To improve the repeatability of surface enhanced Raman spectroscopy (SERS) and ensure the homogeneity of the mixed adsorption of surface enhanced metal nanoparticles and analytes, a hardware system for automatic mixing control of surface enhancer and micro liquid samples was developed to ensure the uniform mixing effect of each sample with the surface enhancer based on the self-built laboratory Raman point detection device. At the same time, the real-time control analysis software was compiled with G language to implement one-button operations of automatic mixing and sampling of trace liquid samples and surface enhancers, spectrum acquisition and data processing by the NI LabVIEW software development tool. Finally, the stability of Raman detection system with automatic mixing of surface enhancers and micro liquid samples was verified by testing nitrofurantoin veterinary drug residues in commercially available honey. Compared with Raman point detection system, the Raman spectrum detection system for automatic mixing of surface enhancer and micro liquid sample exhibited a more satisfactory result that the lowest variation coefficients of Raman characteristic peaks for nitrofurantoin and sulfamethoxypyridazine were 0.0322 and 0.0361, respectively. In addition, the linear regression models of nitrofurantoin and sulfamethoxypyridazine in honey were established. The optimal validation set determination coefficients were 0.9619 and 0.9790, and the root mean square errors were 0.6723mg/kg and 0.5186mg/kg, respectively. These findings demonstrated that the system studied could significantly enhance the stability of Raman spectra, and the obtained quantitative prediction models of veterinary drugs were accurate，and it can offer technical support for the rapid quantitative detection of liquid samples by surface enhanced Raman spectroscopy.