Research On Control Method Of Unmanned Craft Based On Fuzzy Support Vector Machine

As a integrated system of control,navigation and path planning,unmanned craft compared with the conventional boat,its complexity and design difficulties have been greatly improved,but based on its own advantages: rapid response rate,good maneuverability and high reliability,so that it has an important position in various military activities.In addition to military use,civil also began to rise gradually,began to serve as water detection,shipping and other related work,and thus the birth of a group of new High-tech industries.As the intelligent control technology is more and more mature,the research on unmanned craft has put forward higher requirements,and needs better maneuver and control ability.This paper mainly aims at the mathematical modeling analysis of unmanned craft in the experiment,and carries on the simulation and the physical experiment on the basis of understanding its concrete model.The real experiment of unmanned craft mainly adopts the Non-linear PID algorithm in the traditional industry,obtains the unmanned craft actual operation effect,basically satisfies the control request.Then,according to the problem of nonlinear PID,the intelligent algorithm of fuzzy support vector machine is proposed,and the available models are obtained by collecting a large number of data sets in the course of unmanned craft.Considering the defect of fuzzy support vector machine in parameter optimization and model,the particle swarm algorithm and weight pruning algorithm are used to improve the whole system running time and reduce the model size.First of all,according to the current research situation of unmanned craft to describe,understand the relevant development.Then,using the commonly used MMG model analysis technology,the experimental unmanned craft is modeled,and the function and characteristic of unmanned craft are deduced by the measured data.Secondly,by comparing with the intelligent architecture of the common robot,the characteristics of the unmanned boat motion control system are obtained and the software and hardware design is completed.The experiment was carried out with the combination of nonlinear PID and Non-linear method.Through the simulation and the physical operation to obtain the relevant experimental results,prove the whole system structure and algorithm feasibility and applicability.Thirdly,according to the disadvantages of PID algorithm,the algorithm of fuzzy support vector machine is presented,and the whole system architecture and algorithm implementation of fuzzy support vector machine are introduced first.According to the characteristics and distribution of the speed heading data of unmanned craft,the fuzzy membership function and support vector machine parameters of the algorithm are presented,and the model training is carried out.The training model is verified by some unused data,and the performance index of the algorithm can be obtained.Finally,in the original fuzzy support vector machine method,based on the parameters to find time too long and training model is too large to propose an improved method.In the aspect of parameter optimization,the particle swarm can be used for quicker searching in the heuristic way,and the weight with little influence on the size of the model can play a better role.The experiments show that the two methods can improve the time and size of SVM model training,and the training precision and generalization ability are not decreased,and the robustness is improved correspondingly.