Fault Diagnosis In Continuous Annealing Furnace Temperature System Based On PCA-SDG

With the rapid development of automatic technology in modern process industries, security and reliability are being greatly needed by people to avoid the large economic losses caused by various accidents or even breakdowns of industrial equipments. Therefore, it is very significant for industrial integrated automation to develop and research process monitoring system which has the integrated functions of control, monitoring and diagnosis. Fault detection technology monitors the conditions of production process, detects the fluctuation and faults and locates the fault sources, then isolates and eliminates them. In this way, catastrophic accidents are prevented, losses are reduced and qualities of products are improved.This thesis is based on the continuous annealing line, the fault diagnosis method based on PCA-SDG and its application in continuous annealing line furnace temperature system is mainly researched. PCA (Principal Component Analysis) can utilize the statistical characteristics of the variable more efficiently. The intrinsic characteristics of the process can be described through the monitoring variables by means of the statistical property. So in this thesis, PCA method is used to monitor the fault. SDG (Signed Directed Graph) is a fault diagnosis method which based on qualitative knowledge. It can express the relationship effectively in complex system. After the system fault is being detected, SDG is used for fault diagnosis. The detailed procedures of the purpose mentioned is as follows:the SDG model of continuous annealing line furnace temperature system is used to explain the residual contribution which produced by PCA, so the nodes’ conditions are made sure; Finally, according to SDG inference rule the system fault is diagnosed. The fault diagnosis method based on PCA-SDG combines the advantages of PCA and SDG, which is not only resolves the problem that PCA can’t separate the fault resource, but also avoids the shortcomings of using SDG alone.Accordiing the simulation in this thesis, the effectiveness and reliability of the PCA-SDG method are verified. The fault diagnosis method based on PCA-SDG reduces diagnosis time and elevates the level of process automation. So PCA-SDG will become an important method in monitoring and diagnosis field in the future.