Inverse Determination Of Material Parameters For Tailor Welded Blanks Based On Decoupled GMDH Algorithm

Along with the increase of car ownership and speed, more and more attention is paid to the safety problem. Due to the high strength and high energy absorption, high strength steels are widely used in energy absorption components. Because of the influence of strain rate effect, the mechanics performance of high strength steel under the crash condition is different from that when it is under the static condition. Therefor, the accurate strain rate parameters are the premise of obtaining the high accuracy simulation.With the advantage of cost saving, excellent efficiency and flexible operability, metamodel technology has been utilized extensively. At the same time, the hybrid numerical method which is based on finite element simulation and metamodel has gradually become the mainstream method for parameters identification. To overcome the disadvantage of unsuitable for high-dimensional problems and not taking the coupling relationship between variables into consideration for the existing metamodel technique, this paper propose a new decoupled group method of data handling (GMDH) algorithm which join the coupling relationship judgment mechanism into the traditional GMDH. And the decoupled GMDH is successfully applied to parameters identification for the tailor welded blanks, specific work is as follows:(1) After the technical features of various reverse methods have been detailed discussed, the numerical method based on finite element simulation and metamodel technique is selected as the final process for this paper; Compared with other algorithms, the group method data handling (GMDH) have great advantage for high-dimensional problems and the GMDH is chosen as the basic metamodel method.(2) There is only one expression form for partial description in traditional GMDH, to deal with this weakness, two type improvement have been developed. For the first improvement, although the diversity of expression has been increased in a certain extent, the risk of deleting effective variables has also increased. While the second category has already deviated from the basic idea of traditional GMDH. In this article, a decoupled GMDH which combine a coupling relationship judgment and GMDH has been proposed. In decoupled GMDH, there will be a judge process before model construction, and the judgment results are used as criteria to determine the coupling terms. With this strategy, the risk of loosing effective variables could be in control while the self-organizing characteristic of GMDH could be also retained.(3) Two type test functions are used to test the decoupled GMDH. The results show that, the decoupled GMDH could achieve recognition of coupling relationship and model construction accurately; The lightweight design for frame case is used to verify the practicability of the method. And the results show that, decoupled GMDH could not only establish high precision metamodel, also could highly link up with optimization algorithm.(4) To be consistent with actual automobile structure, the energy absorption of the tailor welded blanks is taken into consideration. Based on the trolley experiment of tailor welded blanks, the decoupled GMDH method has been applied to metamodel construction between variables and target parameters. Then the optimization algorithm is used to search the best parameters combination for different materials. Compared with the test results, the decoupled GMDH has high precision.