In the maintenance decision-making problem of a multi-state system, it is more in line with the actual situation to consider imperfect maintenance. However, under the influence of imperfect maintenance, the failure mechanism and maintenance process of the multi-state system are more complicated. Further discussion is needed about how to deal with the relationship between the states of various components and the overall state of the system, and how to make the optimal maintenance strategy. To address this issue, this paper proposes a multi-state system maintenance decision optimization method considering imperfect maintenance. Firstly, an imperfect maintenance model is established based on quasi-renewal theory. Secondly, the mapping relationship between the state of the components and the system is decoupled combining universal generating function with the non-homogeneous continuous-time Markov model, the reliability evaluation of the multi-state system is realized. Finally, the maintenance decision optimizing model considering imperfect maintenance is constructed, the optimal maintenance strategy of the multi-state system is obtained by optimizing the number of imperfect maintenance actions to maximize the expected benefit of the whole life cycle.