In view of the strong coupling and hard terminal constraints inherent in the cooperative rendezvous between two spacecrafts,this paper proposes a multi-impulse cooperative rendezvous control algorithm based on an improved model predictive static programming (MPSP) algorithm,denoted as cooperative impulsive model predictive static programming (CI-MPSP).A variational linearization technique is introduced to overcome the limitations of traditional MPSP methods in jointly optimizing the impulse vectors and timings of multiple spacecrafts.An analytical relationship among the terminal rendezvous deviations and the deviations in the impulse vectors,impulse application times and terminal time of two spacecrafts is established,thereby achieving the coordinated optimization of impulse control and timing.The proposed CI-MPSP algorithm is used to establish a unified solution framework that accommodates both cooperative and active-passive rendezvous modes,and it can be extended to multi-spacecraft rendezvous scenarios.Numerical simulation results demonstrate that,compared with the Gauss pseudospectral method (GPM) and the genetic algorithm (GA),the CI-MPSP algorithm significantly improves the computational efficiency while achieving the optimal system fuel consumption.Furthermore,when CI-MPSP method is applied in the cooperative rendezvous mode rather than in the active-passive mode,the total fuel consumption is reduced by approximately 5.3% while enhancing mission flexibility and robustness.

With the development of the low-altitude economy,the unmanned aerial vehicle (UAV) clusters have been widely used in urban traffic monitoring and logistics transportation.However,GNSS signals are susceptible to blockage and interference in complex urban environments,leading to a significant degradation in positioning accuracy and stability,which severely constrains the application of UAV clusters s.The China SatNet low earth orbit (LEO) satellite constellation can provide high-precision navigation services for UAV clusters; nevertheless,how to mitigate the time asynchrony and achieve the heterogeneous fusion of multi-type navigation sources remain the critical challenges in the positioning of UAV clusters.To address these issues,an information-geometry-aided cooperative positioning method based on LEO satellites for UAV clusters is proposed.Firstly,a unified information-geometric probabilistic model is constructed for LEO satellite navigation,inertial navigation,and ADS navigation,enabling the standardized parameter representation and uncertainty modeling across multi-source navigation data.Secondly,a pseudo-range cooperative transfer and distributed fusion framework is developed based on the factor graph theory to enhance the utilization efficiency of pseudo-range within UAV cluster.Finally,the adaptive weighted fusion of multi-source positioning results is achieved under the criterion of information-geometric optimality.Experimental results demonstrate that the proposed method significantly improves the average positioning accuracy.Under abrupt error scenarios,it converges rapidly within 3-4 computation epochs,with the error increment be controlled within 0.06 m,exhibiting superior robustness and stability.

To address the challenge of real-time path planning for unmanned aerial vehicles (UAVs) in multi-target reconnaissance and assessment missions,a fast path planning method based on improved ant colony optimization (IACO) algorithm is proposed.The multi-target assessment path planning problem is transformed into an open traveling salesman problem (OTSP).The differential deployment strategy of ant colony based on convex hull clustering is adopted to improve the diversity of initial optimization,and a dual pheromone update mechanism with positive and negative feedback is introduced to accelerate the convergence speed and avoid falling into local optimum.Taking into account the tactical application constraints of UAVs,a straight-line target passing strategy and an arc turning transition path generation method are designed to realize the rapid construction from the reference path to the flyable path.Simulated results show that the proposed algorithm has favorable optimization accuracy and robustness with fast iterative convergence,and its real-time computational capability meets the on-board application requirements of missile-borne platforms.This research provides a new insight for the intelligent navigation and mission planning of UAVs,and possesses significant engineering application value.

To address the problem of slow convergence of integrated navigation caused by the large initial misalignment angles for air defense guided projectiles,an inertial navigation/ground-based radar integrated navigation algorithm based on the launch-centered earth-fixed (LCEF) frame is proposed.A mechanization equation of strapdown inertial navigation in the LCEF frame as well as the attitude,velocity and position error equations are established.Based on this,a state equation and a measurement equation of the inertial navigation/ground-based radar integrated navigation are constructed.Taking the azimuth angle,elevation angle,radial range and radial velocity (AERV) provided by the ground-based radar as measurements,the nonlinearity of the measurement equation is handled via Taylor expansion,and the measurement matrix is derived.Monte Carlo simulations are carried out.The simulated results show that the proposed algorithm enables the integrated navigation attitude error to converge within 2° (1σ),the velocity error to converge within 1m/s (1σ) and the position error to converge within 10m (1σ) under an initial roll angle error of 100°.The proposed algorithm can achieve high-precision integrated navigation for air defense guided projectiles.The inertial navigation/radar integrated navigation algorithm corrects the inertial navigation errors only using the measurement information from ground-based radar.It is suitable for low-cost,high-dynamic air defense guided projectiles and provides navigation technical support for countering low-altitude,slow-moving and small unmanned aerial vehicles and their swarms.

To address the problem of rapid attack envelope estimation for glide vehicles,this paper proposes a dual-layer rapid attack envelope estimation method based on the hp-adaptive Radau pseudospectral method (hp-RPM) and a multilayer perceptron (MLP).On the basis of hp-RPM,an offline attack envelope computation method is established by formulating the terminal azimuth angle as the optimization objective,thereby simplifying the generation of the offline attack envelope dataset.Subsequently,an MLP surrogate model is trained using the high-fidelity offline dataset.The prediction accuracy and computational efficiency of the proposed model are verified through Monte Carlo simulation.Compared with the conventional method that repeatedly solves the nonlinear optimal control problems using direct methods,the proposed method significantly reduces the computational burden in the online stage.It achieves millisecond-level rapid estimation while maintaining high prediction accuracy,providing a new perspective for mission planning and real-time decision-making of glide vehicles.

The coupling influence laws of the explosion depth and the slope angle of pit wall on the detonation-driven behaviors of the shell body and soil medium in multi-layer soil media are investigated.The internal explosion behavior of the earth-penetrating propjectile in multi-layer soil is simulated using the SPH-FEM solid coupling algorithm,and the characteristic parameters of the fragment scattering are obtained through simulation.A calculation model for the fragment scattering is constructed,and the distribution law of the soil landing points is obtained.The results show that the scattering distance of soil medium is longer than that of the shell fragments,and it is the main factor determining the scattering field of the earth-penetrating projectile.The explosion depth has a significant nonlinear inhibitory effect on the scattering range.When the explosion depth exceeds 0.7 m,the conversion rate of the explosive energy to the kinetic energy of the surface soil decreases from 6.32% to less than 5%,and the flight distance of the soil in the tail direction converges to within 150 m.The slope angle of pit wall affects the scattering direction and landing point distribution.As the slope angle decreases,the scattering range increases first and then decreases,and the conversion rate of kinetic energy β decreases from 13.26% to 1.71%.The study reveals the influence laws of the explosion depth and the slope angle of pit wall on the scattering field of multi-layer soil,which can provide useful references for the structural design,material selection and safety protection of the earth-penetrating projectile.

To address the issue of low level of UAV chaff jamming self-defense decision-making intelligence,an intelligent decision-making algorithm based on deep reinforcement learning is proposed.Firstly,a scenario model for the unmanned aerial vehicles (UAVs) using chaff to jam the counter missile radar seekers is constructed.The UAV is regarded as an agent,and the state space,action space and reward function are designed according to the mission characteristics.Furthermore,a chaff jamming self-defense decision-making method with a data buffer function is proposed based on the proximal policy optimization (PPO) algorithm.The simulated results show that the proposed algorithm can converge to the optimal strategy within a limited number of training episodes,and significantly improve the survival success rate of UAVs.The data buffer function can also effectively accelerate the convergence speed of the algorithm.Compared with other deep reinforcement learning algorithm,the proposed algorithm exhibits distinct advantages and is more suitable for the mission scenario.

Guidance systems face significant difficulties when intercepting highly maneuverable targets.Deep reinforcement learning (DRL) is able to improve the guidance performance through its interactive trial-and-error capability and the suitably designed reward functions.To address the safety hazards caused by the lack of physical constraints in pure end-to-end intelligent controllers,this paper proposes a composite residual guidance law integrating DRL.The guidance law uses the augmented proportional navigation (APN) with a variable structure interacting multiple model (VSIMM),ensuring that the agent only outputs bounded residual commands for predictive correction.In addition,the first-order derivatives of target acceleration and the model probability residuals serve as the inputs to the neural networks.A multi-level reward function based on a zero-effort miss (ZEM) term and a saturation penalty is constructed to prevent the terminal divergence of engagements,and an adaptive curriculum learning process is designed circumvent catastrophic forgetting.Monte Carlo simulation findings in the presence of strong interference show that,in response to various types of unknown extreme maneuvers,The guidance law effectively suppresses the long-tailed error divergence and the overall mean and median miss distances.The results demonstrate that the guidance law achieves good performance in terms of ensuring guidance safety and attaining maximum interception precision.

The determination of internal blast loading is the fundamental basis and prerequisite for the related research on internal explosions. A numerical calculation model for the internal explosion flow field of a cylindrical charge within a confined structure is established to analyze the blast loading characteristics of cylindrical charges in the confined structures.The rationalities of the calculation model and parameters are verified by comparing with the experimental data.The internal explosion flow fields generated during the explosion of cylindrical charges at scaled distances ranging from 0.1 to 0.97m/kg^1/3 are calculated,and the evolution process of the internal explosion flow field is analyzed.The empirical formulas for predicting the peak overpressure and impulse (applicable range:scaled distance of 0.1-0.97m/kg^1/3) are derived based on the fitted calculation data.The distribution characteristics of blast loading on the inner wall of the structure are also analyzed.Furthermore,the internal explosion flow fieldsof cylindrical charges with different length-to-diameter ratios ranging from 0.6 to 2.0 are calculated. The influence of the charge’s length-to-diameter ratio on the peak overpressure and impulse of the internal blast loading is compared and analyzed.The evaluation formulas describing the effect of the length-to-diameter ratio on reflected peak overpressure are developed by fitting the calculation data.The results indicate that the variation in the length-to-diameter ratio of the cylindrical charge has a relatively minor impact on the impulse compared to its effect on the overpressure peak,with a maximum difference not exceeding 9%.The relevant findings can provide a reference for the research and design of structures resistant to internal explosions.

A two-stage preamble synchronization method is proposed for the direct-sequence spread-spectrum binary phase-shift keying (DSSS-BPSK) system in the millimeter-wave projectile-borne burst data link,where large Doppler frequency offset,low data rate,and limited preamble overhead make frame acquisition,frequency-offset estimation,and fine timing difficult to balance.The proposed preamble consists of a short repeated pseudo-noise (PN) segment and a long PN segment.The short repeated PN segment is used for coarse acquisition and coarse frequency-offset estimation,and the long PN segment is used for local fine timing after coarse frequency-offset compensation.The preamble parameters are determined according to the ambiguity-free frequency-offset range,repeated-correlation stability and long-PN processing gain.Simulations under the conditions of additive white Gaussian noise (AWGN),Rician fading,multipath propagation,time-varying Doppler,phase noise and fixed-point quantization show that the equivalent synchronization cost of the two-stage structure is about 1/124.88 of that of the single long-PN two-dimensional search structure under the same preamble overhead,while the stable acquisition and frequency-offset estimation performances are maintained under complex channel and radio-frequency impairments.The proposed method provides a reference for synchronization design of millimeter-wave projectile-borne data link under the large frequency offset,low data rate and resource-constrained conditions.

To address the difficulty of achieving the real-time and accurate radar jamming effect evaluation in adaptive jamming and cognitive electronic countermeasure closed-loop systems,this paper proposes a radar jamming effect evaluation method based bidirectional long short-term memory network (BiLSTM) and model-agnostic meta-learning (MAML).The proposed method takes the time-series sequences of radar parameters obtained through reconnaissance as input,and employs BiLSTM to extract the dynamic evolution characteristics of parameters such as pulse repetition interval,carrier frequency,and pulse width,thereby enabling radar operating mode recognition and jamming effect evaluation.A MAML-based meta-learning framework is introduced to overcome the scarcity of samples for emerging radar systems and the limited transfer adaptability of conventional deep learning models.Through meta-training,the framework learns the initialization parameters of model with strong generalization capability,enabling the evaluation model to rapidly adapt to new radar scenarios with only a small number of samples and a few gradient updates.A jamming effect evaluation procedure is then constructed for five typical radar operating modes,and the simulation experiments are carried out in guided radar confrontation scenarios.The experimental results show that the proposed method can effectively improve the accuracy of radar operating mode recognition and the rapid adaptation capability in new scenarios,providing effective support for the adaptive jamming decision-making and the closed-loop optimization of cognitive electronic countermeasures.