Introducing an energy consumption optimization strategy in quadrotor UAV trajectory planning to extend its flight longevity can significantly improve its practicality and economy.To this end, this paper proposes an energy-optimized trajectory planning method based on the Segmented Gaussian Pseudospectral Method (SGPM).Firstly, a comprehensive kinematics-energy consumption model of a quadrotor UAV is constructed, which can quantitatively describe the coupling relationship between UAV kinematics and energy consumption.Secondly, based on this model, the trajectory planning problem considering energy consumption optimization is transformed into a continuous time optimization problem with constraints.Finally, SGPM is designed to obtain the optimized trajectory, in which the sliding selection of optimization sub-intervals effectively reduces the computation time in complicated flight missions with a large number of waypoints, and at the same time improves the smoothness of the trajectory and the accuracy of waypoint tracking.The simulation and experimental results show that:1) Compared with the conventional fixed-waypoint flight, the proposed method can significantly reduce the flight energy consumption;and 2) Compared with the commonly-used optimization algorithms, the proposed algorithm ensures the optimization effect of energy consumption with shorter computation time and higher waypoint tracking accuracy.

When a wideband phased array radar is detecting high-speed targets, it is likely to encounter the problems of aperture traversing and range migration, and the fast-time frequency of the echo is coupled with spatiotemporal variables.To solve the problems, a joint correction method is proposed.The method uses frequency compensation and 1D Time-Scaling (TS) for decoupling when the radar works in narrow-transmission mode, and uses 2D TS to decouple the echo in wide-transmission mode of the radar.As for the differentiation of radar transmission modes, the corresponding defining conditions are given.The simulation results show that the proposed method can solve the cross-element and cross-pulse problems in the echo caused by aperture traversing and range migration, and the loss of Signal-to-Noise Ratio (SNR) at the threshold for effective target detection after the joint correction is no more than 3 dB, compared with that of the ideal state.

In order to solve the problems of detail blurring and color deviation of images processed by the existing dehazing methods, a dual-attention dehazing network based on feature enhancement is proposed.In this network, an encoder-decoder structure is used to design a dual-attention feature enhancement module, in which the Ghost module is used to replace the nonlinear convolution to realize the lightweight processing of the model.The Receptive Field Block (RFB) fully integrates the characteristics of different scales.Dual attention mechanism is introduced to realize cross-channel and spatial interaction of information, so as to ensure the performance of the model and suppress the noise features.The RESIDE dataset is used for network training and testing.The experimental results show that the proposed algorithm has excellent performance in both subjective visual and objective evaluation indicators, which can effectively improve the feature extraction ability of the network, realize the color restoration of foggy images in different scenes, and enhance the contrast and clarity of the image.

This paper presents an enhanced path planning strategy, namely Improved Artificial Potential Field-QRRT* (IAPF-QRRT*), to address the problem of the existing Q-RRT* algorithm in meeting both reachability and safety requirements during the path planning process.The IAPF-QRRT* strategy utilizes the Q-RRT* algorithm to obtain a set of critical discrete path points that connect the starting and ending points, which offers improved initial solutions and faster convergence speed compared with the traditional RRT* algorithm. Additionally, a novel non-potential orthogonal vector field is derived by enhancing the conventional Artificial Potential Field (APF) method.Under specific conditions, the repulsive vector field is orthogonal to the attractive vector field, thereby enhancing safety along the critical path points.Comparative analysis with other algorithms validates the effectiveness of the proposed IAPF-QRRT* strategy through numerical simulation experiments.

The Infrared Patch-Image (IPI) small target detection method has a wide range of applications in the fields of data chain, early warning, guidance, etc.For example, data chain can be used to accurately transmit IPI of small targets to the radar.In order to further improve the effects of infrared small target detection under complex background conditions, this paper proposes an IPI small target detection algorithm combining reweighting with local a priori.Firstly, the weighted Schatten p norm is used to constrain the background patch image.Secondly, the prior information of local contrast and the weighted l1 norm are introduced to suppress sparse non-target points, which further enhances the sparsity of the target image and further improves the performance of the algorithm model.The simulation results show that the proposed algorithm has better results than the existing classic algorithms in background clutter suppressing and accurate target detecting.

Target detection for aerial images has high application value in military and civilian fields.To solve the problems of low detection accuracy and inaccurate positioning due to factors such as small size of the targets, wide scale ranges, and background interference in UAV aerial images, a target detection algorithm for UAV aerial images based on the improved YOLOv8n is proposed.Firstly, the C2f module is improved, and the Deformable Convolutional Network(DCN) is used to replace the convolution in its Bottleneck to adapt to the deformation and scale variations of the objects in aerial images.The LSK attention mechanism is introduced into the backbone to dynamically adjust the spatial receptive field, thereby more flexibly adapting to the differences in background information requirements of different targets at the feature extraction stage.Then, the neck structure is improved, a shallow detection layer is added and the big target detection layer is removed, so that the network can more effectively capture the features of small targets to improve detection accuracy.Finally, the WIoU loss function is introduced to make the model focus more on low-quality samples and obtain higher detection accuracy. Comparative experiments and ablation experiments were conducted on the VisDrone2019 dataset.The mAP50 value is increased by 5.2 percentage points compared with that of the baseline model, the parameter count is reduced by 20%, and the detection speed reaches 87 frames per second, which can meet the real-time detection requirements. Comparative experiments were conducted with mainstream algorithms, and its performance is better than that of current mainstream algorithms.A generalization experiment was conducted on the DOTA dataset, and the mAP50 is increased by 1.7 percentage points, proving that the algorithm is versatile.

To address the problems that the existing Siamese network tracking algorithm conducts similarity matching by merely employing the features of the last layer of the backbone network and it is lack of effective template update strategy, a Siamese network tracking algorithm is proposed based on multi-layer feature fusion and adaptive template update.Firstly, a novel zero padding unit is developed by combining deep over-parameterized convolution, and the deeper foreground features and semantic background are extracted.Secondly, a novel global-local feature fusion module is proposed for fully aggregating the global and local information of shallow layer features and capturing rich superficial features and transitional features of the middle layer.An adaptive template update mechanism is used to online update the template.Assessment is made on public benchmark dataset to verify the effectiveness of the algorithm and the experimental results show that, the proposed algorithm achieves the accuracy of 0.878 and 0.588 on the OTB2015 and VOT2018 datasets respectively, and the average overlap rate on the GOT10K dataset reaches 0.526, outperforming other algorithms.

To address the issues of slow initial path formation, high failure rate, and poor path quality of Informed-RRT* algorithm, a point selection strategy based on the artificial potential field method is proposed to select high-quality sampling points.The greedy strategy with bidirectional direct connection and the dynamic step size strategy are introduced, so as to quickly obtain the initial path and enter the traversal optimization stage as soon as possible.Then, by implementing new sampling strategies and evaluation functions, the planned path is guaranteed to be more optimal.Finally, the path is optimized, so that the obtained path is more suitable for the operation process of mobile robots.The simulation results show that the improved algorithm has better performance than the Informed-RRT* algorithm.All the success rates of the improved algorithm in different environments are 100%.It is also proved that the convergence rate and path quality of the improved algorithm are better than those of the original algorithm under limited sampling times.

Aiming at the problem that the deep learning single-stage detection algorithm has insufficient feature extraction ability and insufficient sample learning for infrared aircraft targets, a target detection algorithm is proposed based on Feature-Enhanced Global Context Mechanism (FEGCM) and sufficient sample learning.FEGCM can obtain feature images containing both global and local information, and the target feature extracting ability of feature extraction network is improved.By adding modulation factor into Focal Loss, it makes full use of some easy negative samples containing target characteristics on the basis of paying attention to the learning of difficult negative samples, so that the samples are learned sufficiently, which helps the detection algorithm learn more meaningful target features.Experiments show that the proposed algorithm has a mAP50 of 96.9% on the self-made infrared aircraft dataset, which can effectively realize infrared aircraft target detection.

Aiming at the problems of low detection accuracy and lack of real-time performance of existing infrared ship detection algorithms, an infrared ship detection algorithm based on improved YOLOv8 is proposed. Firstly, the Multiscale Coordinate Attention (MCA) mechanism designed in this paper is introduced into the backbone network of YOLOv8 to enhance the capability of multi-scale feature extraction.Secondly, the YOLOv8 detection head is designed with shared parameters and re-parameterization, so as to improve the detection efficiency of the detection head.Then, the neck network of YOLOv8 is improved by using BiFPN structure, and the feature expression capability of the network is enhanced by bidirectional information flow and learnable weights.Finally, Faster Block is used to improve the C2f module of YOLOv8, which can maintain the accuracy while reducing the quantity of parameters and improve the detection speed of the model.The algorithm is tested on the infrared ship data set, and the mAP value reaches 93.1%, which is 2.5 percentage points higher than that of the original model, and the quantity of parameters is 32.6% lower than the original model.The experimental results show that the improved algorithm is much better than the original algorithm, which proves the effectiveness of the improved algorithm.

In order to improve the accuracy of decision-making of UUVs, a multi-layer cascaded fusion enhancement network for enhancing underwater images is constructed.An attention-guided color enhancement module is designed and it is combined with multi-layer cascaded enhancement architecture to enhance feature reuse while extracting multi-scale features from images.Secondly, a global adjustment module is designed to combine the Swin Transformer unit with the expansion convolution to improve the overall enhancement effect of the network on degraded images.Finally, the feature information extracted from each module is fused and enhanced by the triple feature aggregation module to obtain the enhanced underwater image.In order to train the model better, the joint loss function is constructed.Comparison experiment results with other underwater image enhancement methods show that the proposed method has good enhancement effect for the problems of color deviation and blurring that exist in underwater images, and is a great promotion of subsequent feature extraction tasks.

Ship target detection based on SAR images continues to face challenges due to environmental complexity, ship target dispersion, and scale diversity.This paper proposes a ship target detection algorithm specifically for SAR images.Firstly, a ship feature refinement module is developed based on deformable convolution to enhance the feature extraction capabilities for ship targets with significant aspect ratios. Secondly, a ship spatial pyramid aggregation structure is integrated at the end of the backbone network, thereby improving the global feature extraction capability for ship targets.Finally, a scale expansion feature pyramid network is designed to facilitate the interaction between shallow and deep feature information of the ship, thereby enhancing the detection capability for multiscale ship targets.Experimental results indicate that the proposed algorithm achieves a mean Average Precision (mAP) of 93.72% and an F1 score of 89.70% on the HRSID dataset, outperforming all the comparative methods and demonstrating effective detection performance.

The diffraction efficiency under oblique incidence is low in traditional optimization methods for Double-Layer Diffraction Optical Elements(DLDOEs).Based on the model of theoretical relationships of diffraction efficiency with incident angles and wavelengths for DLDOE, an angle-optimized design method is proposed. By analyzing the influence of different incident angles and micro-structure heights on diffraction efficiency in the designing process, optimal micro-structure heights are selected, which significantly enhance the diffraction efficiency at large field-of-view.In order to validate the above theory, the proposed angle-optimized designing method is applied alongside the method of equal diffraction efficiency under long and short wavelengths, and the bandwidth-integrated average diffraction efficiency maximization method, and the DLDOE in both the visible light and infrared spectrum are designed.The results demonstrate that the angle-optimized design method can effectively increase the angular bandwidth-integrated average diffraction efficiency of DLDOE, thereby enhancing the imaging quality of optical systems.Finally, based on the angle-optimized designing method, a long-focus hybrid refraction-diffraction system is designed, which achieves high image quality with high diffraction efficiency.

In Integrated Modular Avionics (IMA) architecture, the definition of system functions is different from the definition of task-oriented distributed computing in the previous joint architecture.The redundancy of functional applications on IMA platform usually represents the redundancy of system functions.In order to determine the redundancy of the functions of the airborne system, four variables are introduced, which are the type of application software, the general processing resource module, the IMA platform chassis and the functional application software redundancy.A task reliability evaluation method is proposed, which can set dependent variables and feedback adjustment in turn.It can scientifically design the redundancy of functional application software in IMA architecture system, which provides a new way to analyze the redundancy of system function in IMA architecture.

An Unmanned Aerial Vehicle (UAV) controller with high safety is a critical factor for ensuring the integrity of UAV combat, which is capable of regulating the UAV’s flight performance within a secure operational boundary.To design a safety controller for the UAV, analysis is made on the closed-loop control system of UAV, and the nonparametric prediction error estimation method is used to identify the actuator modules in the system, thus to demonstrate that the identified model is unbiased with the actual model. Furthermore, the concept of UAV safety is clarified, and a safety controller for UAVs is designed based on the identified model of the closed-loop control system.Finally, taking the pitch channel of UAV flight attitude angles as an example, simulation is made to validate the performance of the safety controller.Experimental results affirm the effectiveness of the UAV safety controller, which can regulate the UAV’s flight performance within a safe range, thereby ensuring the safety of UAV flight.

A Fractional-order Reduced-order Active Disturbance Rejection Control (FO-RADRC) algorithm is proposed to improve the dynamic performance and anti-interference ability of the photoelectric stabilized platform. First, a Reduced-order Linear Extended State Observer (RLESO) is adopted to improve the observation efficiency and accuracy by using the measurability and reliability of the system output signal.Then, a Fractional-order PD (FOPD) is used as the control law, thus enhancing the system stability and control performance. Numerical simulation experiments are carried out, and the proposed algorithm is compared with the Linear Active Disturbance Rejection Control (LADRC) algorithm and the Reduced-order Linear Active Disturbance Rejection Control (RLADRC) algorithm.The results show that the proposed algorithm is superior to the contrast algorithms in terms of dynamic performance indicators, and has stronger disturbance suppression ability under the equivalent disturbance at different frequencies.These results verify the effectiveness and innovation of the improved algorithm in improving the control effect of the photoelectric stabilized platform.

In the area of military aviation, complicated tasks pose challenges to the path planning of robotic arms.To solve the problems of low learning efficiency and low sample utilization of Twin Delayed Deep Deterministic policy gradient (TD3) algorithm, an improved TD3 algorithm of Recurrent-TD3 is proposed.Firstly, Long Short Term Memory (LSTM) is integrated into strategy network and value network to capture time series information of aviation control tasks, enhance its response ability to time series changes, and enable it to consider historical actions and states in decision-making, and improve the representation ability of the network.Then, Hindsight Experience Replay (HER) is integrated into the TD3 algorithm to avoid the difficulty in learning the sparse rewards in tasks, thereby making more efficient use of the samples by converting the experience of not reaching the goals into the experience of reaching the new goal.Finally, a collision detection process based on the bounding box is designed to improve the safety of robotic arm military aviation missions.The experiments show that this method can find a collision-free path faster than other methods, and the average path length is the shortest.

In response to the problem of inaccurate results caused by subjective factors in the evaluation of navigation equipment in the past, this paper proposes an evaluation method that combines Analytic Hierarchy Process (AHP) with Quality Function Deployment (QFD), with task requirement satisfaction degree as the evaluation indicator of navigation equipment capability.This method combines qualitative and quantitative analysis, based on the mapping process of navigation equipment of support task to capability requirements to tactical/technical indicators, proposes the evaluation steps of determining mapping domain indicators, calculating the importance of combat support capability, determining relationship matrix, and calculating the satisfaction degree of task requirements, and provides the method for calculating the satisfaction degree of requirements. Finally, an example of land-based Beidou navigation equipment support is analyzed, and the results prove the feasibility of this method, which is of great significance for the performance optimization and practical use of navigation equipment.