Research On Trajectory Tracking Filters Of Near-Space Hypersonic Glide Vehicle

In recent years,Near-Space Hypersonic Glide Vehicle?NSGHGV?,represented by HTV-2,has gradually become a research hotspot in academic and engineering fields due to its fast flight speed,strong maneuvering ability and diverse maneuvering styles.As an offensive weapon,NSHGV has great development potential and strong penetration ability,representing the future development direction of long-range strategic weapons.Therefore,there is great significance to carry out relevant technical research on NSHGV interception for safeguarding national security.NSHGV trajectory tracking method is the core technology of anti-NSHGV,and it is an important guarantee and prerequisite for accurate prediction and interception of NSHGV.This dissertation takes NSHGV as the object,and conducts an in-depth research on its trajectory tracking method.The main contents are listed as follows:According to the motion characteristics and observation conditions of NSHGV in the boost phase,target motion models such as gravity turning model and space-based infrared early warning satellite?SBIRS?measurement model are established,and then the extended Kalman filtering?EKF?algorithm is given.Since the tracking system is nonlinear,the EKF has a large truncation error.Therefore,the dissertation establish the general framework of deterministic sampling filtering method based on the idea of deterministic sampling approximate Gaussian distribution,and then the unscented Kalman filter and cubature Kalman filter are proposed based on the unscented transformation and spherical-radial cubature rules.After that,the trajectory tracking mathematical simulation in the boost stage is conducted to compare and analyze the tracking accuracy of different target motion models and various nonlinear filters.Due to the channel congestion or interference factors,the measurements from SBIRS may arrive at the ground data processing center with random delay.In order to solve this problem,the dissertation gets rid of the traditional idea which based on the state propagation,and rewrites the measurement equation with the Bernoulli random variables.Then the randomly delayed system model is established.After that,the dissertation deduces the relation equations between the ideal measurements,the practical measurements and the correlation covariances.Considering the high nonlinear approximation accuracy of the five-order spherical-radial cubature rule,this dissertation uses this rule to deduce the Randomly Delayed High-order Cubature Kalman Filtering?RD-HCKF?algorithm under the criterion of minimum mean square error,and the RD-HCKF can solve the problem of delay filtering when the state model is inaccurate and the measurement delay time is unknown.The glide stage is the main stage of NSHGV,and its movement mode mainly includes equilibrium glide and skip glide.Based on the observation conditions of the glide phase,a ground-based radar measurement model is established.Considering the switch of the fling mode,this dissertation makes an in-depth research of the Interactive Multiple Models?IMM?algorithm,and uses this algorithm to design the glide tracking filter.In addition,the corresponded mathematical simulation is carried out to verify the effectiveness of the tracking algorithm and the target motion models in the glide phase.NSHGV can use aerodynamic force to adjust the motion parameters such as the skipping period in the process of skip gliding,so its mode space is continuous.The traditional multi-models algorithm has some problems in the process of model set design,such as inaccurate model coverage or too large model set.Therefore,the dissertation proposed a Hybrid Grid Multiple Models?HGMM?estimation algorithm.The HGMM establishes the coarse model set based on the prior information,adaptively designs the fine model set based on the moment matching method with the estimation results from both model sets.After that,the overall state estimation can be obtained through the weighted fusion of the estimation results from the coarse model set and fine model set.The simulation results indicate that HGMM has higher estimation accuracy than the single model algorithm and the traditional multi-models algorithm.Under the influence of target reflection at different positions and the electronic circuit performance,the measurement noise of radar usually exhibits non-Gaussian characteristics,mainly glint noise or impulse noise.As for the glint noise case,the Huber filtering which is a combined ₁l and l₂ norm estimator is researched.The measurement update which under the framework of Kalman filter is converted into the linear regression,and then the high-order cubature Huber-based filtering method is deduced.As for the impulse noise case,the measurement update of high-order cubature Kalman filter is transformed into the solution of obtaining the maximum value of the cost function based on the Maximum Correntropy Criterion?MCC?,and then the maximum correntropy high-degree cubature filter?MCHCF?is deduced.Moreover,this dissertation also applies these two robust high order cubature filters to the trajectory tracking of equilibrium glide target.The mathematical simulation verifies the advantages of these two filtering algorithms in robustness and accuracy under the case of non-Gaussian noise.