Research On The Early-Warning In The Supply-Chain Based On Ratiocinative Model

With a large number of information systems being deployed in supply chain, supply chain risk management has been transforming from past passive mode to nowadays active mode. The passive mode aims at reducing the lost causing by risk. However the passive mode intends to forecast and prevent risk. Almost all the relevant researches are in the management area. Discusses focus on the definition of the supply chain risk; the elements producing supply chain risks; the classification of these elements or the content of the supply chain risk management. While these researches can only service for strategy level, they can't support the operation level actions in the real supply chain early-warning application. They hardly mention how to implement in the information system.At the beginning, we scan the content of supply chain and supply chain risk in the management field. Then we analyze the algorithms based on the traits of the supply chain application. Firstly raise an open and multi-algorithms adaptive method. Through our research to achieve such goal: In condition that there is no change of target and plan in supply chain, use minimal cost in operation level to effectively prevent supply chain risk and continuously improve supply chain management, in order to strengthen competitive capability.In this thesis, we choose decision tree, ANN(Artificial Neural Networks) and Bayes method as samples to profoundly analyze the classify algorithms'adaptive character in the supply chain early-warning application. Based on the analysis, we raise a layered and divided-conquer method. At the macro-layer, we settle the question how to calculate the risk corresponding to the topography of supply chain. At the micro-layer, we resolve the problem how to adopt different ratiocination models to calculate the risks of various supply chain nodes. The cooperation of two layers is also taken into account. We prove such method is feasible by simple realization. At the end of thesis, we will give a conclusion and an expectation.