摘要:For the composite glass box girder, the generalized Bayesian objective function of elastic constants of the structure was derived based on layered shell element theory. Mechanical performances of the composite glass box girder were solved by layered shell element method. Combined with quadratic parabolic interpolation search scheme of optimized step length, the adaptive Powell’s optimization theory was taken to complete the stochastic identification of elastic constants of composite glass box girder. Then the adaptive Powell’s identification steps of elastic constants of the structure were presented in detail and the adaptive Powell’s identification procedure was accomplished. From some classic examples, it is finally achieved that the adaptive Powell’s identification of elastic constants of composite glass box girder has perfect convergence and numerical stability, which testifies that the adaptive Powell’s identification theory of elastic constants of composite glass box girder is correct and reliable. The stochastic characteristics of systematic responses and elastic constants are well deliberated in generalized Bayesian objective function. And in iterative processes, the adaptive Powell’s identification is irrelevant with the complicated partial differentiation of the systematic responses from the layered shell element model to the elastic constants, which proves high computation efficiency.