In order to relieve the wear of brake shoe and guarantee the safety of PMSMdriven rail vehicles with demagnetization fault, a novel healthoriented optimal control strategy is presented through similarity based modeling (SBM). The proposed methodology estimates the flux of PMSM and generates the Fv profile of pneumatic brake with the similarity operation between historical data and online monitoring data. The feasibility and benefit of the developed scheme is validated through a simulated experiment of the Syntegra PMSM rail vehicle. Simulation results show that accurate alarm monitoring and damage prognosis of demagnetization and realtime control optimization can be reached, which can effectively relieve the brake shoe wear.