Abstract: There are complex dependencies among multi-component series systems, so developing the optimal maintenance strategy to maintain the minimum maintenance cost is one of the keys to equipment scientific management. The method considers economic dependence and structural dependence among components, and develops a finite time system maintenance optimization model to maximize the saving of total maintenance cost. The method uses the genetic algorithm to optimize the components of the system and solve the non-deterministic polynomial (NP) hard problem. Finally, the method applies the rolling plan, combining the longterm plan with the short-term plan to realize the dynamic real-time update of the maintenance plan. According to the case, the method is effective and feasible, and worth learning in engineering practice.