Maneuvering Modeling Of Underwater Vehicle Based On Support Vector Machine

With the continuous exploration and development of the ocean,people realize that the underwater vehicle plays a very important role in the process.Control system is the key part in the process of tudy of underwater vehicle.The establishment of an appropriate mathematical model is the first problem to solve in the whole control system.As one of the three theories of cybernetics,system identification has become an important method of mathematical modeling of the system since it generated,and also has a good use in the test.To some systems,their operation mechanism is not very clear,so derive their equivalent mathematical model through the input-output data,this method is also called black-box modeling.For some other research objects,people accumulate a lot of experience and knowledge through previous research results and existing technical methods,and have a very complete theoretical basis,the structure of the model has been widely accepted by the field.When we study such problems,we can only establish complete mathematical model by identifying some parameters.In recent years,the method of system identification is used for modeling the underwater vehicle,and has achieved good results.In the previous modeling,the system identification method is usually used to identify the coefficients in the model.With the development of artificial intelligence,black-box modeling of the system is easier,this paper uses an artificial intelligence method-support vector machine for black-boxmodeling of underwater vehicle.This paper study the establishment of motion model of a underwater vehicle based on the method of system identification of support vector machine?Support Vector Machines,SVM?,combining Z maneuvering and circular motion.First of all,simulate the 10o/10o maneuvering motion of the underwater vehicle,and establish the model of maneuvering motion by black-box modeling using the obtained data,then predict the maneuverability of the underwater vehicle.Compared the simulation data with the forecast results,verify the correctness of the method of black-box modeling based on support vector machine.Second simulatehe 20o/20o maneuvering motion and 15 circular motionof the underwater vehicle,establish the model by black-box modeling,the consistence of simulation result and prediction result verify the Generalization of SVM.Finally,the maneuvering hydrodynamics of the underwater vehicle is identified,and the input-output data used for training support vector machine is obtained by simulating the 25^o/5^omanipulating motion.The trained support vector machine is used to predict the hydrodynamics of 5^o/1^o,10^o/1^o,10^o/5^o,15^o/5^o manipulating motion.The results show that the support vector machine has good nonlinear mapping ability and modeling ability.The selection of structural parameters and kernel parameters of support vector machine has adopted the method of trial and error in the past,but it affects the accuracy of prediction.In order to reduce the difficulty of choosing structure parameter and kernel parameter of support vector machine,using the modern swarm intelligence algorithm-Fruit fly optimization algorithm to optimize the parameters of support vector machine,and applying to the black-box modeling of maneuverability motion of underwater vehicle.The predicted results obtained by parameter optimization are compared with results by the method of trial and error,which shows that the optimized results are more accurate.