As distributed power sources connected to the distribution network, the magnitude of the fault current and the direction of the power flow have been changed, and the relay protection setting of the distribution network has become more difficult. Aiming at the optimization problem of inverse time overcurrent protection setting for the distribution network containing DG, the uncertain factors of distribution network failure, the inherent properties of relays and the four requirements of relay protection is considered. The concept of "global historical average optimal solution" and dynamic inertia weight is introduced in the update process of particle swarm optimization. Finally a fixed value optimization method for inverse time overcurrent protection in the distribution network containing DG based on improved particle swarm optimization is proposed. The conclusion shows that the improved particle swarm algorithm can effectively prevent the fixed value solution from falling into the local optimal dilemma, and it is suitable for the fixed value optimization problem of the inverse time overcurrent protection in the distribution network containing DG.