- 學位論文
汽缸孔內壁缺陷之自動化影像檢測機台研究與開發
The Research and Development of the Automated Images Inspection Machine for the Defects of Cylinder Surface Wall
摘要
自動化光學檢測(Automatic Optical Inspection, AOI)扮演著快速檢驗的關鍵角色，可以取代人工檢測與減少生產成本，達成高速高精度的檢測，進而大幅提升生產的品質與產能。本研究整合機電系統、取像系統與檢測軟體，對汽缸內孔壁表面之缺陷建立出一套自動化光學檢測系統。 本論文所要檢測之產品為某一割草機引擎之汽缸壁，其汽缸壁在加工過程中皆會產生一些不良品。本研究開發之自動化孔內壁缺陷檢測方式是使用網路攝影機以側面斜角窺視對汽缸壁上半層之區域取像及使用攝影機(CCD)搭配側向全景鏡頭深入汽缸壁下半層之區域取像，做機械視覺缺陷檢測，並運用可程式邏輯控制器(PLC)作為檢測機台之自動化模組，NI Vision Builder AI 為影像處理與檢測模組，將其整合，進而達到自動化孔壁面缺陷檢測之目的。最後根據不同缺陷樣貌、環境條件及影像處理方法進行缺陷辨識檢測。 藉由此系統的建立，對於此產品以自動化檢測方式能在實驗驗證階段完成辨識六種缺陷，且平均約12秒的時間檢測完一樣品，達到此產品的生產線需求。
並列摘要
The automated optical inspection plays a key role in rapid inspection, can replace manual inspection and reduce production costs, to achieve high-speed and high-precision inspection, thus greatly improve production quality and capacity. In this study, the integration of mechanical and electrical systems, Inspection systems and image-taking software, the cylinder wall surface of defects to establish a set of automated optical inspection systems. In this paper, the product is to be inspection mower engine cylinder wall, its cylinder wall are during processing will produce some defective products. The research and development of the automated inspection for the defects of cylinder wall surface is to use a webcam is based on the laterally bevel angle method to capture image of cylinder top half wall, and use CCD and borescope probes be controlled by introducing into the cavity of lower half wall to capture image, to do machine vision inspection defect. The use programmable logic controller (PLC) as the machine's automation module and NI Vision Builder AI for the image processing and inspection modules, integrate them, achieve automated images inspection machine purposes. Finally, depending on the method of identification of defects appearance, environmental conditions and image processing. This system for product in an automated inspection method can be completed in the experimental validation phase with more than identify the six kinds of defects, and Average time of about 12 seconds to detect a complete sample, to meet production demand for this product line.