Improved genetic algorithm, combined with artificial neural network, is present for the optimal
design of 2.5D braided composite. Dispersal simulation data, including maximal stresses and elastics
properties, are adopted by artificial neural network for the calculation of strength property. Based on
calculation method of strength mentioned above and other calculation models for other mechanical
properties, genetic algorithm is employed for the design of structure parameters of 2.5D braided
composite, such as wrap fiber density, fill fiber density and interface strength. These structure optimal
parameters are finally optimized for practical application.