Abstract: With the advent of an era of intelligence and information, it is so difficult for current logistics network to meet increasingly various customer demand. For improving the inventory replenishment level of enterprises, the concept of Physical Internet is introduced. Different from traditional logistics network, the logistics network of Physical Internet has characteristics of interconnection, openness and sharing. According to source selection strategy, the suppliers or PI-hubs can be selected by the stockout PI-hubs or retailers dynamically. In order to solve the problem of inventory replenishment in Physical Internet, a hybrid replenishment strategy of Installation-Echelon is proposed. Besides, the dynamic inventory replenishment model is constructed in Physical Internet. The objective function is represented by minimizing inventory replenishment cost and the decision variables are expressed by replenishment quantities and reorder points. Aiming to solving the large-scale dynamic mixed integer nonlinear programming model, several replenishment predetermined plans are obtained by particle swarm optimization algorithm. Then predetermined plans are circularly brought into the simulation environment to optimize. Finally, taking the China Aviation Equipment Import and Export Corporation as an example, the solvability of the model and the feasibility of the algorithm are verified. The results show that source selection strategy in Physical Internet has a significant impact on replenishment cost and replenishment strategy, which provides a reference value for enterprises to formulate inventory replenishment strategy and select source under different conditions.

Key words: Physical Internet; dynamic installation-echelon hybrid replenishment strategy; mixed integer nonlinear programming; particle swarm-simulation optimization algorithm