Quantitative identification of three-dimensional subsurface defect based on the fuzzy inference of thermal process

Based on the inverse heat transfer methodology, the quantitative identification of three-dimensional subsurface defect is studied by using the finite element method and fuzzy inference method. For the quantitative identification problem of subsurface defect, a fuzzy inference system is established. A set of fuzzy inference units are created, through which the decentralized fuzzy inference processes are executed from the deviations between measured and computed temperature to the corresponding inference components. These fuzzy inference components are weighted and synthesized to gain the compensations of the defect parameters. Numerical tests are performed to study the effects of defect size, the initial guess value, the measurement errors, the number of measurement points, the defect shape and the convective heat transfer on the identification results. Comparison with the Levenberg-Marquardt method (L-MM) is also conducted. The results show that the established method in this paper can significantly reduce the dependence of identification results on the number of measurement points and improve the anti-interference ability on the measurement errors, also has better anti-illposedness characteristic and higher computational efficiency.