Research On Cash Replenishment Planning And Distribution Joint Optimization For Automatic Teller Machine

Mobile payment and online shopping have impacted the cash business of banks,which greatly increasing the complexity of ATM cash management.Banks want to reduce their cash input on the basis of high cash security ratio.Meanwhile,outsourcing service companies,which is responsible for ATM operations and maintenance,want to minimize their operating costs on the basis of meeting bank goals.On the guidance of the actual industry scenario,this article proposes a solution for the ATM cash replenishment optimization,distribution route optimization,joint optimization of replenishment and distribution,which can reduce operating costs(reflected by the number of replenishment tasks and vehicles,total travel time and distance)and cash costs(reflected by averaged inventory of ATM)and improved the service level of ATMs(reflected by the rates of out-of-cash and full-of-cash occurrence).In the research of ATM cash replenishment optimization,the daily usage forecasting model considers the characteristics of daily cash usage data such as seasonality,holidays,outliers and empty values.Error analysis shows that forecasting errors will not have a significant impact on replenishment plan.The large number of recycling ATMs(both cash withdrawal and cash deposit)has caused serious risk of lout-of-cash and full-of-cash.A special inventory management model that considers the upper and lower safety stock and replenishment number is established.Numerical experiments show that the model can save costs and improve service levels.By setting common key parameters for all ATMs,the trade-offs between banks and outsourcing service companies are achieved.In the research on the optimization of cash transport vehicle,a VRP optimization model that considers task time windows,vehicle time windows,and variable travel time is suitable for the constraints of the cash replenishment scenario.An improved artificial bee colony is established to optimize it.Compared with the traditional artificial bee colony algorithm and simulated annealing algorithm,the improvedartificial bee colony algorithm achieves the characteristics of faster convergence speed and better jumping out of the local optimal solution.Numerical experiments show that the model can optimize the number of vehicles and make full use of transportation resources.By adjusting the weights of different key indicators,the different optimization indicators are balanced.Finally,we innovatively propose to jointly optimize the ATM cash replenishment plan and distribution to further improve the overall efficiency and service level of the cash management service.Aiming at the huge solution complexity of the joint optimization problem,the idea of a two-stage method is used to solve it.Firstly,a machine learning method is used to establish replenishment plan according to similarity of out-of-cash probability matrix and actual geographical distance.The distribution route of each group is calculated by a heuristic algorithm.The idea of this model is more in line with the actual demands of banks and outsourcing service company.Numerical experiments not only reflect the trade-offs of different goals through adjusting parameters,but also establish the methodology of the optimization of ATM cash replenishment planning and distribution under different data sizes and different demand.