Sound absorption performance of polyurethane foam could be improved by adding a prepositive microperforated polymethyl methacrylate panel to form a composite sound-absorbing structure. A theoretical sound absorption model of polyurethane foam and that of the composite structure were constructed by the transfer matrix method based on the Johnson–Champoux–Allard model and Maa’s theory. Acoustic parameter identification of the polyurethane foam and structural parameter optimization of the composite structures were obtained by the cuckoo search algorithm. The identified porosity and static flow resistivity were 0.958 and 13078 Pa·s/m2 respectively, and their accuracies were proved by the experimental validation. Sound absorption characteristics of the composite structures were verified by finite element simulation in virtual acoustic laboratory and validated through standing wave tube measurement in AWA6128A detector. Consistencies among the theoretical data, simulation data, and experimental data of sound absorption coefficients of the composite structures proved the effectiveness of the theoretical sound absorption model, cuckoo search algorithm, and finite element simulation method. Comparisons of actual average sound absorption coefficients of the optimal composite structure with those of the original polyurethane foam proved the practicability of this identification and optimization method, which was propitious to promote its practical application in noise reduction.