影像相關法應用於結構監測

離岸風力發電機組裝設於嚴苛的海面上。經常受到海風吹襲或海浪的沖刷，這些都是可能使結構受損的條件。一旦因為損壞停機，對於供電產生極大影響；再加上風機結構運維成本十分昂貴，定期監測健康狀態，及早發現結構損傷並及早修復才能降低成本。但風機為動輒數十公尺高的大型結構，需要數以百計的感測器才能完整監測，要想監測完整風場所需的成本相當可觀。考慮到目前影像辨識技術已逐漸成熟，混合式的監測系統也許可以解決成本過高的問題。若想對結構進行拍攝，勢必要使用船或無人駕駛飛行器等等的載具搭配攝影機。此時不論是海浪對船體所造成的波動或是無人駕駛飛行器本身飛行時自身的晃動都會對影像造成影響，這些晃動可能使後續在辨識健康狀態時發生誤差甚至誤判，因此消除影像晃動在監測系統中是必要的。本研究針對固定頻率下的晃動條件進行測試，以相機錄影搭配晃動補償及影像追蹤的演算法對縮尺模型進行監測，除了測試此方法的可行性，也透過改變拍攝距離、焦距等參數，希望能找出相機能有效監測的極限。

關鍵字

離岸風電；結構健康監測；數位影像相關法；影像晃動補償

並列摘要

Offshore wind turbines are constructed on rough seas. Hits from wind and waves all might create structural damage. Once the damage causes the turbine to stop generating power, it will hugely affect the power supply. Moreover, the structural maintenance cost is extremely expensive, the only way to lower the cost is to constantly monitor the structural health and find damages as soon as possible. A turbine is a huge structure and would need hundreds of sensors to monitor completely, the cost to monitor a whole wind farm will be immense. The visual recognition techniques are much more developed, thus a hybrid system of visual and sensors could be the solution to this problem. Because of the vehicles that carry the camera will face environmental loads such as wind and waves, the noise could jeopardize experimental results. Therefore, it is essential to apply video stabilization techniques. This research aims to test the video stabilization algorithms by applying known vibrations on the camera, while analyzing a scaled down model. In addition, this research also changed different parameters like distance and focal distance, in hopes to find the limits of relevant monitoring.

並列關鍵字

Offshore Wind Turbine ； Structural Health Monitoring ； Digital Image Correlation ； Image Stabilization

主題瀏覽