Research On AUV Intelligent Control And Parameter Optimization Technology

With increasing knowledge of the sea and rapid development of high technology,autonomous underwater vehicles(AUVs)have been widely applied to military and civil fields including mine detection,intelligence gathering,seabed resource development as well as dam and pipeline detection.In respect of the growing dependence on AUVs and increasingly complicated operation environment,there have been higher requirements on AUV motion control performance.Admirable motion control performance is known as the premise for AUVs to complete different operation tasks.In order to improve AUV motion control performance,studies are carried out in this dissertation in the following aspects.a)In order to facilitate the study on AUV motion control and parameter optimization technology,the platform of the research object is introduced,on the basis of which the hardware and software architecture of the control system is constructed in accordance with the sensors and actuators equipped on the research object.b)Studies are carried out on sparse-representation-based filtering technology.For the interest of higher precision of sensor output,theories of sparse representation and measurement of data sparsity are introduced.The existence and uniqueness of the optimal sparse representation solution is certified.On the basis of K-Singular Value Decomposition algorithm,the realization of over-complete redundant dictionary training is elaborated.A sparse-representation-based filtering model is established.The research object’s Doppler Velocity Log data in simulation experiments is processed and the Doppler Velocity Log data in sea trials is processed in real time with the sparse-representation-based filtering technology.c)Studies are conducted with the focus on a novel sigmoid-function-based control method based on sliding mode structure.Given that the residual buoyancy is not taken into consideration in the classic sigmoid-function-based control model and the existence of considerable residual buoyancy at time of high-speed motion degrades the control effect,a novel sigmoid-function-based control method is put forward with sliding mode structure and residual buoyancy included in the AUV motion control model.Lyapunov stability analysis is conducted for the novel sigmoid-function-based control method.In view of the joint actuators of thruster and rudders,thrust allocation strategies are also described.Simulation experiments and sea trials are carried out with the proposed novel control method.d)A control parameter optimization method based on improved artificial fish school algorithm is put forward.Due to the shortcomings of existing research on multiple peaks during AUV control parameter optimization,the original artificial fish school algorithm is improved while its excellent global searching ability is remained.By introducing adaptive visual and step length,and ignoring the crowded factor to intensify the acts of following and gathering,the searching precision and efficiency of optimized parameters are improved.The model of control parameter optimization is established.Simulation experiments and sea trials are carried out with control parameters optimized by the improved artificial fish school algorithm.e)Research on Least Squares Support Vector Regression(LSSVR)interactive network is also carried out.In order to enhance the offline and online learning speed and precision of the controller,a control method based on LSSVR interactive network is therefore proposed.The function of each module and interaction flow of the network is introduced.Based on LSSVR online learning method,dynamic characteristics of the research object are identified.Offline design and online optimization of the control law are realized.Simulation experiments and sea trials are conducted with the proposed control method.f)Contrast sea trials for the LSSVR interactive network control method,the parameter optimization method based on improved artificial fish school algorithm and the novel sigmoid-function-based control method with the classic sigmoid-function-based control method.Based on the sea trial results,contrast analysis for the three proposed control methods with the classic sigmoid-function-based method is conducted from different perspectives.