Research On Underwater Acoustic Channel Modeling Based On Digital Twin

Underwater acoustic communication technology plays a vital role in marine research and development.However,the underwater acoustic channel has complex time-varying,frequency-varying and space-varying characteristics,which seriously affect the reliability and effectiveness of the underwater acoustic communication system.Therefore,constructing a model that accurately reflects the characteristics of the underwater acoustic channel is the key to designing an underwater acoustic communication system.Aiming at the problem that the traditional underwater acoustic channel modeling method is limited by the application conditions and cannot satisfy the simulation accuracy and environmental adaptability;this paper applies the digital twin technology to the underwater acoustic field and carries out the digital twin modeling with the underwater acoustic channel as the core.The research aims to build an underwater acoustic channel model with high fidelity and high fitness.By studying the digital twin modeling method that combines the ray model of the underwater acoustic channel and the machine learning model,it breaks through the technical difficulty that the simulation of the underwater acoustic channel is completely dependent on the physical model and realizes the accurate mapping of the underwater acoustic channel in the digital space.The main research content of the paper around the underwater acoustic channel digital twin modeling method includes the following aspects:The theoretical research on digital twin technology is carried out,and the definition and composition structure of digital twin given in different application fields in recent years are summarized.On this basis,this paper proposes a digital twin framework suitable for underwater acoustic channel reconstruction,which mainly includes two parts: the acoustic ray model and the machine learning model and provides a technical route for underwater acoustic channel modeling.Aiming at the problem that the accuracy of the environmental twin data affects the perception of the ocean environment by the underwater acoustic channel model,the absolute median method and the moving average filter are firstly used to detect and correct abnormal temperature,salinity,and depth data.At the same time,in order to improve the accuracy of the estimated sound velocity,this paper uses harmony search,improved harmony search,particle swarm,and harmony-particle swarm hybrid algorithm to optimize the parameters of Medwin and Leroy’s sound velocity empirical formula,compare the performance of the algorithm and select the convergence speed and optimization ability The balanced harmonic-particle swarm algorithm optimizes the Medwin formula to obtain the sound velocity that fits the real ocean environment.In view of the lack of received data and the difficulty of training and testing the digital twin model of the underwater acoustic channel,this paper first uses the processing method of additional environmental noise for data enhancement and constructs the measured received signal data sets with different signal-to-noise ratios to effectively overcome the insufficient amount of measured received data.The problem is to fully provide real underwater acoustic channel characteristics.In response to the issue of incomplete acquisition of marine environmental elements,which makes it impossible for the physical model of the underwater acoustic channel to fully perceive the state of the underwater acoustic channel,based on the ray model of the underwater acoustic channel,this paper adds weak disturbances to the measured marine environment elements to meet the consistency and quantity requirements of the marine environment.Time-invariant simulated received signal sample set.By combining Generative Adversarial Network(GAN),Deep Convolutional Wasserstein Generative Adversarial Network(DCWGAN)and Deep Convolutional Wasserstein Generative Adversarial Network with Gradient Penalty(DCWGAN-GP),the underwater acoustic channel digital twin construction based on GANs is carried out.model research.The three models were trained offline and tested online with the help of the 2022 Laoshan measured and simulated data sets,and finally the DCWGAN-GP model with the best test results was selected as the machine learning model in the digital twin framework.The experimental results verify that the digital twin model of the underwater acoustic channel has high fidelity.Using the Zhanjiang sea test data in 2022,it is verified that the digital twin model of the underwater acoustic channel based on DCWGAN-GP has environmental adaptability,and the model can effectively analyze and predict the state of the real underwater acoustic channel.