Research And Implementation Of Early-warning And Decision-support Of Supply Chain By Using System Dynamics

With the rapid development of information technology and the widespread use of Internet technology, supply chain management solutions have increasingly become a popular topic in both industrial and academic field. Effective management of a supply chain requires a high degree of supply chain visualization. By monitoring the supply chain activities in real-time, both the unexpected variations which have been exposed and those that have not been revealed because of the delay phenomenon among supply chain nodes are required to be detected and early-warned. The existing researches mainly focus in the variations exposed at operational level and take interceptive decisions to remedy the exceptions. However, the existing technology is relatively simple in detecting the unexpected variations at an early stage and taking short-term tactical preventive decisions. This paper focuses on the research and development of supply chain early-warning and decision-support technology at tactical level, detecting the unwanted situations and potential risks in advance, which brings the possibility of taking preventive decisions to mitigate the variations.According to the current status in supply chain management, this paper proposes a methodology used for supply chain early-warning and decision-support at tactical level by combing system dynamics and neural networks. By monitoring the dynamic trends instead of static values of supply chain performance indicators, early-warnings for potential risks can be reported. By fine-tuning the supply chain tactical strategy in short-term, the unwanted impact caused by uncertainty can be mitigated. This paper presents the supply chain monitoring model and adopts system dynamics simulation model for supply chain modeling. The supply chain early-warning technology presented by this paper reports early-warnings based on the prediction of the dynamic trends of monitored indicators. Initially, a supply chain model is built using system dynamics. Then in order to meet the real-time monitoring requirement and integration need of enterprise applications, a neural network that is equivalent to the system dynamics model is built and acts as the kernel of monitoring and early-warning module. This paper also designs the decision support technology based on the learning ability of neural networks. Through mapping the weight adjustment of neural networks into decision-making of supply chain, the short-term tactical decision technology can be put in practice to mitigate the unwanted situations reported by early-warnings.This paper firstly analyzes the existing supply chain early-warning and decision-support technology and points out that not only the exposed exceptions should be paid attention to, but also the dynamic trends of the monitored indicators is worthy to be highlighted. Then, a wide range of supply chain modeling methods is analyzed. In order to meet the requirement of tactical modeling, system dynamics is used to model the supply chain. For the requirement of real-time monitoring, system dynamics model is then transformed into equivalent neural networks and the input sequence is designed to adapt to the necessary of time-varying input. The architecture of monitoring and early-warning module is proposed. On the basis of neural networks used by monitoring and early-warning module, the learning and self-adjust ability of neural networks is used to implement the tactical short-term decision-support technology. Finally, a case study of the manufacturing industry and details of system implementation are presented to illustrate the methodology and architecture.