Static random-access memory (SRAM) -based FPGAs are widely used due to their low development cost, short design cycles, and broad adaptability. In applications such as aerospace, military, high-energy nuclear physics, and terrestrial communication systems, these devices face additional requirements for radiation resistance. Therefore, studying soft error protection methods for SRAM-based FPGAs to enhance system reliability has become urgent. This paper categorizes radiation hardening techniques into three levels: hardware-level techniques, system-level techniques, and software-level techniques. It also reviews research in these categories, including hardware hardening, triple modular redundancy (TMR), scrubbing techniques, and algorithm optimization methods. The advantages and disadvantages of each technique are analyzed, and the feasibility and necessity of using in-place fault tolerance as a supplementary technique to further strengthen the TMR + scrubbing design are discussed. Finally, the paper summarizes and provides an outlook on the application of machine learning techniques in this field and proposes a software fault tolerance method based on reinforcement learning. This research aims to offer technical support for the further enhancement of FPGA software radiation resistance methods and provide a reference for researchers in radiation hardening and related fields.

The instrument landing system (ILS) protection area is an important barrier to ensure the safe landing of aircraft. In view of the insufficiency in the theoretical study on the delineation of the protection area of the localizer (LOC) in the ILS, the LOC protection area is systematically studied for the first time using the physical optics (PO) method, in accordance with the requirements of the Annex 10 of the Convention on International Civil Aviation. First, the LOC radiation field is divided into far-field and near-field regions, and the LOC near-field radiation pattern is derived and solved to avoid errors caused by applying far-field conditions when obstacles are actually in the near-field. Then, based on the vertical surface characteristics of aircraft waiting around the runway, the surfaces are discretized into small facets. The electromagnetic scattering field caused by these aircraft, considered as obstacles in different positions and orientations, is calculated using the PO method. The LOC protection area is delineated according to the impact of these scattering fields on the difference in depth modulation (DDM) of the signal received by approaching aircraft. Finally, simulation experiments are conducted using 20-element and 24-element LOC arrays, with the A380, B737-8, and B787-8 serving as the obstacles. The simulation results show high consistency with the protection area requirements specified in Annex 10 of the Convention on International Civil Aviation, with the maximum deviation in the sensitive area being about 100 m and in the critical area about 50 m. Compared with simulations performed using the professional software ATOLL, the overall trend of the protection area is basically the same. The experimental results validate the correctness of the proposed conceptual approach, theoretical modeling, and technical processing, and can provide an important theoretical basis for LOC protection area delineation.

High-speed on/off valves (HSVs) are widely used in the aerospace industry. The large eddy currents make it difficult to improve their response time, and the relationship between response time and various structural parameters such as armature diameter and spring preload is complex. To address this issue, the influence of eddy currents on the dynamic characteristics of HSVs is analyzed. Based on this, a multi-slit armature structure is designed to reduce eddy current loss, which can accelerate the opening and closing of the electro-mechanical transducer in HSVs. The effects of the interaction between armature diameter, coil turns, spring preload, and spring stiffness on the opening and closing characteristics of the electro-mechanical transducer are then analyzed. Subsequently, the correlation between each factor and the opening and closing times is quantified using grey correlation analysis. The results show that the multi-slit armature structure can reduce eddy current loss by 50.07% and shorten the opening and closing time by 15%. Moreover, it is found that spring preload has the highest correlation with closing time, while armature diameter is most correlated with reset time. These findings can serve as a basis for optimizing the structure of the electro-mechanical transducer and further improving the dynamic characteristics of HSVs.

In this paper, a finite-time tracking control scheme based on the multi-dimensional Taylor network (MTN) is proposed for path tracking control of unmanned vehicles with model uncertainty and measurement noise. First, the MTN model is used to characterize the uncertainties of the unmanned vehicle model, and the improved back propagation (BP) algorithm is adopted as its learning algorithm. Second, an adaptive MTN filter is designed to suppress the measurement noise. MTN serves as the filter, and the least mean square (LMS) algorithm is employed as its learning algorithm. Then, the MTN finite-time controller is designed for precise path tracking control of the unmanned vehicle, which can track the reference trajectory quickly and accurately. Based on finite-time control theory, the convergence of the system is proved. Finally, unmanned vehicle simulation experiments are conducted to verify the effectiveness of the proposed method.

Addressing the issue of low attractiveness and inefficient airport transit connections due to current schedule optimizations that do not fully consider passengers’ selection preferences, an investigation into flight schedule optimization influenced by passengers’ transit duration and transit service selection preferences was conducted. Starting from the actual selection preferences of transfer passengers, a selection behavior experiment was conducted to collect data, and a conditional Logit model was constructed to analyze the flight characteristics that influence passengers’ choices for transit flights. Based on the findings of the selection preference analysis, a passenger attractiveness parameter for transit flights was defined. Subsequently, a flight schedule optimization model was developed with the objectives of maximizing the attractiveness of transit flights, maximizing the number of connectable flight pairs, and minimizing the total flight schedule adjustment. By comparing the solution effectiveness of particle swarm, non-dominated sorting genetic algorithm Ⅱ（NSGA-Ⅱ）, and NSGA-Ⅲ, a flight schedule optimization scheme that considers the selection preferences of transfer passengers was proposed. The results indicate that fares, transit duration, and transit facilitation services are the primary factors affecting passengers’ selection behavior. The results demonstrate that the proposed optimized flight schedule is significantly more attractive to transfer passengers, with the attractiveness of transit flights increasing by 391.22%, the number of connectable flight pairs increasing by 31.28%, and the airport’s transit capacity being effectively enhanced. Additionally, 26.52% of flights were adjusted, with an average adjustment of 12.35 minutes, which falls within the airlines’ acceptable range. This study offers new perspectives and methods for flight schedule optimization, contributing to the enhancement of Chinese hub airports’ transit capacity and facilitating passengers’ travel experiences.

Due to hardware limitations, infrared images typically suffer from low resolution and blurred details when captured. Although visible light images can effectively guide the super-resolution reconstruction of infrared images, differences in image detail caused by their distinct imaging principles often result in issues such as blurring and ghosting during reconstruction. This paper proposes a super-resolution reconstruction network for infrared images based on visible image guidance and recursive fusion. In this network, a flowing Fourier residual module is designed to extract different frequency information from infrared and visible images using modules at different depths, enabling each module to focus on the appropriate frequency information. Simultaneously, a hybrid attention module is employed to capture detailed information in multimodal images from both channel and spatial perspectives, and to fuse it in a complementary manner. Based on this, a global recursive fusion branch is designed to model the correlations across multiple feature layers and adaptively fuse them, thus generating clearer high-resolution infrared images. Experimental results show that compared with comparison methods like Deep-ISTA and PAG-SR, the proposed method demonstrates better level in objective evaluation indicators. In terms of subjective visual comparison, the images reconstructed by this method exhibit clearer textures, fewer artifacts, and better object discrimination in complex environments.

Aiming at the problem that the aircraft is easy to deviate from the runway when landing and taxiing under slippery crosswind conditions, this paper takes the large civil aircraft A320 as the research object and uses Simulink to establish a dynamic model of aircraft landing and taxiing, including four components: wheels, landing gear, aircraft body, and pilot. Based on this model, a human-machine closed-loop simulation of aircraft taxiing under wet runway and crosswind conditions is conducted. An analysis is carried out on the landing roll attitude and yaw distance of the aircraft under different water film thicknesses, unbalanced friction, and crosswind intensities. The following conclusions are drawn: the water film thickness has a significant impact on the lateral controllability of the aircraft. A larger water thickness leads to increased peak and final values of the yaw angle, thus significantly increasing both the yaw distance and roll distance. A water film thickness of 10 mm or more can cause roll oscillations during the later stages of landing and taxiing, which greatly affects the aircraft’s lateral stability. Pavement friction imbalance has a considerable impact on runway excursion risk, with the maximum yaw distance reaching 24.75 m. The aircraft has already deviated from the runway, severely compromising landing and taxiing safety. The stronger the crosswind, the less effective the pilot’s attitude control becomes. When the crosswind speed increases to 13.9 m/s, the roll angle exceeds the safety threshold of 6° at 7.7 seconds into the taxiing phase. At this point, the aircraft’s engine or wingtip may have touched the ground, and the roll angle peaks at 6.57°. However, under this crosswind intensity, the yaw distance does not exceed half the width of the runway. This indicates that crosswind intensity has a significant effect on roll angle, far exceeding its impact on yaw distance. In wet runway management, strict control should be applied to water film thickness, with takeoff and landing prohibited at or above 13 mm, and crosswind speeds should be kept below 13.9 m/s.

A residual multi-scale dual-stream network model based on silhouette differential fusion flow and silhouette flow is proposed to address the challenge of convolutional neural networks being unable to fully capture and utilize spatiotemporal information in gait recognition of low-quality infrared images. Firstly, a fine-grained segmentation strategy combining Faster R-CNN and Deeplab v3+ algorithms is applied at the model’s input to extract silhouettes, thus reducing the impact of noise and preventing feature loss. Secondly, a gait silhouette differential fusion module is added to the branch network of the model to capture the differences and changes between adjacent silhouette frames. Then, residual units and multi-scale feature fusion techniques are employed in the feature extraction section of the model to deepen the network layers and extract spatiotemporal information at different granularities. Finally, the multi-scale pyramid mapping module is utilized to further enhance the model’s ability to represent both local and global features. Experimental results from four different walking conditions on the CASIA-C dataset show that the average gait recognition rate of the proposed method is 98.85%, outperforming current mainstream methods.

To solve the problem of position tracking control of the surface permanent magnet synchronous motor (PMSM) servo system, a compound control strategy based on non-singular fast terminal sliding mode control (NFTSMC) using an improved fast super-twisting algorithm (STA) and adaptive extend sliding mode disturbance observer (AESMDO) is proposed. Firstly, the mathematical model of surface PMSM with disturbance is established. Secondly, a fast super-twisting nonsingular fast terminal sliding mode controller is designed to prevent singularity and chattering. The improved fast STA is used as the switching control law of the approach phase, which has a faster approach speed than the traditional second-order sliding mode. In order to improve the system’s ability to resist disturbance, finally, an AESMDO is designed to estimate the disturbance and compensate by the feedforward method. The stability and convergence of the system in finite time are proved by Lyapunov’s theorem, and the experiment is carried out. The outcomes demonstrate how well the developed controller can monitor and regulate the system’s specified value, effectively remove the buffeting phenomenon, and increase the system’s resilience.

To enhance the channel structure of air-fuel heat exchangers, numerical studies of the heat transfer of supercritical RP-3 aviation kerosene in twisted spiral tubes were carried out. The study focused on examining the impact of petal number, spiral pitch, and kerosene pressure on heat transmission. The circumferential distribution characteristic and mechanism of wall temperature were revealed through the distributions of channel-section temperature, bulk flow velocity, secondary flow velocity, and turbulent kinetic energy. Axial evolutions of swirl intensity and secondary flow intensity were used to investigate the swirl flow effect on heat transfer. Compared with circular tubes, the enhanced heat transfer effect of twisted spiral tubes was characterized by the comprehensive heat transfer coefficient performance evaluation criteria(PEC). The heat transfer correlations for twisted spiral tubes with three, four, and five lobes were proposed. The findings demonstrated that two asymmetric wall temperature waveforms emerge within the petals when the wall temperature approaches the pseudo-critical temperature, indicating heat transfer deterioration brought on by the gas-like coating with significant heat transfer resistance. Increasing kerosene pressure, increasing spiral pitch, and reducing the petal number all weaken the swirl flow effect. The PEC ranges from 1.05 to 1.65, and the higher the kerosene pressure, the smaller the spiral pitch, and the fewer petals, the more favorable it is to enhance the comprehensive heat transfer effect of twisted spiral tubes.

Abstract: In this paper, a coordinated control method for transition flight position and attitude based on appointed-time prescribed performance control（ATPPC） is proposed for the tilt transition flight control problem of quad tilt-rotor UAVs. Firstly, establish a six degree of freedom nonlinear motion/dynamics model for a quad tilt-rotor UAV and complete affine nonlinear processing. Second, a three-dimensional safe transition corridor of "nacelle angle flight speed angle of attack" is established to ensure the transition safety of the UAV through reasonable matching of flight speed, nacelle inclination angle, and angle of attack. This is done by adding the angle of attack as the third dimension to the traditional two-dimensional transition corridor of "nacelle angle flight speed" in order to accurately present the relationship between the tilting process and aerodynamic characteristics with these characteristics. Then, in response to the efficient allocation problem of the fixed wing/rotor heterogeneous redundant control during the tilt transition process, a sequential quadratic programming algorithm is used to transform the control allocation problem into a nonlinear optimization problem with multiple constraints. By solving the optimal solution of the optimization problem, accurate mapping of control force and torque to the rotor speed, rudder surface, and other executing mechanisms is achieved. On this basis, a position and attitude coordination control method for quad tilt-rotor UAVs based on appointed-time prescribed performance control is proposed, and the stability of the closed-loop system is analyzed. Finally, a safety profile is designed within the three-dimensional transition corridor as a position and attitude coordination control command. The proposed control method is compared with the nonlinear dynamic inverse method in simulation experiments. The findings shown that the suggested approach provides benefits over conventional techniques in addition to being practical for attaining transition flight control of quad tilt-rotor UAVs.

Addressing the issues of insufficient time-series features and incomplete fault information in fault diagnosis methods for electromechanical actuators (EMAs) based on traditional machine learning and deep learning, a fault diagnosis method for EMAs based on multi-source signal fusion with gated recurrent unit (GRU) and an improved attention mechanism is proposed. First, the collected signals from different sensors are divided into separate channels, and the time-series features of each channel’s signal are extracted using GRU. The self-attention mechanism is then introduced to further distinguish the important relationships between different time points of the signal. A multi-channel attention mechanism is employed to adaptively fuse the features from different channels. Finally, fault diagnosis is achieved through the classifier. Experimental results based on the test rig dataset show that the diagnostic accuracy improves by 10% compared to the single-sensor model and by 5.2% compared to the model without the attention mechanism. Compared to classical machine learning, deep learning and recent improvements in deep learning-based algorithms from the past two years, the diagnostic accuracy of the proposed model exceeds 98.5%, demonstrating optimal diagnostic performance.

To address the problem of insufficient multi-scale feature representation in liver tumor detection, we propose a liver tumor CT image detection method that integrates reparameterized convolution, weighted bidirectional feature pyramid, and attention mechanism. Firstly, data augmentation is used to improve the problem of small sample size and enhance the generalization ability of the model. Secondly, to enhance the ability to extract multi-scale features, the weighted bidirectional feature pyramid network is utilized to merge the image's shallow and deep features. Then, a parameter-free attention mechanism is introduced in feature fusion to focus on the key features of liver tumors. Finally, reparameterized convolution and shapeaware intersection over union (SIoU) loss functions are used to improve tumor detection and localization accuracy. The mean average precision（mAP）of this method on LT3DM and LiTS2017 datasets reached 92.9% and 92.2%, respectively, which is 2.3% and 1.8% higher than that of the YOLOv5 model. The experimental results indicate that this method has a greater ability to detect liver tumors than standard detection models.

Ground taxiing takeoff and landing is a major way for unmanned aerial vehicles (UAVs) to take off and recover. The runway maintenance and lateral deviation correction control during the taxiing process play a key role in the safety of the aircraft. A deviation correction control method is proposed to solve the lateral deviation correction control problem for high subsonic UAVs. The real-time cross-track error is included as a feedback compensation, and the nonlinear guiding algorithm serves as the primary feedforward in the method’s design. Simulation is carried out, to verify the proposed method. This study introduces the linear extended state observer (LESO) in the active disturbance rejection control to mitigate the impact of unknown disturbances, including actuator control inaccuracy. The nonlinear uncertain term during the deviation correction control process is estimated by the observer, and feedback compensation is added to the control law. Through simulation and actual taxiing experiments, the proposed method can realize high-precision stable correction control under high taxiing speed, even in the presence of initial position deviation and actuator control error, so as to meet the need for autonomous take-off and landing control for high subsonic UAVs.

With the development of the modern computational fluid dynamic (CFD), the research and application of black box testing technology in CFD software system testing are significant in improving software quality. The basic principles of various black box testing technologies including equivalence partitioning, boundary value analysis, decision table, and state transform diagram were introduced. The application of these technologies to determine test conditions and create valid test cases in order to satisfy testing requirements and function coverage in national numerical wind tunnel (NNW) software testing was then thoroughly explained. This allowed for the prompt and efficient identification of failures resulting from inconsistencies between the design and implementation of NNW software and requirements. Engineering practice has demonstrated the great value of black testing technologies, which when used to NNW software system testing may effectively increase testing efficiency and coverage.

To address the insufficient feature learning of traditional random forests when processing high-dimensional and structurally complex medical images, this paper proposes an abdominal multi-organ detection method utilizing deep differentiable random forests methdo. This method first ingeniously fused deep learning with random forest, employing a global-local dual encoder to extract high-level features that were subsequently fed into a differentiable random forest for tree partitioning. The designed decision attention assigned weights to each decision tree, and all parameters were learned through backpropagation to construct the final end-to-end model. Unlike traditional random forests, this method performs tree node splitting in the form of probabilities, and multiple decision trees vote with weighted parameters. This backpropagation learning of node splitting parameters and voting weight parameters can avoid the local optimum brought by traditional random forest leaf node splitting, allowing the deep differentiable random forest to search for the global optimum. Ultimately, experiments were carried out on two public medical image multi-organ datasets (AbdomenCT-1K and AMOS2022). The findings reveal that, in comparison to the benchmark method, the average WD value for five organs in AbdomenCT-1K decreases by 0.7−2.66 mm, while the average WD value for seven organs in AMOS2022 declines by 0.67−2.68 mm. These demonstrate that the proposed method achieves superior detection accuracy.

Accurate prediction of remaining useful life (RUL) is critical to the stability, reliability, and safety of aircraft engines. A new deep learning model called BSVAR is suggested for RUL prediction in order to address the issue that current RUL prediction techniques are unable to properly utilize the deterioration information of sensor data. The deep degradation information of sensor data is extracted using a bidirectional long-short-term memory (Bi-LSTM) networks and self-attention based variational autoencoder (SVAE). With the utilization of variational inference, the sensor data is clustered according to the implied degradation information, meanwhile, the latent space can be generated. The combination of the Bi-LSTM, the SVAE, and the regressor is used to establish a RUL prediction model to sufficiently extract the degradation features of sensor data to improve the prediction accuracy. Results from experimental validation on the aero-engine C-MAPSS dataset demonstrate that the suggested approach outperforms the current RUL prediction approaches in terms of prediction performance and can identify the engines’ degree of degradation in the latent space.

For tiny tilt-rotor aircraft, the tilt transition mode is a crucial phase of flying during mode switching. To investigate the aerodynamic characteristics of the rotor under different tilt attitudes during the tilt transition process, the steady aerodynamic experiments were conducted in the tilt angle range of 0-90° for a small fixed-pitch rotor. Firstly, a fan-array open-wind-wall system was established to simulate the flow conditions. Using a hot wire and a five-hole probe, the airflow quality was assessed, and the test procedure was confirmed. Multi-dimensional aerodynamic data of the rotor were measured using a high-resolution six-component force/moment sensor, which enabled the acquisition of aerodynamic characteristics under various combinations of air flow speeds, rotational speeds, and tilt angles. The results indicate that the influence of tilt angle on axial aerodynamic characteristics is related to the air flow velocity in the tilt angle range of 0-90°. When the air flow velocity is greater than 5 m/s, the axial thrust decreases with the increase of tilt angle. The axial torque is less affected by the air flow velocity than the axial thrust. Local negative thrust phenomena are more likely to occur under conditions of air flow velocity greater than 7 m/s and rotational speeds below 2500 r/min. The horizontal lateral force and pitching moment in the rotating plane decrease with increasing tilt angle, with values in the 0-45° tilt angle being significantly higher than those in the second half. In the ground coordinate system, the rotor's vertical lift coefficient decreases rapidly after 15° tilt angle, and the influence of horizontal velocity on the vertical lift coefficient is less than 20% under the condition of 5000 r/min. The horizontal thrust coefficient remains relatively constant between 0° and 30° tilt, and beyond 60° tilt angle, the horizontal thrust increases gradually, with its value being strongly influenced by rotational speed and velocity.

Shape stealth is the primary factor determining the stealth performance of an aircraft. In the shape design of modern military aircraft, both aerodynamic performance and stealth performance need to be comprehensively considered. Aerodynamic stealth optimization is a complex multimodal problem. Given the high computational cost of global optimization and the tendency of gradient optimization to fall into local optima, an optimal design method combining global multimodal optimization for airfoils and gradient optimization for layouts is proposed. To address the insufficient local search capability of the classical multimodal particle swarm optimization algorithm, the fitness-euclidean distance ratio ring topology local-best particle swarm optimization (FER-R3PSO) algorithm is introduced by combining the ring topology particle swarm optimization algorithm with the fitness-euclidean distance particle swarm optimization algorithm, which enhances the local search capability of the classical multimodal particle swarm optimization algorithm. To combine the multimodal search algorithm with surrogate models and reduce the computational burden of using the multimodal algorithm in engineering applications, a surrogate model point addition strategy and a peak extraction method suitable for the multimodal search algorithm are proposed. Function tests, airfoil stealth optimization, and standard aerodynamic examples are used to verify the effectiveness of the multimodal algorithm. The global/gradient coupling design for the flying wing layout is proposed, and aerodynamic stealth optimization based on the adjoint method is carried out using the three-dimensional shape obtained through global multimodal algorithm optimization of the airfoil assembly as the initial stage. The optimization results show that, compared to gradient optimization, the proposed method can achieve a shape with superior aerodynamic and stealth characteristics at a lower computational cost.

Considering the insufficient efficiency of creep calculation with the finite element methods in the reliability analysis process of turbine blades, a principle for predicting blade creep strain based on the creep information of finite element nodes was proposed. And then a model for creep strain prediction was built based on multilayer perceptron (MLP) learning network. Such model can predict the later creep strain of the finite element nodes according to their earlier creep strain. Compare to the formula fitting method, machine learning method is utilized in the proposed model. Results suggest that the proposed model performed better in the creep prediction with less creep history information. Since more creep history information can be learned in the more sever working condition, it is helpful to improve the performance of machine learning model by enhance the level of temperature field and rotation speed. Finally, the working condition with the highest level of temperature field and rotation speed was regarded as the training condition. The MLP model predicted the creep strain at 6～100 h with the input of creep history information before 6 h, whose error for the finite element nodes with large creep strain was about 10%. And 43%～48% computation time of the finite element computation were economized with the proposed model, which enhance the reliability analysis efficiency of the creep strain of the turbine blade.

Real-time satellite clock estimation based on regional networks is a prerequisite for regional precise point positioning (PPP) services such as the BDS-3 PPP-B2b service. However, issues like low precision and persistent bias plague regional real-time satellite clock estimate, which is constrained by the dispersion of ground stations. Group delay variation (GDV), which varies with satellite elevation angle, is the low-frequency part of the code multipath. The existence of GDV will affect the convergence of the regional network clock estimation and needs to be corrected. Therefore, this contribution extracts and models GDV between 15 regional stations and BDS-3/ global positioning system（GPS） satellites, and applies the GDV model to real-time clock estimation based on regional observation, and analyzes its impact. The results show that, after considering the GDV correction, the root mean square (RMS) of BDS-3 and GPS satellite clock offsets are reduced by 24% and 18%, respectively; the standard deviation (STD) is increased by 56% and 52% for BDS-3 and GPS, respectively. Finally, the clock offset is used in kinematic PPP. The BDS-3-only positioning accuracy is improved by 37% and 19% in the horizontal and vertical directions, respectively, by using the clock offset that accounts for the GDV correction. A similar conclusion can also be found in the GPS-only case, whose positioning accuracy is increased by 49% and 27% in the horizontal and vertical direction, respectively, and in the GPS/BDS-3 case, whose accuracy is increased by 61% and 32%.

Sculptured surface iso-segmentation method based on boundary point tracking has applicability to different segmentation situations. However, it exhibits low efficiency in the application of sculptured surface iso-segmentation. Therefore, this paper presents a geometric cognitive computing algorithm for surface iso-segmentation. Firstly, the input surface is sampled by isoparametric grid sampling, and then the boundary points are calculated. Secondly, the boundary point calculation group is extracted from the sampling points, and then the boundary points are calculated and recognized. Thirdly, the boundary point group is extracted from the boundary points, and then the boundary edges are constructed and recognized. Finally, the inner boundary loop is extracted and recognized from the boundary edges to split the input surface, and the resulting iso-surfaces are recognized. Attributes including principal curvatures, Gaussian curvature, mean curvature, and machinability are used to design a condition set for the iso-segmentation of several surfaces. The segmentation efficiency of the proposed method is 38.44% higher than that of the existing method. Test results show that the proposed method can achieve higher segmentation efficiency in the application of sculptured surface iso-segmentation.

Subspace clustering can assign high-dimensional data to different low-dimensional subspaces, which has extensive applications. The majority of subspace clustering techniques usually make the assumption that each variable in high-dimensional data has an equal impact on the clustering process. However, this assumption is not suitable for practical applications. To address the above issue, this paper proposes a self-weighted scaled simplex representation subspace clustering method. The self-expressive method is used to reconstruct the weighted data after each variable has been given an appropriate weight based on differences in relevance. In addition, a sparsity regularization term is utilized to control the sparsity of weights. Simultaneously, the scaled simplex representation is introduced to obtain a more reliable coefficient matrix. The enhanced Lagrangian multipliers approach is used to optimize all of these phases and combine them into a single framework. Experimental results on real-world datasets demonstrate that the proposed algorithm has better clustering performance than existing clustering methods.

An integrated optimization design methodology is developed for variable camber wing using active aeroelastic wing (AAW) technology, based on a genetic optimization algorithm. The optimization design for low aspect ratio aircraft’s scale model’s trim using multiple variable camber wings is conducted in a steady rolling maneuver. The goal is to reduce the structural mass while taking into account the limitations of the wing root’s moment, flutter speed, and the equivalent deflection angle of the morphing leading and trailing edges. A comparison between the new method and the conventional design method with single variable camber is also presented. The results showed that morphing the leading and trailing edges could increase control efficiency by 34.71% when compared to the traditional control surface. By coordinating the morphing of multiple leading and trailing edges, the AAW technology could fully utilize the flexibility of the wing structure to effectively reduce maneuver load and the wing’s structural mass by 12.9%.

To predict and regulate the space vehicle’s undesirable coupling motion produced by tiny amplitude sloshing of liquid propellant, a transfer function model identification study is carried out based on the findings of numerical calculations. For an example of a horizontally placed kerosene tank in a vehicle, during the level flight in the re-entry phase, a force effect transfer function description model is established with the range of rotational motion of interest, based on which a sloshing suppression design is carried out to increase sloshing damping and weaken the impact of sloshing on flight. The findings show that the transfer function can capture the impacts of the liquid propellant sloshing force, which can be utilized to anticipate space vehicle flight characteristics and design sloshing suppression.

Aiming at the problem of recognizing the intention of space non-cooperative targets without prior information, an intention recognition method based on fuzzy reasoning is proposed to enable traceable identification of target intentions. The three indicators “flying around”, “strike”, and “angle” are used as situation assessment metrics. An intention recognition dataset is constructed by integrating characteristic information of typical close-approach behaviors of space non-cooperative targets with domain expert knowledge. The intention-related feature data is fuzzified through cluster analysis and membership function assignment. A fuzzy decision tree algorithm is then employed to build the intention recognition model, enabling accurate and traceable identification. Finally, numerical simulations are conducted to validate the effectiveness of the proposed method, demonstrating superior recognition accuracy compared to the baseline algorithm.

To evaluate the performance of broadcast ionospheric models in existing global navigation satellite systems (GNSS), post-processed global ionospheric map (GIM) products provided by the International GNSS Service (IGS) Analysis Center are used. Long-term accuracy evaluations of ionospheric total electron content (TEC) and positioning domains are conducted for the BeiDou system, global positioning system (GPS), and Galileo broadcast ionospheric models. It is observed that during the solar activity ramp-up, the accuracy of all four models (BeiDou system Klobuchar 8-parameter model (BDSK8), GPS Klobuchar 8-parameter model (GPSK8), BeiDou global ionospheric model (BDGIM), NeQuickG) declines to varying extents, with distinct characteristics observed in different latitude bands. The results indicate that the BDGIM exhibits the best performance, with root mean square (RMS) increasing to 7.27, 4.43, and 4.00 TECU in the three latitude bands, and three-dimensional positioning errors increasing to 6.62, 2.81, and 3.51 meters, respectively. The BDSK8 is most influenced by solar activity, with an RMS amplification factor reaching up to 4 to 5 times. Finally, a Bagging regression tree learner is used to model and predict errors in various broadcast ionospheric models based on space physics parameters. The BDGIM model demonstrates the best predictive performance, with root mean square errors (RMSE) of 2.13, 1.23, and 1.47 TECU for the three latitude bands, and relative errors (RE) of 18%, 15%, and 14%, respectively. The NeQuickG model ranks second, with RMSE values of 4.60, 2.27, and 1.47 TECU, and RE of 17%, 18%, and 21%, respectively. The predictive accuracy of the two Klobuchar models is unsatisfactory.

This paper addresses the issue of segment congestion caused by increasing air traffic flow and focuses on predicting segment congestion situation. The spatiotemporal graph convolution model effectively extracts the spatial features of flight segments and captures the temporal characteristics of traffic data. First, the regional sector segment network is converted into the segment topology graph input to the model, where the information of the nodes represents the edge information in the segment network. A new metric for measuring segment congestion is introduced: the average flight time per unit length of segment. Then, a four-dimensional feature set is defined to represent the node information in the segment topology graph. Finally, a spatiotemporal graph convolutional neural network, consisting of a graph convolutional network (GCN) and gated recurrent unit (GRU), is employed to predict the congestion index. Experimental results show that, compared to the traditional autoregressive integrated moving average (ARIMA) model and GRU, the proposed model achieves optimal performance in short-term prediction (15 min), with the root mean square error (RMSE) reduced by 21.95% and 1.44%, and the mean absolute error (MAE) reduced by 23.36% and 3.74%, respectively. The method fully leverages the input features affecting segment congestion and captures the spatiotemporal correlations within segment traffic flows, significantly improving the accuracy of congestion predictions and providing technical support for effectively monitoring and managing traffic flow in each segment.

A new rapid reachability evaluation method is proposed for the multiple-warhead ballistic missile to solve the problem of rapid reachability decisions for MIRVs(Multiple Independently Targetable Re-entry Vehicle) in ballistic data calculation. Firstly, the MIRV model, controlled by the pitching and yawing angle of the missile, is established to fit the actual reachability area and analyze its characteristics. The second is a programmable method that uses the enhanced Ray method to determine the geometrical link between targets and reachable areas. Based on this, an integrated dynamical fit algorithm is shown and the method of expanding inaccessible region for reachability evaluation based on the idea of expanding boundary to targets is proposed in order to achieve the quick computation of multi-target's reachability. Finally, MATLAB simulation results verify the quickness and accuracy of the reachability evaluation in combination with the proposed decision algorithm.

A coupling characteristic representation method based on the statistical characteristics of radar scattering cross-section-coupling quantity is defined with the aim of addressing the issue of influencing the electromagnetic coupling characteristics between targets when multiple targets in the air make position and attitude adjustments. The method of combining the simulation test with the infield measurement is used to investigate the changing law of the electromagnetic coupling characteristics of the two-aircraft formation. The influence relationship between some formation factors and electromagnetic coupling characteristics in a certain spatial range is obtained, and the validity of the coupling quantity representation method is verified. The findings have engineering application value and may inspire future study on the assessment and improvement of multi-objective RCS coupling performance.

This study suggests the development program of a sleeve type spiral deployment coiled mast using the combination of threaded cylinder and straight guide cylinder, with the goal of designing the sleeve type spiral deployment coiled mast and addressing the crucial issue of whether the mast can be smoothly retracted. In order to predict the overall performance of the ground prototype, this paper adopts the virtual prototype technology, based on ADAMS software, establishes the dynamic simulation model of the sleeve type spiral deployment coiled mast, simulates the retraction and extension process of the mast, and analyzes the deformation characteristics of the crossbar in the transition zone in the post-flexion state. Finally, the ground test of the mast retraction and extension was carried out by the assembled ground prototype. The destabilizing deformation of the triangular cross frame during the mast's deformation process was used to determine the working principle of the mast using elemental instability. The simulation results were compared with the experimentally obtained relationship curves between the change in the mast's length with the chassis angle during the retraction and extension process and the change in the mast's coiled radius in the transition zone, confirming that the simulation model was in the post-flexion state. Comparison is made with the simulation results to verify the correctness of the simulation model and the feasibility of the design scheme.

It is crucial to assess how airfoil form uncertainty affects aerodynamic properties in order to suggest shape inspection indicators throughout the aircraft's whole life cycle. The traditional shape parameterization method has many variables and the calculation of aerodynamic characteristics is complex, making it difficult to directly analyze the uncertainty propagation between shape and aerodynamic characteristics. Class shape transform (CST) is applied to reduce the necessary variables for airfoils parameterization. Uncertain shape is described with random shape parameters. Aerodynamic coefficients are calculated through computational fluid dynamics (CFD) to generate samples. To create surrogate models for aerodynamics calculations, the sparse polynomial chaos expansion approach with least angle regression (LAR) is utilized. The coefficients of the polynomial chaos are used to directly determine the global sensitivity of the shape parameters, and the sensitivity results are used to pick crucial points for shape inspection. Examples show that controlling the deviation intervals of only two key points can reduce the variance of the lift coefficient by 25% and increase the average value by 3.8%, providing a reasonable inspection method for actual processing and maintenance.

Reducing the comprehensive cost of spare parts is one of the main objectives of cost optimization research in the support phase of complex systems. Especially in uncertain scenarios, such as large-scale equipment training and batch production of newly developed products, the demand for spare parts and the associated cost changes are harder to predict. Therefore, a Markov chain can be used to predict demand and complete the ordering process in combination with the emergency order strategy. However, there is a coupling effect between regular and emergency orders, which affects the total spare parts cost, creating a complex optimization problem. To address this issue, this paper proposes an optimization model and solves it using mixed integer linear programming. Finally, the model is applied and validated using data from a real-world scenario. The results show that the total cost is reduced by 295,600 yuan, with a 40.63% reduction in inventory storage costs. This demonstrates that the proposed model effectively lowers spare parts costs, improves system economic efficiency, and provides technical support for optimizing spare parts costs in complex systems.

In this paper, a heuristic constraint conversion scheduling method (HCCS) is presented for time-triggered flows in time-sensitive networks (TSNs). The HCCS method establishes a port scheduling order based on path and per-flow temporal constraints, ensuring that the transmission time of flow instances through high-priority ports is unaffected by low-priority ports. For flow instances passing through the same port, their scheduling order is determined by the flow priorities and their release times within the hyper-period. The constraints of flow instances are then converted into schedulable time intervals, within which heuristic searches are conducted to identify a time slot of sufficient length to determine the gate-controlled forwarding times of the corresponding ports. Each conversion addresses the constraints of individual flow instances, so the resulting time intervals confine the search space, significantly improving constraint-solving efficiency. Moreover, constraint conversion allows the HCCS method to be applied to both single-queue and multi-queue configurations, making it suitable for both zero-jitter and jitter-limited scheduling scenarios. Comparative simulations with the existing heuristic multi-queue scheduler (HERMES) algorithm show that HCCS increases the proportion of schedulable flows and achieves a computation speedup of approximately 3 to 25 times over HERMES, depending on the traffic scale.

Wireless local area network (WLAN) is one of the candidate technologies for wireless interconnection in wireless avionics intra-communications (WAIC) cabins. Drawing from the access mechanism defined by the IEEE 802.11ax standard and improving the uplink multi-user orthogonal frequency division multiple access (OFDMA), a proportional fairness-based minimum padding overhead (PF-MPO) method is proposed. This method clusters multiple users and improves system throughput while ensuring proportional fairness between clusters. In addition, it can reduce padding overhead by adjusting the transmission duration of nodes within the clusters. Moreover, a waiting delay PF-MPO (WD-PF-MPO) method is extended from the PF-MPO method by setting a waiting delay threshold to allow users with high delays to transmit immediately. Simulation results show that, compared with the maximum carrier-to-interference (MAX C/I) method, which focuses solely on maximizing average throughput, PF-MPO and its extension can increase the system’s Jain fairness index to above 0.9, while reducing the maximum transmission delay by at least 25% compared to MAX C/I. Furthermore, WD-PF-MPO can further decrease the maximum transmission delay while maintaining fairness.

The stratospheric wind field has an important effect on the liftoff and long-endurance station-keeping of stratospheric aerostats. By using the reanalysis data, a wind field prediction model based on long short-term memory (LSTM) network is built. Proper orthogonal decomposition (POD) is applied to extract key features of the wind field, enhancing the model’s ability to detect and represent wind field characteristics. The generalization ability of the reanalysis-based prediction model is evaluated using actual radiosonde data. Taking four consecutive years of reanalysis wind field data from the Kashgar region as a case study, a comparative analysis is conducted between the LSTM-based model and the POD-LSTM-based model. The results show that the POD-LSTM model effectively captures wind field features and provides more accurate predictions. Moreover, the POD-LSTM model demonstrates stronger generalization ability when predicting real wind field conditions. These findings offer a practical approach for accurate short-term stratospheric wind field prediction in scenarios where historical real wind data are limited.

The enhanced multi-strategy dwarf mongoose optimization algorithm (EDMO) is a proposed solution to the dwarf mongoose optimization algorithm's (DMO) low convergence efficiency and susceptibility to local optima. This algorithm employs a random opposite learning strategy to amplify the diversity and quality of the mongoose population, bolstering its global search capability and enhancing convergence accuracy. Concurrently, an adaptive approach is deployed to update the babysitter exchange coefficient, striking a balance between global exploration and local exploitation. In the latter stages of iteration, the algorithm capitalizes on the foraging behavior of the slime mold, optimizing between local and global optimal solutions. By solving the CEC2017 test function set, different algorithms are compared. The findings demonstrate that in terms of optimization accuracy, optimization speed, and resilience, EDMO which combines the three strategies performs better than the sophisticated algorithms under comparison. Through the experimental verification of UAV three-dimensional path planning, the EDMO algorithm performs better than the original DMO algorithm in local search, and the flight path generated is more stable.