Research On Underwater Vehicle Flocking Control Algorithm Based On Channel Prediction

With the wide attention of ocean resources,autonomous underwater vehicle(AUV)is an important tool for underwater operations.Due to the multi-dimensional requirements of underwater tasks,a collaborative flocking system of AUVs is urgently needed.Flocking of AUVs refers to the phenomenon that multiple AUVs accomplish flocking collaboration behavior by self-organization using limited environmental information and efficient control strategies.AUVs in a flocking system coordinate a non-centralized scheduling population through information and action interaction with neighboring individuals.Current research focuses on the networking and stability control of AUVs,ignoring the quality of underwater acoustic communication channel.Based on the prediction of signal to noise ratio(SNR)parameters for underwater acoustic channels,the flocking control of AUVs is studied in this paper.The main contents are as follows:(1)Set up a flocking experimental platform for AUVs,which is composed of three AUVs.The AUV is mainly composed of host controller,positioning unit,wireless data transmission unit and electrical unit.The host computer receives the status information of AUV.The flocking experimental platform has completed the test and carried out many experiments.The experimental platform has the functions of track tracking,positioning,collision avoidance and path planning to support theoretical validation.(2)Aiming at the problem of attenuation of communication channel of AUV flocking system,a reinforcement learning estimator for predicting the SNR parameters of underwater acoustic communication based on value iteration is designed.The channel parameters are estimated with limited underwater acoustic channel information,considering the characteristics of path loss,shadow fading and multipath fading.The estimator predicts SNR of AUV at the location of the buoy in the determined area to provide communication support for the AUV flocking system.Simulation results show that,compared with the least-squares estimation method,the value iteration-based reinforcement learning estimation method can effectively avoid obtaining local minimum values,which proves the validity of the theory.(3)To solve the channel attenuation and unknown model constraints of AUV flocking movement,a model predictive AUV flocking controller based on SNR prediction for reinforcement learning is designed.An optimal framework for joint optimization of underwater communication and flocking cooperative control is constructed,including SNR prediction of underwater acoustic communication and AUV flocking control.The AUV flocking controller links the communication quality with the flocking stability to achieve a balance between flocking stability and communication efficiency.Considering the impact of obstacles in the environment,AUV flocking control strategy based on reinforcement learning algorithm can avoid collisions.At the same time,by predicting the AUV model,the dependence on the AUV model parameters is avoided.The validity of the theory is verified by simulation,and the feasibility of AUV flocking control strategy is verified by experiments.