Modeling And Fault Detection For Electroslag Remelting Process

Electroslag remelting (ESR) is one of the primary methods of electroslag metallurgy, the melting steel of it has high purity, dense structure, and homogeneous composition, smooth surface and excellent mechanical properties and becomes the first choice of high-grade steel and special steel. Electroslag furnace with double electrode series can increase the power factor to above0.9for its unique power supply, which effectively decreases the consumed electricity and improves the efficiency.This paper is researched based upon practical engineering project, and the research object is electroslag furnace with double electrode series. Through further study of the principle, characteristics and technology of the ESR, the lumped dynamic model of ESR, which uses current and voltage as inputs and electrode melting speed as output, is established. The model contains magnetic voltage regulator, hydraulic pressure system, electrodes location, slag resistance, slag pool temperature and electrode melting speed. According to electroslag metallurgy craft and data from previous experiments and particular calculation and derivation, the value of each parameter in the model is determined. By building the simulation model of the system in Matlab/Simulink environment, doing simulation and analyzing the result, which is compared with the actual ESR process, we can verify the validity of the model. It can provide an object model for the research of control methods and others research of electroslag furnace.In order to guarantee the quality of the ingot, we need to monitor the ESR process. In recent years, the process monitoring and fault diagnosis methods based on multivariate statistical analysis have been widely applied in the industrial process. However, the application the multivariate statistical analysis methods in the ESR process are very few. According to the ESR process characteristics, we divided the whole ESR process into three stages in this paper, the slag period and the rapid rise current stage of normal smelting period is the first stage, the stability stage of the normal melting period is the second stage, the filling shrinkage period is the third stage. The first and the third stage, which with serious nonlinear characteristics, can be considered as a batch process, we can use multi-way kernel principal component analysis (MKPCA) method for monitoring. The second stage can be considered as a relatively stable continuous process, we can use principal component analysis (PCA) method for monitoring. Finally, modeling with the10batch data which were generated by electrodes radius random fluctuations, and taking the cooling water flow fault, sensor fault and hydraulic system leak fault of the ESR process as example to do process monitoring, the monitoring results are quite well and can verify the validity of the method.Finally, conclude with a summary and some further research areas in this thesis.