The tasks of disaster relief campaign within military operations other than war comprise emergency roadway repair and relief distribution; however, there exists interrelationship between the two kinds of tasks, and some other challenges, such as fuzzy stochastic demand and variable network structure, are also confronted in scheduling of this kind of campaign. According to the above situation, a bi-level hierarchy optimization model for emergency resource scheduling in a network with non-deterministic attributes was proposed, and a hybrid intelligent heuristic algorithm which combines fuzzy random simulation with improved genetic algorithm was also designed. At last, a data case was presented to highlight the significance of the proposed model as well as the efficacy of the designed algorithm.