Moving Horizon Estimation And Application On The Cooperation Localization Of Multiple Unmanned Underwater Vehicles

With the increasing scale of the complex ocean exploration and development projects,a single Unmanned Underwater Vehicle(UUV)more and more difficult to meet people's needs in recent years,more UUV cooperation technology has received more and more attention,including cooperation localization,multi UUVs technology as cooperation has become a hot issue.The main purpose of multiple co-located in UUV group UUV ensure positioning accuracy on the basis of positioning costs as low as possible.Receding Moving Horizon Estimation(MHE)is a superior performance of the rolling state optimal estimation method in nonlinear system,has its unique advantages Uncertain measurement system and state restraint system is becoming a nonlinear uncertain restraint system state estimation hotspot.MHE paper focuses on multiple UUV expand co-localization of research.The main work includes:We study the whole question of the optimal estimation of the probability of information significance,in order to avoid the whole problem in data processing amount of information over time,tired by the lack of a cost function is derived by introducing reach when receding horizon estimation theory and composition in the form of,in Karush-Kuhn-Tucker(KKT)theorem,based on the MHE state estimation problem is transformed into quadratic optimization problem,the calculation method of MHE quadratic programming problems.From the perspective of the numerical calculation is derived MHE and Kalman filter equivalence,study results show that:the linear no constraint in dealing with the problem,when estimating the length of the time domain,MHE and Kalman filter are equivalent.MHE studied state estimation stability problems,and the use of Cauchy-Schwarz inequality in detail proved MHE state estimation error boundedness and convergence.Analysis of the co-location of multiple UUV information processing program,clock synchronization method and acoustic communication model,combined with co-location of the sensor characteristics,in the Cartesian coordinates use of the system Total Observability Matrix(TOM)main UUV from collaborative multi-location system observability in-depth analysis.Derived based on a Central Extend Kalman Filter(CEKF)multi UUV co-location method to analyze the shortcomings of this approach,according to whether the relative measurement information updates will be co-located in two phases is proposed respectively,as well as the use of UKF and MHE which based on UKF two optimization methods getting state estimation,real ship tests to verify the effectiveness of the design.Proposing the use of the rolling fields estimated measurement uncertainty solve problems more colocated UUV acoustic communication system due to the delay,packet loss.The establishment of underwater acoustic communication delay,packet loss process mathematical model,a detailed demonstration of the feasibility of the UUV end only with a limited number,in accordance with certain rules for processing acoustic communication systems to achieve optimum measurement equivalent estimated,on the basis of on,respectively,the Delayed Extend Kalman Filter(DEKF)and Delayed Moving Horizon Estimation(DMHE)to achieve positioning errors due to acoustic communication caused amended.By changing the sequence order of unknowns,the KKT matrix becomes a sparse band structure,proposed a special structure LU decomposition iterative realization MHE problem solving,simulation show that this method reduces the calculation time.We study the attitude estimation method based on the research of the UUV's Micro Inertial Measurement Unit(MIMU)and Three Axis Magnetometer(TAM)combined measurement system.QKF proposed state estimation method using posture conversion relation sampling period to achieve linearized state space model,but this approach in dealing with the quaternion normalization constraint will introduce a new estimation error.Then designed based Cubature point in the feasible region constraint volume Constraint Cubature Kalman filter(CCKF)method implementation quaternion normalization constraint state estimation,but this method is relatively large amount of calculation.Based on the final design of the quaternion normalization constraint constraint explicit expression of Constraint Moving Horizon Estimation(CMHE)method,simulation results verify the feasibility of this approach.Research based on KKT block matrix TQ domain CMHE.First,according to the nature of the state of constraint will be divided into two types of conventional and threshold constraint,the special structure of the matrix decomposition decomposition on the basis of TQ,are designed to increase or decrease each iteration affect certain constraints on the KKT conditions and the block matrix equation processing method so that each iteration requires only KKT block matrix rows or columns specific location for processing you can achieve positive changes in response to the state of constraint,the simulation proved that this method greatly reduces the computational complexity,improved CMHE UUV based collaborative computing efficiency multi-location.