基于群体智能协商算法的金融系统网络抗风险研究

英文摘要(限3000 Characters)： The collective and cooperation behavior of living beings, such as flocks of bird， shoals of fish, herds of wildebeest has certain advantages, including avoiding predators, increasing the chance of finding food, saving energy, etc. Inspired by biological behaviors, swarming intelligent has become an active research area in the past few years. Swarming intelligent has been widely used to control or to organize large scale distributed systems, such as controlling the traffic flow, airport flights, network communication and managing economic\social systems. Through the literature review, it is found that the swarming intelligence has been used to research contagion problems in finance network. Most of the research outcomes are related to the failure of finance system with different structures. But there is no effective crisis management to protect the finance network. Our research will focus on the contagion prediction and crisis protection of finance network. A distributed controller will be designed to predict and protect the finance network from contagion destruction of finance crisis. In this project，we propose to model the finance network as a graph. For the purpose of analyzing, the finance agents and funds transaction channels between them are modeled as vertices and edges of a graph and the finance network can be analyzed by using the graph theory. The graph of a finance network represents the interaction relationship between the finance agents. By using the characteristic matrix of graph, we can theoretically analyze the finance network and create a mathematical model of finance network. Based on the finance model, a consensus algorithm will be used to create the prediction and protection mechanism for the finance network. The consensus algorithm is very important to address agreement problems in the area of distributed computation. It is widely used in the field of distribute intelligent systems in which groups of agents need to agree upon certain quantities of interest. The quantities might be related to the movements of the individual agents or might be related to the properties of agents. In this project, the consensus will be embedded into a "Kalman filter" to estimate and reach an agreement on the crisis level among the finance network. Furthermore, the consensus algorithm will be used to design the crisis protection mechanism which is triggered by consensus crisis prediction result.

本课题以群体智能算法为基础，建立分布式金融系统网络的风险预警机制和抗风险保护机制。 首先，利用图论和控制理论建立金融系统网络的连续化数学模型。 并以金融系统的资金流动形式为研究对象，探究金融系统在金融危机冲击下的资金流动模式、危机传染方式和传染路径，从而为风险预警机制以及抗风险对策研究奠定基础。 其次，利用群体协商式算法实现分布式金融系统网络预警机制，并采用群体协商式卡尔曼滤波器实现风险预测。 最后采用群体协商算法建立分布式金融系统的保护机制。 本项目将揭示在金融危机冲击下，分布式金融系统的资金流动特征与形式。 并利用群体智能算法从宏观上实现对整个金融系统的预警和保护。 该项目的研究成果可以为国内金融系统的控制和保护策略提供科学依据。 对于推动我国金融系统网络的建设和发展，加强国内金融系统的抗风险能力具有重要意义。

作为新兴的智能算法， 群体智能被大量的应用于控制大规模分布式系统。 本项目利用群体智能算法建立分布式金融系统网络的风险预警机制和抗风险保护机制。 课题组采用图论建立了金融系统网络的连续化数学模型，并利用该模型建立了金融网络系统的模拟平台。 通过模拟实验，课题组深入探究了金融网络系统的资金流动形式，以及在金融危机冲击下的资金流动模式、危机传染方式和传染路径。 并利用群体协商式算法实现分布式金融系统网络预警机制和保护机制。 本项目揭示了在金融危机冲击下，分布式金融系统的资金流动特征与形式。 并利用群体智能算法从宏观上实现了对整个金融系统的预警和保护。 其研究成果可以为国内金融系统的控制和保护策略提供科学依据。 对于推动我国金融系统网络的建设和发展，加强国内金融系统的抗风险能力具有重要意义。