Fault Diagnosis Method Based On Complex Networks Theory And System Implementation

The research and application of complex network fields are the hot issues in recent years. In this paper, by setting the network model of chemical process, using complex network theory, implement the fault diagnosis and system brittleness, thus overcome the disadvantages of traditional method that cannot overall describe the process system or extract the topology knowledge among data. Avoiding the strict assumptions of Gaussian distribution in the method based on multivariate statistics, developing a new method and application in research in the field of chemical process, has important theoretical significance and application value.The research content in this paper is as follows:①Research the complex network theory, propose a fault diagnosis method based on complex network. Using partial correlation coefficient determine the relationship between variables, determine the adjacency matrix, generate network model, start from the network topology, distinguish whether system running state has been changed, compare with difference of characteristic parameters between two states, and find the variable of failure;②Research the system brittleness analysis based on network topology, establish complex network model of system, and on this basis, using the method for identifying the key node and edge in complex networks, proposed the brittleness index, has certain guiding significance in monitoring important equipments and links, prevention of major system failure, construction and evaluation of complex system; ③Develop a fault diagnosis and brittleness analysis software based on Python, using TE process model as the application object, imitating the process to verify the effectiveness of the method proposed in this paper.Now, using complex networks to diagnosis fault in complex process system is still in the attempt phase, still need lots of theory research and experiment verification.