Sweep is a common engineering technique for enhancing the stall margin of fan blades. However, conventional sweep introduces a chordwise displacement that deviates the blade from its intended meridional design. Furthermore, its effectiveness is limited by practical constraints such as inter-row spacing and structural strength. To address these challenges, this study investigates the individual effects of leading-edge and throat-line sweep based on an analysis of NASA Rotor 67, revealing that both can independently enhance the stall margin. Building on this understanding, a novel decoupled leading edge/throat sweep design method is proposed. This method provides greater flexibility for introducing beneficial sweep effects within a prescribed meridional contour to enhance stall margin. A highly loaded single stage fan with a pressure ratio of 2.5 was designed using this methodology. Experimental results demonstrate that the fan stage achieves a peak efficiency of 88.1% and a stall margin of 11.8%. These results are supported by a detailed numerical analysis that elucidates the internal flow mechanisms within the single stage fan.
This paper investigates the challenges of nonlinear filtering for INS/CNS integrated navigation when confronted with measurement outliers. An improved robust cubature Kalman filter (ICKF) is introduced, incorporating a modified cubature point update framework (MUF) and a hypothesis test (HT). The ICKF adopts a three-tier architecture. In the detection tier, a dual-mode hypothesis test identifies and classifies anomalies as slow-varying or abrupt. In the estimation tier, a Karush-Kuhn-Tucker (KKT)-constrained Maximum Likelihood (ML) estimator converts detection results into adaptive noise covariance. In the update tier, an MUF uses the adapted covariance to construct posterior cubature points without resampling. As the noise covariance updates dynamically, the prediction cubature points error translates into the posterior cubature points error, improving resilience to measurement outliers. Unlike existing methods that passively attenuate outliers or apply hypothesis testing in a decoupled manner, the ICKF establishes a tight coupling between a dual-mode HT and an MUF, where dynamically estimated noise covariance drives posterior cubature point construction without resampling. Its effectiveness is validated through simulations and semi-physical experiments on the INS/CNS navigation system.
Interrupted Sampling Repeater Jamming (ISRJ) poses a severe challenge to aerospace radar target detection because it is generated by intermittently sampling and retransmitting the radar waveform, resulting in high similarity to genuine target echoes and strong time-frequency overlap. To address phase-coherent target–jamming separation under such conditions, this paper proposes a Complex-Spectral Residual Temporal Convolutional Network (CR-TCN). The complex STFT spectrum is represented by its real and imaginary components and jointly reconstructed so that phase-sensitive information required for subsequent radar processing is retained during separation. To exploit the temporal structure of ISRJ, a dilated TCN is employed to capture dependencies over multiple temporal scales associated with intermittent sampling and repeated forwarding. A residual spectral formulation is further adopted to estimate the jamming-induced deviation from the mixed spectrum rather than directly reconstructing the complete target spectrum. In addition, a hierarchical Global Attention–Local Attention (GA-LA) module combines long-range contextual modeling with local feature refinement, enabling the network to characterize repeated forwarding patterns and transient interrupted slices. Experimental results demonstrate that CR-TCN effectively suppresses ISRJ while preserving target waveform characteristics under a wide range of jamming conditions. The recovered echoes retain sufficient signal fidelity for subsequent pulse compression and coherent integration, indicating the potential of the proposed framework for radar target–jamming separation in complex electromagnetic environments.
This paper presents a coverage barrier residual learning (CBRL) framework for 3D cooperative interception of a maneuvering target by multiple interceptors. The proposed guidance scheme estimates terminal target acceleration with a first-order backward finite difference from sequential target velocity observations and uses reinforcement learning as a bounded residual optimizer around a cooperative proportional navigation baseline. By combining a composite time-to-go predictor with a target-velocity-frame zero-effort-miss projection, the framework establishes a low-dimensional physical representation where coverage reduces to a 2D regulation task around dynamically assigned defense points. We reformulate the zeroing control barrier function inequality as a dense soft funnel score rather than an online hard constraint, which shapes policy training without requiring online quadratic programming or brittle invariance claims, and design a bounded convergence reward that increases with interception progress toward a plateau. This structure delivers improved terminal guidance accuracy while preserving the baseline’s interpretability and bounded command envelope.
Failed spacecraft often exhibit complex tumbling motions, which pose severe threats to the safety of on-orbit servicing, thus detumbling constitutes a critical first step. A contactless detumbling approach utilizing the plasma plume generated by Hall thrusters offers promising potential owing to its operational safety and low energy consumption. However, the traditional locally optimal planning and disturbance-sensitive control strategies significantly limit its efficiency. Therefore, this paper investigates the planning and control problem in the plasma plume detumbling manipulation. Firstly, a lightweight system dynamics model incorporating plasma plume impact forces is established, which avoids complex high dimensional numerical computations. By dividing the detumbling cycle into operational and waiting periods, a Global Optimal Energy Detumbling Planning (GOEDP) method is proposed, which coordinates the suppression of both spin and nutation kinetic energy throughout the entire detumbling process, with its underlying principle further elucidated through a graphical approach. A rigorous planning formulation is then established to dynamically optimize the plume aiming points, thereby improving the average operational efficiency. To address actuator saturation and environmental disturbances, an adaptive prescribed performance framework is designed to improve the nonsingular terminal sliding mode controller, enhancing the robustness of the service spacecraft. Simulation results demonstrate that the proposed GOEDP and control method can stabilize a several-hundred-kilogram target using plasma plume within 1.5 h, achieving an 18.2 % efficiency improvement over the traditional approach.
