With　the　development　of　urban　rail　transit　systems,　Communication-Based　Train　Control　(CBTC)　system　choose　to　deploy　Ad　Hoc　network　alongside　the　track　for　train-to-trackside　communication.　However,　due　to　the　mobility　of　the　train,　how　to　efficiently　send　information　to　the　train　by　Ad　Hoc　network　still　remains　a　challenge.　Considering　the　dynamic　characteristics　of　the　train　and　multiple　optimization　objectives,　we　propose　a　mobility-aware　multi-objective　Deep　Deterministic　Policy　Gradient　(DDPG)　algorithm　for　routing　to　optimize　delay,　throughput　and　energy　consumption.　We　first　set　up　the　clustering　routing　model　according　to　the　dynamic　routing　scenario.　To　solve　the　problem　of　route　selection,　Markov　decision　process　(MDP)　models　are　constructed　for　intra-cluster　optimization　and　inter-cluster　optimization　respectively,　and　train　operating　conditions　are　considered　in　inter-cluster　MDP.　Then　we　propose　a　multi-objective　DDPG　routing　algorithm　to　get　the　optimal　routing,　where　delay,　throughput　and　energy　consumption　are　designed　as　a　three-dimensional　vector.　Simulation　results　indicate　that　our　scheme　optimizes　multiple　objectives　in　a　balanced　manner,　and　shows　better　performance　compared　with　other　schemes.　©　2023　IEEE.