Abstract: The direction of arrival (DOA) of the target signal is a critical parameter for simulating scale targets using towed flexible linear arrays. However, under low signal-to-noise ratio (SNR) conditions, distorted arrays will result in deteriorated performance in DOA estimation. To address this issue, this paper proposes a DOA estimation method that integrates array shape estimation and signal decomposition denoising to enhance estimation accuracy. First, by deploying heading sensors within the flexible array to obtain heading information of the array elements, the array shape is estimated using linear interpolation. Then, successive variational mode decomposition (SVDM) is applied to extract the modal components of the signals received by the receivers, and appropriate components are selected for cross-correlation operations to calculate the time delay. Finally, the time delay values are substituted into the DOA estimation model to obtain the final results. Simulation experiments validate that, under significant array distortion and an SNR of 0 dB, the proposed method achieves a root mean square error of less than 1.21° for direction of arrival estimation, representing a significant improvement in accuracy compared to noisy ideal arrays.