Optimization And Application Of Deep Neural Network Models For Computer Visio

Since the underactuated ship not only has multiple input and output,nonlinearity,and non-integrity,but also includes uncertain items of the model and external environmental interference,and the trajectory tracking requires a specified time to reach the target position,the traditional control method cannot be directly applied to the underactuated ship.Therefore,it is of great significance to study the trajectory tracking control problem of ships.This thesis mainly completed the following research work:1)The three-degree-of-freedom mathematical model of surface ships and some relevant knowledge required for the analysis and design of ship trajectory tracking problems were sorted out,including RBF(Radial Basis Function,RBF)neural network theory and BLF(Barrier Lyapunov Function,BLF),etc.Relevant control theory,etc.,lay the foundation for the follow-up research.2)For the underactuated ship trajectory tracking control problem with model uncertainties and external environment interference,firstly,in order to realize the output-limited control of the ship trajectory,BLF is used to constrain the distance deviation and heading angle;then combined with the dynamic surface control method,a trajectory tracking control method for underactuated ships based on RBF adaptive neural network is proposed,and the stability analysis is carried out;finally,it is verified by simulation experiments that the proposed control strategy can effectively solve dynamic uncertainties,unknown environmental disturbances and constraint output problems,and has the advantages of simple calculation and easy engineering implementation.3)For the track tracking control problem of underactuated ships under the influence of ocean current drift,firstly,in order to better estimate the uncertain items in the ship model motion,the series-parallel model in system identification is used to predict the ship longitudinal velocity and bow angular velocity,and the prediction error and tracking error constitute the adaptive law of combined learning;then the RBF neural network is used to approximate the uncertain items of the system,and then the output of the RBF neural network and the estimated value of the disturbance observer are used for the ship trajectory tracking controller Design;Finally,through the comparison of simulation experiments,the combined learning adaptive RBF neural network control method has better control characteristics,and is more suitable for trajectory tracking with higher precision requirements.4)For the underactuated ship model whose inertia matrix and damping matrix are off-diagonal matrices,it is transformed into a form in which the off-diagonal elements are zero through matrix coordinate conversion,and then a combined learning underactuated ship model is designed using BLF and RBF neural networks.The trajectory tracking of the driving ship outputs a limited control strategy.In order to verify the effectiveness of the above control strategy,a container ship was used as the simulation test object.The results show that the control method can achieve the constraints control of ship trajectory tracking output under wind and wave disturbance,and expands the scope of application of the control method in asymmetric ship models.