Search

Under the tide led by artificial intelligence, the application of deep learning methods in underwater target recognition has gained increasing attention. However, obtaining underwater sample data from actual scientific research is extremely difficult because of environmental constraints, time, cost, and other factors. An insufficient number of samples leads to poor training effects in deep learning models. Generative adversarial networks(GAN), a new artificial intelligence technology, have a wide range of applications in data enhancement, image generation, and other fields. However, the traditional GAN model is unsuitable for underwater echo samples and cannot be used directly for sample data generation. Therefore, this study proposed an underwater echo signal generation method based on an improved GAN for a small sample to address the problem of insufficient underwater target data. Combining the characteristics of the echo signal, the study designed and built a convolution unit-based GAN model. Furthermore, an echo signal generation simulation experiment was carried out using the pool experimental test data. Finally, the effectiveness of the generated signal was verified at the level of the signal waveform and amplitude probability distribution. The experimental results show that the improved GAN model proposed in this study is suitable for generating highly realistic original echo signals in the case of small samples, which provides a new idea for the active detection and recognition of underwater targets.

Ship wake detection is an effective method for undersea vehicles to detect and track surface ships. However, traditional wake detection methods based on time-domain features are limited by subjective experience and complex varying marine environments with certain limitations in the intelligent detection of ship wakes. Aiming at the problem of insufficient accuracy and efficiency of traditional wake detection methods in complex environments, this study introduces a deep learning theory to improve the independent learning and adaptive capabilities of the model. Subsequently, a novel ship wake detection method based on transfer learning and time-frequency analysis is proposed, considering the difficulty of obtaining ship wake samples. First, in this method, the time-frequency domain characteristics of the wake signal are extracted using short-time Fourier transform. Then, through the strategy of parameter freezing and fine-tuning, transfer learning based on a pre-trained convolutional neural network is completed. Finally, the effective detection of ship wakes using a small sample dataset is realized. Combining the wake detection experiment with actual measured data, the results show that, when compared with the traditional wake detection method, the correct rate of the wake detection method based on time-frequency analysis and transfer learning is increased by approximately 10%, with the highest reaching 97.49%. It combines the characteristics of a low time cost, low sample demand, and high recognition performance.