應用機器視覺於機械工件焊道品質之檢測：「農機修護技能競賽」零件製作項目之輔助評分

本研究應用機器視覺建立一套農機修護技能競賽零件製作評分之輔助系統，可以輔助裁判做出快速且正確的評分。本系統可以針對加工元件進行編號辨識，同時對焊道品質客觀分析，並將量測結果進行比對評分後輸入至評分表中。利用機器視覺針對25組「九十學年度台灣區中等職業學校農科技藝競賽農機修護職類」參賽選手之加工元件部分作品。應用本系統對加工元件進行量測與評分試驗，同時以人工方式進行焊道品質分析。評分項目包括：(1)焊接中斷次數、(2)焊道蝕孔數、與(3)平整度等，(1)、(2)兩項每項佔十分，第(3)項佔五分。由實驗結果顯示，編號辨識與焊接中斷次數之正確率為100%，焊道平整度準確率為92.8%，焊道蝕孔數平均誤差為1.92個，轉換為配分，平均誤差達51.2%，造成誤差原因在於取像死角，致使無法擷取到位在圓柱鐵柱面的蝕孔影像。每組樣本完成評分約需15 sec，人工評分約需40 sec，機器評分速度約為人工評分的2.7倍。本系統具有攜帶方便及組合容易等優點，可攜帶至比賽現場進行加工成品之評分。依實驗樣本而言，本系統約可取代25%人為主觀評分之工作。

關鍵字

機器視覺；加工工件；評分系統

並列摘要

A quality inspection system with machine vision for machine parts had been developed in this study. This system can help judge to evaluate machined parts correctly and fast. The system is composed of calibrations for image gray value and space, recognition of digits, positioning procedure for parts, and inspecting weld quality. The tasks of inspection include recognizing digits, and evaluating weld quality. The results were saved in the table. Twenty-five sets of machined parts were tested. The data were compared to the ones obtained manually. The results showed that all the digits were recognized correctly and the numbers of discontinuity were counted without any error. For the rest results, in the worst case, the averaged error percentage for counting the number of cavities of welds was 51.2%. The erroneous counts were about 1.92. This was because the CCD camera was not capable of grabbing the image for the cavities on the surface of cylindrical part. To process one sample with the system took about 15 sec compared to 40 sec manually. It was about two times fast. The system is easy to carry and to assemble. In some situation when time is urgent such as the grading task in the part-machining contest, the system can be used. The system can replace about 25% of manual task.

並列關鍵字

Machine Vision ； Machine part ； Grade system

國際替代計量

應用機器視覺於機械工件焊道品質之檢測：「農機修護技能競賽」零件製作項目之輔助評分

全文下載

主題瀏覽