摘要：An intelligent response surface methodology (IRSM) was proposed to achieve the most competitive metal forming products, in which artificial intelligence technologies are introduced into the optimization process. It is used as simple and inexpensive replacement for computationally expensive simulation model. In IRSM, the optimal design space can be reduced greatly without any prior information about function distribution. Also, by identifying the approximation error region, new design points can be supplemented correspondingly to improve the response surface model effectively. The procedure is iterated until the accuracy reaches the desired threshold value. Thus, the global optimization can be performed based on this substitute model. Finally, we present an optimization design example about roll forming of a "U" channel product.

摘要：The inverse heat conduction method is one of methods to identify the casting simu- lation parameters. A new inverse method was presented according to the Tikhonov regularization theory. One appropriate regularized functional was established, and the functional was solved by the sensitivity coeffcient and Newton-Raphson iteration method. Moreover, the orthogonal experimental design was used to estimate the ap- propriate initial value and variation domain of each variable to decrease the number of iteration and improve the identification accuracy and effciency. It illustrated a detailed case of AlSi7Mg sand mold casting and the temperature measurement ex- periment was done. The physical properties of sand mold and the interfacial heat transfer coeffcient were identified at the meantime. The results indicated that the new regularization method was effcient in overcoming the ill-posedness of the inverse heat conduction problem and improving the stability and accuracy of the solutions.