A Research On Robust Optimization Model For Fund Allocation Of China UnionPay Network

In recent years, a revolutionary change has happened in the bank services following by the continual development and improvement of computer networks and inter-bank payment and cleaning system. The service network of banks has transferred from physical infrastructures to network payment systems such as ATM and POS. Bankcards and their services have become the important products and central business. The operation conditions of bankcard network play an important role in the management of banks. Thus, banks should allocate proper fund for the bankcard network to maintain its daily operation in the preconditions of realizing the maximal returns and anti-risk ability and satisfy the fund demands of bankcards in the network, which is the urgent problems to be solved by bankers.It has been considered in this dissertation that the fund allocation problem in the environment of China UnionPay network. The dissertation takes the uncertain fund demand of China UnionPay network as a breaking point and carries on a preliminary research on it. According to the decomposition and analysis of fund demand of bankcards in the China UnionPay network, the dissertation makes use of simulation-and-optimization-based hybrid scenario generation algorithm to generate the fund demand scenario matrices of bankcards and establishes the robust optimization model for fund allocation of China UnionPay network. On the foundation of this, the dissertation clusters the scenarios generated from simulation with the help of SPSS 13.0 and obtains a two stage scenario trees of five scenarios. The uncertain fund demands in the network are described by the scenario trees. Moreover, the genetic algorithm and optimization toolboxes of MATLAB 7.1 are used for solving the model. It has been discussed that the influences of different model robust weights on total return of network system, variance of total return of network system and square sum of demand departures in the paper. Furthermore, the anti-jamming features, stability and reliability of the robust optimization model are verified.