Fuzzy Estimation For The Thermal Boundary Conditions Of Slab Continuous Casting Mold

In the field of metallurgical engineering, mold as the core part of the castingsystem has a significant impact on the process of solidification, slab surface quality andproduction efficiency of caster. Numerical simulation method has been widely used tostudy the casting solidification process. To obtain the reliable simulation results, wemust accurately determine the boundary conditions of heat transfer interface betweenthe slab and the mold. However, it is difficult to measure the heat transfer boundarycondition directly, especially the heat flux distribution at the interface between mold andmetal, since the actual working environment is very complex.Using the inverse heat conduction problem (IHCP) method to estimate the heattransfer boundary conditions at the interface between the slab and the mold according tothe measurement information of thermocouple buried into the mold has become a viablesolution.Decentralized fuzzy inference (DFI) is a solving method for the IHCP with a goodresistance to ill-posed on the basis of the fuzzy set theory. DFI method belongs to thescope of uncertainty reasoning method. Compared with the traditional methods for theIHCP, it has good processing ablitity for the non-deterministic information, takeadvantage of the inaccurate information for reasoning and decision-making, and possesgood fault tolerance and robustness for reasoning.Inversion for the heat transferboundary conditions at the wall of the slab mold is studied by using the DFI method inthis paper. A synthesizing weighted approach based on the normal distribution isestablished in terms of the problems of the exsiting QDFI method based on qualitativeanalysis. Finally, a decentralized fuzzy inference (NDFI) method based on normaldistribution weighted approach is formed. The main work is as follows:①A two-dimensional heat transfer model for the slab mold is established, andfinite element analysis (FEA) is used to slove the mold temperature distribution. Thegrid-independent verification for the heat conduction model is conducted.The influenceof the withdrawal speed, the cooling water flow rate and the cooling tank arrangementon the slab mold temperature distribution are taken into consideration through thenumerical simulation experiments.②The basic theory of DFI method is introduced, and the heat flux distribution atthe wall of slab mold is obtained by the QDFI method based on qualitative analysis. Numerical tests have been carried out to consider the effects of the temperaturemeasuring point configuration, the initial guesses of heat flux distribution andmeasurement errors on the inversion results. The results of the QDFI are also comparedwith CGM, and the limitations and problems of the QDFI and CGM for sloving theIHCP are summarized.③In order to futher improve the application effect of DFI method on the IHCP,and for the problems of QDFI method, a synthesizing weighted approach based onnormal distribution function to determine the effect relationship matrix is established.Finally, a decentralized fuzzy inference (NDFI) method based on normal distributionweighted approach is formed. The inverse problem of plate heat transfer boundarycondition is solved by the established NDFI method. The effectiveness of the proposedNDFI method for solving the IHCP is verified by retrievaling the heat transfer boundarycondition of the plate.④The NDFI method is applied to the inverse problem for heat transfer conditionof slab mold. Numerical tests have been carried out to consider the effects of thetemperature measuring point configuration, the initial guesses of heat flux distributionand measurement errors on the inversion results. Comparisons with both the existingQDFI method are also conducted, and they all show the validity of the established NDFIinversion method for solving the inverse problem for the thermal boundary conditiondistribution of slab mold.