In this paper, the influence law of vibration characteristics of different positions on surface roughness is studied. And a surface roughness prediction model based on vibration parameters is established. The experiment was carried on in a TB9 titanium alloy system with cemented carbide tools. The cutting vibration at different positions was detected by 8704B25 and 3333A2 acceleration sensors. The vibration acceleration signals were processed and analyzed by Matlab. TR2000 high-precision surface roughness meter was used to measure the surface roughness of the machined parts. The experiment results show that there is a significant linear relationship between the vibration characteristics of the turning system at different positions and the surface roughness; Pearson correlation coefficients between root mean square value of tool vibration acceleration, root mean square value of spindle vibration acceleration and root mean square value of rear guide vibration acceleration and surface roughness are 0.379 93, 0.331 90 and 0.181 95, respectively; and the average percentage error of surface roughness prediction model is less than 3%. It is concludes that the turning vibration of the tool, the spindle and the rear guide all have a certain effect on the surface roughness in turning titanium alloy wire; the order of the influence of the vibration characteristics at different positions of the turning system on the surface roughness is cutter > spindle > rear guide. It can be seen that the closer the vibration to the cutting position, the greater the impact on the surface roughness of the turning process; the surface roughness prediction model based on vibration parameters has a higher accuracy and can be used as a surface roughness prediction model.