混凝是水處理中固液分離的重要程序之一，而在混凝過程中存在著許多因子影響混凝的好壞，其中混凝劑的加藥量是最難控制的一個，目前是由瓶杯實驗及操作經驗來決定加藥量，常有過量加藥的問題，造成藥劑量及後續汙泥處理成本的提高。目前雖有對於控制加藥量的設備，但問題在其接觸性的偵測造成訊號上的干擾，而非接觸式的偵測儀器成本太高，低成本的非接觸式偵測系統實在有發展的必要。 本研究建立一套光學監測的系統來觀察並配合影像分析來分析混凝中的膠羽，對光學監測所得之影像分析其訊號變化，並將訊號與混凝結果做比較。研究中利用高嶺土配製人工濁度水進行混凝，以網路攝影機拍攝數位影像並分析其光訊號的變化。 實驗結果說明當濃度上升時，偵測的訊號亦跟著上升，而在一定濃度下訊號會隨著顆粒愈大而降低。在混凝過程中膠羽顆粒的變化反應在訊號上，而與混凝的結果比較下可發現，良好的混凝其訊號值會出現明顯且快速的下降趨勢接著在一固定值上下震動，而失敗的混凝則無此趨勢僅上下跳動。另外由偵測值及混凝結果可發現，要達到良好的混凝，混凝藥劑量的添加小於傳統瓶杯所決定的最佳加藥量。在碎形維度的結果分析方面，由於碎形相似皆為1.9故無法看出其差別。 Coagulation process is one of the important procedures of solid/liquid separation in water/wastewater treatment. There are many factors affects the success in coagulation among them, the dosage of coagulant is most difficult to control. The dosage of coagulant is usually determined by jar test or the experience of operators, which always leads to an overdose of coagulant. But overdosing causes high cost of coagulant, also high cost on sludge disposing. So far, existing equipments either high interference or high cost. There is necessary to develop a technique of low interference and low cost. In this research, to set up an optical image monitoring technique for coagulation, operate with image analysis observation and analysis the flocs in coagulation. Turbid water is made by kaolin particles dilute with tap water. Analyze the image which is captured by web camera, and compare analyzed data with the result of coagulation. The results shows that the RGB values increased as the concentration of particles increased. Also, the RGB values decreased as the particles size increased when the mass concentration of particles remained the same. Therefore, as particles grew during coagulation, the RGB values changes correspondingly. Compare with the RGB values and coagulation, in successful coagulation there is a clear decreased trend of the RGB values and then jumped up and down about a value. But there is no trend in failed coagulation. Besides, from the RGB values and the situation of coagulation, the dosage of coagulant is less the dosage decided by jar test. On the results of fractal dimension, the flocs look similar and the fractal dimension of flocs were about 1.9.
