This paper proposes a model prediction method for tracking wear trajectories under lubricated conditions, based on the Archard wear model and the theory of elastohydrodynamic lubrication. Tracking wear in such a system was challenging due to the random nature of the tooth surface condition and the unstable force. The Archard wear model usually ignored the change in wear depth at the pitch point as it was inherently dedicated to the relationship between relative sliding distance and normal force. To overcome these problems, the proposed approach embedded the force analysis model of the micro convex body with time-varying reference into the constrained optimization process of the Archard model. The model was therefore called the Archard wear optimization model. By embedding the time-varying characteristics of the micro convex into the wear model, the wear depth at the pitch point was calculated and the model was optimized. In addition to inheriting the advantages of the Archard wear model, the proposed model ensures accurate tracking of key points, and is experimentally evaluated. Meanwhile, lubricating oil supply coefficient have been introduced to study tooth surface wear under different lubrication conditions.