Abstracts:A deep learning-based source number estimation method is presented in this article, where the deep generative adversarial network (GAN) combined with semi-supervised learning is applied, modifying the classifier in an adversarial way. Different from the traditional eigenvalue-based methods, the linear shrinkage coefficient established under the general asymptotic theory framework is utilized as the input feature of the network, which produces more distinct classification features, and therefore achieves satisfactory classification performance under conditions of small number of labels and samples, and low signal-to-noise ratios. It is shown that 30$\%$ label rate is able to achieve a performance close to fully supervised learning.

Abstracts:In order to reduce the loss of maneuverability of hypersonic flight vehicle (HFV) when the angle of attack (AOA) is constrained, a dynamic protection control method based on command limiting is proposed in this article. First, a protection system is constructed based on the closed-loop dynamic approximation model, and the dynamic command bound is obtained online and fed back into the control system. Then, an adaptive AOA constraint controller based on command limiting is designed in the backstepping framework. Taking the advantage of bound estimation, the proposed control method only needs one adaptive law to deal with the lumped uncertainty. To avoid the violation of AOA limitation, the virtual command designed in the control system will be constrained and modified by using the bound obtained from the protection system, and a compensation signal is designed to attenuate the effect of command constraint. Rather than setting a conservative and constant value on the AOA command bound, the proposed constraint control scheme predicts the command bound dynamically and modifies the actual command online, which brings the advantage that the remaining AOA in a given interval can be fully utilized. Besides, the proposed controller does not have strict limitations on the initial state value. Even if the AOA exceeds the bound, the proposed algorithm remains effective. Finally, simulation results demonstrate the effectiveness of the proposed method.

Abstracts:This article addresses the challenge of velocity control in fixed canard decoupled dual-spin projectiles, focusing on the nonlinearities and parameter uncertainties of the system. A motor-generator test bench and a dynamical model are developed to emulate the dual-spin system accurately, enabling thorough experimentation and analysis. Robust control strategies, specifically sliding mode control and super-twisting control, are implemented and compared against traditional linear proportional-integral (PI), proportional-integral-derivative (PID), and nonlinear dynamic inversion controllers. Performance evaluations are conducted through two experiments: first, tracking a step reference speed with a fixed driving speed, and second, replicating real-world dual-spin behavior by tracking a variable speed trajectory with varying driving speeds. Results demonstrate that the super-twisting controller and PID controller outperform other strategies, exhibiting minimal reaching time and lower steady-state error during both step reference and disturbance tests.

Abstracts:Stepped-frequency chirp radar achieves range high resolution through wideband synthesis, yet it harbors systemic errors. These errors coupled with the motion errors of high-speed space target, render error calibration more challenging and compromise the quality of imaging. To tackle this problem, this article proposes the coupled error self-calibration method for high-speed space target imaging in stepped-frequency based on minimum entropy. First, a parameterized model of echo signals incorporating complex coupled errors is established. This model not only takes into account the errors introduced by the amplitude–phase response characteristics of the stepped-frequency radar system, but also considers errors arising from the motion of high-speed targets. Then, analytical relationships between entropy and errors after range pulse compression of subband data, after 2-D imaging of subband data, and after high-resolution synthesis of all subband images are constructed. This stepwise processing strategy decomposed complex errors into three distinct components. Subsequently, employing an adaptive matrix estimation method to separately estimate and calibrate the three decoupled error components ensures a gradual improvement in imaging quality. Finally, the effectiveness of the proposed method is verified through computer simulation and a real experiment.

Abstracts:In this article, a robust singularity-free fixed-time satellite attitude control in a geometric control framework is proposed. The satellite attitude is represented by the rotation matrices defined on the special orthogonal matrix group $\text {SO}(3)$ to achieve unique, global, and kinematic singularity-free attitude representation. A nonsingular fixed-time sliding manifold for the attitude control system evolving on the Lie group $\text {SO}(3)\times \mathbb {R}^{3}$ is proposed. A complete mathematical analysis is carried out to prove and establish that the proposed sliding manifold is a Lie subgroup of $\text {SO}(3)\times \mathbb {R}^{3}$. Along this manifold, the reduced-order dynamics is almost globally fixed-time attractive to a residual set containing the stable equilibrium point. Based on the proposed sliding manifold, a nonsingular geometric fixed-time sliding mode control (NGFTSMC) law is proposed for the satellite attitude control in the presence of external disturbance torques. It is proved through five propositions that the NGFTSMC facilitates an attitude maneuver on the phase space $\text {SO}(3)\times \mathbb {R}^{3}$, excluding a set of measure zero. This guarantees almost global, unwinding-free, and singularity-free closed-loop attitude dynamics. Simulation results are presented and compared with a quaternion-based fixed-time sliding mode control to show superiority of the NGFTSMC in terms of the time of convergence, unwinding, and control effort.

Abstracts:As a tradeoff between the conventional phased-array radar and multiple-input multiple-output radar, the phased array multiple-input multiple-output (PA-MIMO) radar has attracted widespread attention. To better manage the coherent processing gain and diversity gain within the system, this article introduces a transceiver subarray configuration strategy. Its essence lies in adjusting the ratio of these two gains through subarray configuration. Initially, we develop a likelihood ratio detector that incorporates channel reciprocity and pulse accumulation, while accounting for diversity gain from subarray configurations. This subsequently leads to the derivation of an implicit radar effective range expression. Leveraging this, we formulate a quality of service-based subarray configuration optimization model, which hinges on the number of elements per subarray. The utility function of the model strikes a balance between fulfilling the task objective and possessing a certain level of low probability of intercept capability. To address this problem, we first design a relaxation and fine-tuning process, and propose an efficient elite social learning-based particle swarm optimization algorithm to find an approximate optimal solution. This algorithm circumvents local optima and inefficient search by emulating the strong uncertainty of particle state superposition. Simulation outcomes underscore the efficacy of our proposed PA-MIMO radar subarray configuration strategy and the enhanced particle swarm optimization algorithm.

Abstracts:This article proposes a prescribed-time chattering-free sliding mode guidance law with terminal angle constraint based on periodic delayed feedback. First, a prescribed-time sliding mode guidance law is presented based on a periodic delayed sliding mode surface, which can ensure the prescribed-time convergence of both the line-of-sight angular rate and the line-of-sight angle. Second, a prescribed-time extended state observer is proposed to acquire the precise disturbance estimation within the prescribed settling time. In contrast to other finite-time or fixed-time disturbance observer, the convergence time of the proposed observer is concise and easy for tuning. Finally, by incorporating the estimated disturbance into the original prescribed-time sliding mode guidance law, a composite guidance law with prescribed-time convergence is developed, where chattering is completely eliminated. In particular, the convergence time of the proposed prescribed-time extended state observer and prescribed-time sliding mode guidance law can be arbitrarily assigned and prescribed in advance. A series of numerical simulations are carried out to verify the efficacy of the proposed guidance law.

Abstracts:This article focuses on the topic of joint direction of arrival (DOA), source association, and attenuation coefficient estimation under multipath environment. Most existing methods adopt the sequential three-phase estimation, resulting in the nuisance dependency between the estimation accuracy of the current phase and the previous phase. Besides, they also require some accurate prior information, including the accurate DOA initialization, and the number of incoherent sources and spatial paths, which is unrealistic in practice. To solve this problem, the joint localization and source association sparse Bayesian learning (JLSA-SBL) algorithm is proposed to integrate the source association process, DOA, and attenuation coefficient estimation into a unified parameter estimation framework. The proposed method exploits the underlying sparsity and coherent structure of the incident signals to achieve more accurate joint parameter estimation. Compared to the previous methods, JLSA-SBL can directly estimate the latent multipath propagation parameters even in the absence of prior information. Besides, the JLSA-SBL also has superior performance in distinguishing the closely spaced multipath signals belonging to different sources. Numerical simulation experiments have been performed to demonstrate the superior performance of the proposed method.

Abstracts:This article proposes an intelligent decision-making approach for a multibranch contingency return trajectory scheme, which is intended to satisfy the contingency requirement during the circumlunar flight phase in the manned lunar missions. First, based on the knowledge of orbital dynamics, a production rule base is constructed with the interval form. An expert system of contingency return trajectory is further designed to assist in the preliminary determination of contingency return schemes. Second, by adopting the fully connected neural network, a contingency return trajectory calculation model is established based on deep learning. Finally, combining the expert system and the calculation model, an intelligent decision-making approach is proposed to achieve rapid decision making of multibranch contingency return trajectories. The simulation shows that the calculation model can accurately generate the contingency return trajectory and has higher calculation efficiency than the traditional method. By using the proposed intelligent decision-making approach, a decision can be made quickly to determine a contingency return trajectory scheme and specific trajectory parameters can be obtained. The research results can provide an effective tool and important references for the decision making of a contingency return trajectory scheme in the future manned lunar missions.

Abstracts:This study investigates a dynamic event-triggered saturation nonlinear observer dedicated to addressing the challenge of event-based nonlinear observation in the presence of measurement noise. We propose an architecture incorporating dynamic event-triggered saturation injection to enhance system robustness against measurement noise while optimizing communication resource utilization by event-based sampling. As a prerequisite for this injection study, we rigorously prove the exponential convergence of system observation error under ideal conditions without measurement noise. Building upon this foundation, we explore the upper bound of convergence for system observation error in the presence of measurement errors. The proposed design is applied to in-orbit drag-free systems, serving as a key technology for significant scientific missions such as gravity field exploration, verification experiments on general relativity in space, and space gravitational wave detection. Drag-free systems face constraints on space transmission resources and require high mission performance with minimal sensor measurement noise, and our proposed method effectively addresses these challenges encountered in studying system dynamics. In addition, we provide a numerical example to illustrate the feasibility and performance of our design by comparing it with a standard observer.