The airborne utility management system is the inevitable developing trend of airborne system. The tasks of the airborne utility management system were partitioned for solving the task assignment problem. The two-level tasks allocation strategy of cycle tasks and un-cycle tasks was put forward because their contributions are difference for the risk coefficient of system. Taking the balance of risk coefficient among the computers as the goal function, the cycle task allocation arithmetic based on the ant colony optimization arithmetic was designed. The ant colony arithmetic was improved by fuzzy-adaptive parameter regulation. The simulative results show that the improved arithmetic can efficiently make it escaped from the local extremum and resolve the UMS task assignment problem of utility management system.