage diagnosis rate and increases the difficulty of detection. Therefore, considering the environment in the measured response, the structural response caused by the degradation of the performance of the sensor is changed and distinguished to improve the accuracy of damage detection. In order to solve the above problems, a series of studies have been carried out and the following results and conclusions have been obtained. The response mode of sensor performance degradation and structural damage in the sensor performance degradation detection control chart has been proposed. Through the development of sensor anomaly detection and performance degradation separation algorithm, it is verified that both structural damage and sensor performance degradation will cause abnormal response. Based on the fact that the structural damage has a global effect on the structural response, and the performance degradation of the sensor has a local effect on the structural response, the abnormal source can be identified according to the change of the control pattern before and after isolating the sensor. The change of control chart mode indicates that the anomaly originates from the degradation of sensor performance, and if the control chart mode does not change, the anomaly comes from structural damage. Furthermore, by combining the numerical test with the laboratory model test, it is found that the control chart mode caused by the performance degradation of the sensor is a non-stationary mode. An improved algorithm for sensor anomaly detection and performance degradation separation is proposed. The improved performance degradation detection control chart index is sensitive to sensor performance degradation, but not sensitive to structural damage, that is, the degradation of sensor performance will cause the control index to exceed the limit in the control chart, but the structure damage will not. Therefore, the abnormal source of the sensor can be judged according to the control index in the improved sensor performance regression control chart. The algorithm simplifies the method and steps of sensor anomaly source judgment, and has a strong persuasion. It proposes a new damage detection strategy: a two-stage damage detection method based on modal parameters and MCPSO (Multi-particle swarm optimization algorithm). First, the damage location is preliminarily located based on modal strain energy, which is proved to be effective and feasible, and it lays a good foundation for the second stage damage degree optimization. The threshold value is used to filter some small damage elements (considered as non-damage elements), which encode the primary damage elements, and MCPSO is used to find out the damage degree by particle optimization and iterative accurate location of the unit. It is proved that the damage identification accuracy is high and the efficiency is high.
【学位授予单位】：福州大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TU317

【相似文献】