Abstract: When designing feedback coefficients of satellite attitude determination and control system,the precision of the system mathematics model can be significant improved by applying uncertain component model.However,traditional optimization algorithm is insufficient in solving problems with uncertain factors.A robust feedback coefficients optimization approach based on Ant Colony Algorithm (ACA) was proposed.The method was applied to feedback coefficient design for magnetic rods of a momentum bias micro-satellite.Comparing with the traditional Powell algorithms,the new method has clear convergence tendency,which means the algorithm based on ACA has better capability in dealing with random factors,therefore,it can be used in coefficient design for a system considering random factors.