Abstract: The classical genetic algorithm (GA) itself has the limitation of weakness local search ability，converging prematurely and reducing convergence speed in the anaphase. The theory of adaptive strategy and niche technique of pre-selection mechanism are introduced into the GA. Niche technique is used to avoid the problem of converging to the part-optimum. Adaptive strategy is presented for the adaptive controlling of the crossover probability and mutation probability. The new type of algorithm is adaptive niche genetic algorithm (ANGA)，for the purpose of maintaining the population diversity and improving the reliability of holistic convergence. This algorithm is verified by three typical examples. Besides, a numerical case of 70m single-layer reticulated dome is given to demonstrate good operation steadiness and strong holistic searching capability，however the consuming time using proposed algorithm is longer than that using ANSYS optimistic built-in module. The algorithm can be applied to section optimization of large and medium reticulated dome with several optimum variables，and can ensure the quality of results.