摘要
隨著工業科技不斷進步，人們對於安全意識的逐漸高漲，對於結構安全的重視也日趨成長，由於振動將影響結構之穩定性，嚴重者甚至會造成結構破壞，因此研究結構的振動，並設法解決減振的問題，在工業界扮演著越來越重要的角色，也是結構學研究領域中，長久以來的重點目標之一。本論文以影像視覺化的方法分析系統振動問題，並以一橫樑為例，以高速攝影機對高速且微小振動體施以高速攝影記錄，並以電腦深度學習的觀念，輔之以 MatLab 軟體處理此大數據的振動影像資料，進行 "影像-數值"轉換，尋找系統最大振幅，以利工程分析。本論文以雷射位移計量測振動橫樑之振幅，並使用高速攝影機，以光學的角度同時進行量測；在收集龐大的視訊資料之後，對於整個系統的振動情形，以影像視覺化的方式進行分析，並與雷射位移計量測的結果對照，以驗證影像視覺化的正確性。本論文提出的方法可用於一般振動物體之振動監測，在一個難以使用感應器或是非接觸式的量測物體上，吾人可使用此光學法，進行全面得振動監測，輔之以大數據及影像處理，相信可提升工程精密度及成品安全。
並列摘要
With the continuous advancement of industrial technology, people’s increasing awareness of safety and the increasing emphasis on structural safety are also growing. Because vibration will affect the stability of structures, the structure’s vibrations will be studied and solutions will be sought. The problem of vibration reduction plays an increasingly important role in the industrial area and is one of the long-term key objectives in structural studies. This study analyzes the vibration problem of the system by means of image visualization and uses a beam as an example. High-speed cameras are used to apply high-speed photographic recordings to high-speed and micro-vibrations. The concept of computer deep learning is complemented by MatLab software. By using the toolbox software in the MatLab, we transfer the Big data of vibration images to the numerical values (the "image-value" conversion) to find the maximum amplitude of the system for engineering analysis. In this thesis, the amplitude of the vibration beam was measured by laser displacement gauge, and the measurement was performed at the same time using a high speed camera. After the huge video data was collected, the vibration of the entire system could be performed qualitatively. The experimental results from the laser displacement gauge measurements were used to verify the correctness of image visualization analysis. The method proposed in this paper can be used for vibration monitoring of general vibration objects. We can use this optical method to perform comprehensive vibration monitoring on a difficult-to-use sensor or non-contact measurement object, supplemented by big data and images. Processing, we believe it can improve the engineering precision and product safety.