The presence of interference sources presents an obstacle to the accurate determination of the location of the desired source. To overcome this challenge, this paper proposes a novel Direction of Arrival (DOA) estimation algorithm. We adopt an approach that involves learning the environmental characteristics before proceeding with DOA estimation, accounting for potential variations in environmental parameters. In detail, the algorithm initially adopts a domain adaptation concept to perform linear mapping of the sampled signals in an interference-laden environment, aiming to remove interference noise. Subsequently, an online linear mapping update scheme based on the Z-test is introduced to ensure the precision of DOA estimation. Finally, the processed signals are estimated using the Multiple Signal Classification (MUSIC) algorithm. Based on these algorithms, a complete online adaptive DOA estimation framework is established. Simulation results confirm the efficacy of the algorithm proposed in this paper, demonstrating its ability to automatically update the mapping matrix and enhance the accuracy of DOA estimation, thereby mitigating the impact of interference noise.