Based on the probability model of enemy aircraft senor on research submarine, an improved ant colony algorithm (ACA) is presented. The algorithm makes sure that the path has the ability of keeping away from the unsafe area far away by the coefficient of pheromone volatile and the index of evading risk . The coefficient matrix of direction to target W is introduced to save running time. Finally, the analysis result of the example indicates that using , , are reasonable and useful. Especially, the right using of , W is helpful and can improve the control ability as well as practicability.