Key Issues On Gre-Based Sdg Fault Diagnosis

This study focus on the Granular Computing (GrC), Graph Theory, Singed Directed Graph (SDG) and their applications in fault detection and diagnosis. The research reaches the cross-frontier of the computer science, information science and graph theory. The research results have great significance both for theoretical study and industrial applications of fault detection and diagnosis for complex systems in the real world.Combined with GrC, Graph Theory and SDG together, a new fault diagnosis method was presented in this paper, which has been successfully applied to the Tennessee-Eastman process (TEP).The research mainly includes but not limited in the following sections:(1) Granule-based SDG model was established and granule was applied to describe the elements in SDG, this is a new description method for SDG model.(2) By introducing GrC, system hierarchical theory, and graph theory into SDG fault diagnosis method, a new hierarchical SDG granular graph model was presented, which not only keep the completeness of the SDG but also simplify the interconnect relation among the system elements. Furthermore, the model was applied to High Pressure Heater System and TEP. The simulation results showed the effectiveness of the proposed algorithm.(3) Improving the existing Granular Matrix-based knowledge reduction algorithm by proposed node important-based and granlularity entropy-based knowledge reduction algorithm, which can help find the minimal attribute set and optimal diagnosis rule base.(4) A similarity based researching and reasoning algorithm was proposed, which can obtain the most possible fault sources by computing and sorting the similarity. This method can improve resolution by effectively reducing or even avoiding conflict probability of reasoning results.(5) By applying Fuzzy theory to the definition of the node state and tributary state, realizing the quantitative fuzzification of qualitative SDG model. Fuzzy set-based consistent path decision making algorithm was also proposed, which result in a fuzzy-SDG-based fault diagnosis method.(6) Design and develop a real time fault predictive diagnosis system based on configuration software, which realize the simulation for TEP, and prove the efficiency of the proposed algorithm.The innovations of this paper are as follows:(1) Proposed a hierarchical SDG fault diagnosis model and established hierarchical SDG model for High Pressure Heater System and TEP.(2) Proposed two knowledge reduction algorithms respectively based on node importance and granularity entropy;(3) Proposed a research and reasoning algorithm based on similarity;(4) Developed a real time predictive fault diagnosis method based on configuration software, which has been successfully applied to the TEP diagnosis.