To counter the problems of Yangtze River ship dangerous goods emergency resources and their own characteristics, taking consideration of the overall relative efficiency of input-output in emergency response resources in the system and making use of data envelopment analysis(DEA) of dangerous goods accident emergency response vessel, this paper evaluates the efficiency of resource allocation, makes the evaluation proceeds with the neural network data for training, and predicts the effect of generating program, so as to obtain the global optimal ship dangerous goods emergency resource allocation optimization program, which can provide a reference for the scientific allocation of dangerous goods emergency resources.