Abstract: To address the limitations of traditional flowing liquid detection methods, such as insufficient real-time performance and poor environmental adaptability, an online detection system for flowing liquids based on visible spectroscopy was developed. By designing experiments covering different colors (red, yellow, blue, purple) and concentrations (high concentration range of 20%–40%, low concentration range of 0.05%–0.6%) the effects and causes of different variables on the detection performance of the online flowing liquid detection system are analyzed. Experimental results showed that the system maintained good repeatability across different colors and concentrations. The yellow sample exhibited the best repeatability, while the blue sample showed the poorest. The repeatability of purple and red samples performed similarly, with repeatability levels between those of yellow and blue. Sample repeatability was relatively poor at boundary concentrations of 20% and 40%, while it performed better in the concentration range of 25%–35%. The lightness (L^*) and saturation (C_ab) of blue and yellow samples exhibited highly linear responses (R²>0.96). The hue angle (h_ab) of blue sample changed rapidly initially and then slowed, while yellow sample exhibited only a small variation in hue angle (<2°). The results demonstrated that the system is capable of achieving online liquid detection in complex flow environments, providing valuable insights for refining liquid color online detection methods in industrial production.