Abstract: The stockpiling state of containers is usually not in the ideal order of the containers to be sent, so relocation of the containers is inevitable. In order to reduce the relocation frequency in the container yards, improve operational efficiency, on the condition that the order of container withdrawal is required, the paper regards every possible state that may be generated when a container is being withdrawn as a node, and takes the number of relocations plus a base as the connection weight of the node. Then, the optimization problem for container relocation is converted into the shortest-path solution problem. The paper brings forward a new approach to find the shortest path based on pulse-coupled neural network, which can be used to solve large-scale practical problem. We can easily work out the shortest path from a source to all other target points, therefore we have the best relocation program. The algorithm lays a solid theoretical basis for the container intelligent control system.