Selection of logistics network nodes based on cloud warehousing under uncertain demand

In order to reduce the effects of uncertain logistics demands on the efficiency and stability of urban delivery network, the problem of selecting logistics network nodes based on cloud warehousing was investigated. Realistic factors such as the capacities of cloud warehouses and vehicles, as well as the stock-out guarantee rate, were taken into account, and a mixed-integer nonlinear stochastic optimization model aiming to minimize the total cost of cloud warehouse leasing and distribution was constructed. A three-layer hybrid algorithm model was designed by integrating plant growth simulated algorithm, neighborhood search algorithm, dynamic programming, and genetic algorithm. In the outer framework, the number and location selection of cloud warehouses were optimized based on an adaptive plant growth simulated algorithm. In the middle framework, the service relationships between cloud warehouses and demand points were optimized based on the improved neighborhood search algorithm incorporating the volume-distance-cost accumulation clustering operator. In the inner framework, the delivery vehicle routes were optimized based on the dynamic programming savings algorithm, and the trunk vehicle routes, the rental areas of each cloud warehouse, and the replenishment cycles were optimized based on the genetic algorithm. Research results show that the total cost of the solution obtained by the algorithm designed reduce from a maximum of 3.673 4 million yuan to 3.508 2 million yuan compared with existing algorithms, and the reduction range is between 2.48% and 4.50% compared with different algorithms. When the number of cloud warehouses and demand points is the same, the distance between production and sales areas is negatively correlated with the trunk transportation cost, and positively correlated with the rented area of cloud warehouses and the replenishment cycle. The unit rent of cloud warehouses is positively correlated with the trunk transportation cost, and negatively correlated with the rented area of cloud warehouses and the replenishment cycle. The more concentrated the distribution of cloud warehouses and demand points, the lower the distribution cost of the cloud warehousing system. Under the same distribution, the number of demand points is positively correlated with the number of selected cloud warehouses and the rented area of cloud warehouses. The research conclusions can provide decision-making references for the design of cloud warehouse logistics networks.