[Objective] Extreme disasters can cause multiple physical failures in distribution networks as well as damage to road infrastructure. Road disruptions impede the mobility of repair crews，thereby affecting the efficiency of distribution network restoration. To improve the fault recovery efficiency of distribution networks post extreme disasters，this paper proposes an optimization strategy that couples road repair progress into multi-fault restoration planning for distribution networks. [Methods] A coupled model of the transportation network and distribution network is established，together with a disaster-impact model for the transportation network. A time-varying load model is then developed based on the spatiotemporal demand characteristics of three user categories. Building on these models，a coordinated restoration optimization framework is formulated that jointly schedules distribution-network repairs and road-repair activities，with the dual objectives of minimizing power loss and shortening overall restoration time. To solve this multi-objective optimization model，a parameter-adaptive Non-dominated Sorting Genetic Algorithm II （NSGA-II） algorithm enhanced by an inflection point strategy is employed. Simulation studies are conducted on a regional distribution line and a transportation network system with 17 nodes. [Conclusions] The results show that the proposed strategy dynamically allocates emergency repair resources based on the evolving restoration status of the coupled network，prevents repair delays，enables coordinated optimization of transportation-network recovery and distribution network restoration. The method effectively reduces the distribution-network repair time and outage load. [Conclusions] The proposed strategy effectively integrates road emergency repairs in the transportation network with fault restoration in the distribution network，enhances system resilience against extreme disasters，and offers valuable reference for distribution network emergency repair.