統計的異常診断手法による学習不要構造ヘルスモニタリング(モニタリング,OS.6 折り紙・針金/スマートストラクチャ)

概要

論文の詳細を見る
This research is about a statistical diagnostic method for a structural health monitoring which is applicable to existing structures from the present moment. In this method, structural condition is statistically diagnosed by change of a response surface which is regression model of sensor's outputs. Change of response surface is statistically tested with F-test. In the F-test, a threshold value of normal/troubled condition is simply decided from a theoretical F-probability distribution. Therefore, this diagnostic method only requires data of intact condition and does not require the complicated modeling and information of troubled condition. To confirm the effectiveness of the presented method, health monitoring of jet fan is conducted. Load cells are mounted to a supporting system of the jet fan. Response surface is calculated from vibration data measured by the load cells. As a result, structural condition of the jet fan is successfully diagnosed with the method.
2004-03-03