<p>Corrosion fatigue of tubing is an important safety factor to be considered in the process of acidizing fracturing in the oilfield. In corrosive media, the significant dispersion of fatigue life makes the prediction results less practical. The reason lies in the uncertainty of crack initiation and propagation affected by corrosion conditions, which makes the error of life prediction model difficult to be effectively controlled. In this study, titanium alloy was selected as the research object to carry out corrosion fatigue test under the low concentration hydrochloric acid corrosion environment commonly used in petroleum engineering. An innovative corrosion fatigue life prediction method with life prediction bias feedback evaluation was proposed by building a support vector machine prediction learning model, so as to effectively predict the corrosion fatigue life of titanium alloys in different concentrations of corrosive media. The results of model prediction reveal the effect of hydrochloric acid concentration on corrosion fatigue and reduce the dispersion of corrosion fatigue life prediction in practical engineering. Compared with Basquin model, its prediction results are more conservative, improve the safety of fatigue structures in corrosive media, and have important engineering application value.</p>