Abstract: The direction of arrival（DOA） of array signal is estimated by using BP neural network and maximum likelihood estimation（ML）. We combined the advantages of chaos mapping and swarm intelligence optimization algorithm to design a sines cosine chaotic double string grasshopper optimization algorithm（SCAGOA）. On the one hand, it can solve the problem of falling into local optimum due to improper selection of weights and thresholds of neural network; on the other hand, it can solve the problem of computational load and low efficiency caused by multidimensional search in maximum likelihood. The output effect and estimation error of two signal sources were discussed through simulation experiments. The generalization ability of different optimization algorithms to SNR were compared. The results show that the optimized BP neural network and maximum likelihood method have better estimation accuracy than other optimization algorithms in DOA estimation.