Supply Chain Inventory Collaboration Under Uncertain Demand

Supply chain collaboration (SCC) has been a growing area of interest amongst researchers and practitioners from varied disciplines over the years. SCC enables multiple companies to work together to share their information, knowledge, risks and rewards, and coordinate their activities to optimize the whole performance of supply chain.SCC consists of different functions and collaborative inventory joint decision making is considered to be an important part of it, since a good inventory decision can minimize the over-stock/under-stock costs and then maximize the profits. However, the inventory coordination is a challenging problem because of the uncertainty of customers’demand. Existing researches typically assumed that the demand is either deterministic or stochastic variable that follows a certain probability distribution, but the former cannot reflect the ever changing market and the later lacks the universality for the inventory cost computation.In this paper, we proposed an approach combining fitness inheritance particle swarm optimization (PSO) and Monte Carlo simulation, as well as Adaptive Sampling for inventory collaboration decision making under uncertain demand. First, Monte Carlo simulation is adopted to construct a universal model to mimic the uncertainty of market demand, and used to evaluate a coordination scheme. This evaluation method is able to mimic any type of stochastic demand and is convenient to calculate the total inventory cost of supply chain. Then a Fitness inheritance PSO combined with adaptive sampling algorithm is proposed to find an inventory coordination scheme. Fitness inheritance can largely reduce the number of particles that need fitness evaluation process by Monte Carlo simulation and the adaptive sampling focus on reducing the sampling times during the fitness evaluation process. Adaptive sampling enables the Monte Carlo simulation to sample accordingly to avoid unnecessary calculation. Various fitness inheritance techniques and adaptive sampling techniques are studied and combined with the PSO and Monte Carlo to construct an effective algorithm for inventory coordination. The experimental results show an excellent performance of our approach in reducing the total cost of supply chain and saving the computation of fitness evaluations.