5G signals can provide accurate relative range/angle measurement information for multiple UAVs,thereby improving their cooperative positioning accuracy when the satellite receiver is malfunctioning.However,the 5G communication clock of each node in the cooperation network is often asynchronous,which will greatly affect the accuracy of range/angle measurement information,thereby reducing the fusion performance.To solve the above problems,a multi-UAV cooperative positioning method based on 5G signal asynchronous clock error compensation is proposed.Firstly,a relative range measurement model based on 5G signal asynchronous clock errors is constructed,and the position observation equation of the follower UAVs is established by using relative range/angle measurement information and altimeter information.Based on this,the joint optimal estimation of the Inertial Navigation System (INS) errors of the follower UAVs and asynchronous clock errors is realized by designing an Iterative Extended Kalman Filter (IEKF),which improves the positioning accuracy of the follower UAVs.The simulation results show that,compared with the traditional method,the proposed method improves the horizontal positioning accuracy of the follower UAVs by 25.3%.

The morphing aircraft is capable of actively adapting to changes in flight tasks, flight environments and flight states, thereby maintaining optimal performance in different tasks. Smart flexible deformable wing technologies, including the integrated flexible skin with good flexibility and load-bearing capacity, the high-power actuator, and the lightweight and high-reliability deformable mechanism, are crucial components of the morphing aircraft. In order to investigate smart flexible deformable wing technologies, based on the main wing deformation modes of the morphing aircraft, the focus and methods of researches on such key technologies of the deformable wing as the deformable skin, the drive technology and the deformable mechanism are analyzed. Additionally, the prospects of the aforementioned areas are identified.

In order to cope with the threat of dense coherent false targets of modern interferers,this paper takes linear FM radar against dexterous noise and C&I interference as the background.Firstly,a preliminary estimation method of interference parameters based on YOLOv5 for composite interference range-Doppler domain distribution feature extraction and interference recognition is proposed.Then,the Doppler parameters are accurately estimated by using the methods of feature region search and spectrum interpolation.The extracted accurate parameters are used to design Doppler filter banks to suppress the interference.Finally,the test statistic is constructed by using the suppressed signal,and the variance test is used to determine whether the Doppler filter weight needs to be updated for adaptive adjustment.The simulation results show that the proposed method can effectively suppress the interference in the environment of low SNR and multiple deception jamming.The study has reference significance for the suppression of dense false target interference.

Random frequency-hopping is a high-performance signal waveform with high anti-interference capability and low interception probability.However,the conventional random frequency-hopping signal is sensitive to target motion,the coupling of target motion and frequency modulation will introduce the phase high-order term in the echo,leading to range profile defocusing.In view of this,this paper proposes a random frequency-hopping radar imaging method based on the joint modulation of Pulse Repetition Time (PRT) and frequency,which eliminates the high-order coupling term of motion and frequency by changing the pulse repetition time based on the conventional random frequency-hopping waveform.Then,the kernel function correlation method and Fast Fourier Transform (FFT) are used for fast imaging,and the imaging process avoids the estimation of the target velocity,which reduces the computational amount and facilitates the engineering implementation.The simulation data and the measured data are used to verify the algorithm,and the experimental results verify the effectiveness of the algorithm.

The existing multi-scale Harris operator has a complex algorithm,large computation amount and low accuracy.To solve the problems,an efficient and simple algorithm is proposed.Firstly,Gaussian kernel function is used to build a multi-scale space for images,and then Harris operator is used to detect feature points in the scale space.The simplified 32-dimension SIFT feature vector is utilized to characterize the feature points.Then,the nearest neighbor method is used for feature matching,and the modified similar triangles method is used to screen the matching points.The improved K-means algorithm is used to group the feature points,so that the feature points within the same group are clustered and the feature points belonging to different groups are far apart.Finally,the modified RANSAC algorithm is used to calculate the transform matrix between the two images.The feature points from different groups are selected,so as to avoid the selected feature points being too close to each other and the algorithm falling into local optimum.The experiments verify the algorithm‘s performance.

To address the issues of external disturbances and model uncertainties during the flight of a quadrotor UAV,an Active Disturbance Rejection Control (ADRC) method is adopted to realize attitude control.Meanwhile,improvements are made to the ADRC.Considering that the error gain in the Extended State Observer (ESO) is too large,which is prone to generate oscillations due to error switching,a nfal function is constructed to replace the fal function to achieve a balance between disturbance rejection and oscillation suppression.Additionally,genetic algorithm is used to tune the parameters of the ADRC,which improves the efficiency of parameter tuning.In the simulation system,it is proved that the improved ADRC method exhibits improved disturbance rejection performance to some extent.The effectiveness of the control method is verified on a Hardware-in-the-Loop (HIL) experimental platform for UAV flight control based on a Links-RT real-time simulation system.

Single-image defogging is widely used in outdoor optical image acquisition equipment.To solve the problems of incomplete defogging and color bias of the existing methods,an image defogging algorithm based on Gaussian model and adaptive atmospheric light curtain is proposed.Firstly,the low-frequency components of the foggy image are separated by discrete cosine transformation,and then the linear relationship between the low-frequency components and the atmospheric light curtain and the Gaussian function are used to establish an adaptive atmospheric light curtain acquisition model to obtain the initial atmospheric light curtain.Then,an optimization function with unknown parameters is constructed by using Gamma correction,and the average saturation prior is applied to the atmospheric scattering model,and the final adaptive atmospheric light curtain is obtained.Finally,a clear fog-free image is obtained by using the atmospheric scattering model.The experiments show that,compared with several classical and advanced algorithms,the proposed algorithm has stronger versatility and is superior to other comparison algorithms on several indicators in objective evaluation,and the restoration results are clearer with natural color.

Using multiple UAVs to explore unknown environments can improve the robustness and execution efficiency of exploration tasks.Different from the heuristic method,the multi-agent deep reinforcement learning method eliminates the process of making rules artificially,and takes the UAVs as agents to independently learn more effective “rules” by interacting with the environment.A multi-threaded simulation environment for multiple UAVs is built to provide an environment for cooperative training of multiple UAVs.A Long and Short Term Memory neural network-based shared Multi-Agent Proximal Policy Optimization (LSTM-MAPPO) method is proposed to adapt to the multi-threaded environment,and the global boundary information is added on the basis of the cooperative LSTM-MAPPO method to increase the exploration area of each episode.The numerical experiment results show that:1) Compared with the existing Multi-Agent Depth Deterministic Policy Gradient (MADDPG) method,it can converge stably in later periods of training under the continuous action;2) Compared with the existing LSTM-MAPPO method,its final reward is stably above 5000;and 3) On three different simulation maps,the trained network can realize the stable exploration of more than 70% of the area during the test.

To solve the problem of low randomness of action selection in the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm,the normal distribution is used for action selection in the TD3 algorithm.Based on the interfered fluid path planning method‘s advantage of high path smoothness,a UAV path planning framework based on the interfered fluid and the TD3 algorithm is proposed.It is used for UAV path planning in unknown dynamic environment,which can realize fast convergence of the UAV path planning scheme.The simulation results show that the improvements to the algorithm dramatically enhance network training efficiency and meet the trajectory quality requirements while ensuring real-time obstacle avoidance,which provide new ideas for the application of Deep Reinforcement Learning (DRL) in path planning tasks.

Under dark vision conditions,the images captured by image acquisition equipment have low visibility.Retinex model enhancement method can map the corresponding reflective image by manipulating the estimated illumination.However,since the noise term is not in consideration,it is prone to amplify the noise of the enhancement results.To solve this problem,an image enhancement algorithm based on nonconvex constraint and noise suppression is proposed.Firstly,a new Retinex model with the noise term is defined.Then,an objective function with l0 regularization constraints is constructed based on the smoothing filter with gradient minimization,so as to obtain illumination images.Then,based on the above procedures,an objective function with l1 regularization constraints is established to separate the noise from the reflective image.Finally,after image reconstruction,the final enhancement results are obtained.The experimental results show that the proposed algorithm not only improves the visual effects of the image,but also has stronger noise suppression ability while retaining more information of the image.

A sliding mode control algorithm based on fixed-time sliding mode observer is proposed for the trajectory tracking control of rotor flight manipulator with self-modeling uncertainties and external disturbances.Firstly,the kinematics and dynamics models of the rotor flight manipulator are established in the generalized coordinate system,and a fixed-time observer is designed to estimate the uncertainties of the model parameters and the external disturbances.The piecewise function and the saturation function are introduced into the sliding surface to suppress the chattering of the fixed-time observer and speed up its convergence.Then,the rotor flight manipulator is divided into three subsystems,that is position,attitude and manipulator.Based on the fixed-time theory,a new fixed-time sliding mode controller is designed for the three subsystems respectively,and an adaptive reaching law is designed to suppress the chattering of the sliding mode controller.Compared with the traditional sliding mode controller,this controller has faster convergence speed and better tracking performance.The stability of the system is proved by Lyapunov stability theory.Finally,the effectiveness of the proposed algorithm is verified by simulation experiments,and it has good effects on the trajectory tracking control of the rotor flight manipulator.

The Flight Management System (FMS) onboard civil aircraft is an important system for controlling the flight trajectory in flight operation.The FMS constructs an economical flight profile based on the airborne Performance Database (PDB) and completes the trajectory control task by interacting with the display system,the auto-flight control system and the navigation system.To solve the airworthiness issue of FMS PDB and ensure the PDB correctness and integrity,one method combining the element-level database testing with system-level functional verification is proposed innovatively to demonstrate that the airborne PDB meets the requirements on system function,performance and airworthiness.The element-level testing technology of the airborne PDB on raw data structural integrity,format correctness and data accuracy is studied,and the system-level airborne PDB integrated functional verification technology is studied,which provide support for the autonomous control of civil aircraft FMS PDB key technologies.

Carrier-based aircraft ammunition loading is an important part of carrier-based aircraft ammunition support,and traditional carrier-based aircraft ammunition loading scheduling usually ignores the constraints existing in the work.To solve this problem,the constraints in the ammunition loading work are analyzed,and the sum of the ammunition loading time of all carrier-based aircraft is taken as the optimization objective,and a model of carrier-based aircraft ammunition loading scheduling is established,which considers different initial loading time of the carrier-based aircraft and the ammunition loading space.An improved genetic algorithm based on chromosome segment coding is proposed,which avoids the generation of a large number of unfeasible solutions and improves the efficiency of the algorithm.Through simulation analysis,the feasibility of the ammunition loading scheduling scheme obtained by the model and the algorithm is verified.

Considering the characteristics of airborne embedded systems,such as high real-time requirements,limited hardware resources and weak data and task security protection,this paper studies TEE-based airborne embedded data security technology.The object of security protection reinforcement is the three stages in the whole life cycle of data in airborne embedded systems,namely,data storage,data transmission and data use.Based on the unique data isolation feature of TEE environment,the encryption algorithm in data storage,the signature checking procedure in data transmission,the sensitive data and applications in data use are isolated from the general execution environment,which ensures the safe and reliable processing of key data between different tasks,and the encryption algorithm used in it is simplified,ensuring better real-time performance under the premise of limited hardware resources.The simulation results show that compared with the previous methods,the proposed algorithm has better safety,reliability and efficiency.

In order to solve the problems of low accuracy and low robustness of specific emitter identification based on a single feature in complex electromagnetic environments,a method of specific emitter identification based on multi-feature fusion is proposed.This method takes the envelope of pulse front edges,the carrier frequency deviation and the harmonic distortion coefficient as the features,designs a multi-channel 1D convolutional neural network,and achieves the identification by fusing the features of different structures.The experiments show that this method has good accuracy even at a low SNR,and can effectively solve the problem of low accuracy of specific emitter identification based on a single feature in complex electromagnetic environments.

Taking the incoherent signal source as the research object,this paper comprehensively analyzes the performance of Direction of Arrival (DOA) estimation of the monostatic coprime array MIMO radar.Firstly,the direction finding signal model of monostatic MIMO radar based on sum coarray and sum-difference coarray is derived on the basis of summarizing the structural characteristics of the coprime array.Then,by using the Multiple Signal Classification (MUSIC) algorithm,the DOA estimation methods of monostatic MIMO radar based on sum coarray and sum-difference coarray are introduced in detail respectively.Finally,through simulation experiments,a comparison with the uniform linear array MIMO radar with the same number of elements is conducted,which verifies the superiority of the DOA estimation performance of the coprime array MIMO radar.

During the operation of airborne LiDAR,the attitude changes of airborne platform can significantly influence the density distribution of point cloud and the accuracy of the reconstructed Digital Surface Model (DSM),so an attitude stabilization device is designed to compensate for them in real time.The designed attitude stabilization device is a nonlinear and strong coupling system.In order to eliminate the coupling effect of the attitude stabilization device and improve its motion control accuracy,a decoupling control strategy based on neural network inverse system is proposed,which obtains satisfying control effects.Firstly,the multi-variable neural network inverse model for dynamics system of the attitude stabilization device is established.Then,a PID closed-loop feedback controller and a feedforward compensator of neural network inverse system are composed to be a feedforward-feedback compound controller,to decouple the control system in real time and to improve the dynamic control performance.Finally,the decoupling control system is testified.The experimental results show that the designed decoupling control method based on neural network inverse system effectively improves the control accuracy of the attitude stabilization device,and has excellent robustness for error interference.

In order to investigate the obstacle-breaking capabilities of multi-rotor UAVs,this paper establishes a multi-rotor UAV efficiency evaluation index system based on the UAV obstacle-breaking combat environment and combat mission requirements.The overall effectiveness evaluation model and the sub-item capability evaluation model are established hierarchically by using correlation and weight analysis.Finally,taking five different types of UAVs as examples,this paper calculates the single-point obstacle-breaking efficiency of UAVs at different proportions of target obstacles,and gives the capability analysis results and suggestions of different multi-rotor UAVs based on the analysis of task completion time.

To solve the problems of a large number of small target samples and inadequate extractable feature information in UAV aerial photography images,a small target detection algorithm based on the improved YOLOv7 is proposed.Firstly,the low-level small target detection layer in the backbone network is integrated into the aggregation network structure,and a header is added to detect extremely small targets.Secondly,channel-spatial attention modules are added to the feature extraction process of the backbone network.At the same time,the feature fusion mode of improving the original connection in feature fusion is introduced,and the output weight of the feature graphs of each level is generated adaptively to dynamically optimize the representation ability of the feature graphs.Finally,the positioning loss function of SIoU is introduced into the prediction process to improve the model‘s detection ability and positioning accuracy.Experimental results show that the mAP50 of the improved model reaches 52.6%,which is 2.8 percentage points higher than that of the baseline YOLOv7 algorithm.The improved model also achieves higher detection accuracy than the mainstream detection methods,and has better performance in small target detection.