In order to solve the accuracy degradation problem of the point feature SLAM algorithm in low texture environments,a UAV binocular vision SLAM algorithm based on line feature constraints is proposed.Firstly,the observation of line features by using a binocular camera is divided into three situations,that is,a monocular camera observing in two frames,a binocular camera observing in two frames,and a binocular camera observing in one frame.Secondly,the residual block and Jacobian matrix in the three situations are calculated respectively.Finally,the line feature constraints proposed in this paper are used to improve the VINS-Fusion algorithm.The experiments are conducted on the EuRoC dataset,and the results show that the proposed algorithm exhibits good performance in over half scenarios of the dataset,and has a 13.2% improvement on average in positioning accuracy compared with the VINS-Fusion algorithm before improvement.

An arbitrary time guidance law is proposed for the interception of maneuvering targets with impact angle constraints.The design of the guidance law adopts the arbitrary time stability theory and the piece-wise sliding mode method,which solves the problem that it is difficult to obtain the convergence time of the finite-time and fixed-time guidance laws,and realizes the arbitrary setting of the convergence time of the line-of-sight angle.The designed guidance law is analyzed and proved by using the relevant stability theory.The effectiveness of the arbitrary time guidance law is verified by numerical simulationand the simulation results are compared with those of the finite-time and fixed-time guidance laws.

An adaptive genetic algorithm based on the greedy algorithm that improves mutation operation is proposed for the case of Multi-UAV searching for targets in specific areas.The search area is rasterized according to the prior intelligence,and the collaborative search model based on the state updating cycle is established considering UAV performance constraints.The 0-1 encoding is introduced to associate the UAV heading control sequence with the search probability.Considering that the repeated detection of a certain area by airborne radar could improve the search probability to a certain extent,the greedy mutation strategy adding the greedy operator is proposed,and the strategy selection threshold is introduced to realize the dynamic adjustment of the mutation strategy according to the variation of the search probability,so as to improve the local search ability of the algorithm in the later stage of the process.The simulation results show that the improved adaptive genetic algorithm performs better and has strong search ability and robustness.

An aircraft target tracking algorithm based on event triggering is designed to estimate the aircraft motion state.Firstly,in order to solve the problem of bandwidth limitation,the measurement information of the sensors is sent to the local estimators only when the trigger condition of random events is met.Secondly,considering the influence of the event triggering mechanism on state updating and covariance updating of estimation errors,the local estimators adopt an improved Unscented Kalman Filter (UKF) algorithm to obtain local state estimation.In addition,the fast covariance intersection method is used to fuse the state estimation of local estimators,and the fused state estimation of the target is obtained.Finally,an example of aircraft target tracking is used for simulation,and the results show that the communication times are reduced by 17% when the algorithm achieves the estimation effect close to that of UKF algorithm.

UAVs can provide target-related image information in such fields as military intelligence and aerial photography detection,providing target information for image processing tasks.An improved UAV image recognition algorithm based on YOLOv5s is proposed to address the issues of complex background,small detection targets and limited extractable features in UAV images.Firstly,the network structure is optimized by using CotNet module to enhance the model's self-learning ability and improve recognition accuracy.Secondly,the Neck network is improved to better utilize the rich information contained in shallow feature maps to locate targets through cross-layer linking and an improvement to feature map resolution.In the detection head section,decoupled detection heads are used to reduce the conflict between localization and classification tasks over feature information utilization in the prediction process.Finally,in order to improve the convergence rate and model accuracy,the aspect ratio of the width to the height of the loss function is optimized based on CIoU and EIoU loss functions.Tests are conducted on the test set of public dataset VisDrone.The proposed algorithm improves mAP50 and mAP50︰95 by 6.1 and 2.9 percentage points respectively in comparison with the original algorithm.The experimental results show that the proposed model can effectively improve the accuracy of UAV image recognition.

In geospatial remote sensing images,the target detection accuracy is low due to dense target distribution,large target scale variation range and too little feature information of small targets.To solve the problem,a multi-scale remote sensing small target detection algorithm cosSTR-YOLOv7 based on Swin Transformer (STR) and YOLOv7 is proposed.YOLOv7 is taken as the baseline network.Firstly,STR module is used to replace E-ELAN module in the backbone network，and it is improved to be cosSTR module by using cosine attention mechanism and post-regularization method，so as to improve the stability of model training.Secondly,a new feature fusion layer is constructed in the Neck part to reduce feature information loss.A small target prediction layer is added to the prediction part to improve the model’s ability of small target detection.Finally,a new SIoU loss function is used to calculate the positioning loss to accelerate the convergence rate of the model.The remote sensing dataset DIOR is used for the experiment,and the experimental results show that the mean Average Precision (mAP) of the proposed algorithm reaches 92.63%which is 3.73 percentage points higher than that of the original YOLOv7 algorithmand the performance of multi-scale small target detection has been significantly improved.

In order to enhance the ability of Particle Swarm Optimization (PSO) to jump out of the local optimum and improve the performance of PSO in solving complex practical engineering problems,and to improve the optimization effect of PSO in path planning problems,a PSO algorithm based on Compression strategy and Mutation strategy,called CMPSO,is proposed.Based on the compression strategy,a new mutation strategy is designed.The fusion of the two strategies effectively realizes the adaptive adjustment of particles,enhances the diversity of the population,and increases the probability of the algorithm jumping out of the local optimum.Through the verification of partial CEC2017 test functions,it is proved that CMPSO has excellent optimization ability.Finally,taking various constraints into consideration,CMPSO is applied to complex UAV path planning.Compared with other improved PSO,CMPSO can obtain shorter paths with less time consumption and cost.It is shown that this method can provide a strong technical support for complex UAV path planning.

Trajectory planning technology is the key to ensure the UAV completing its mission successfully.An improved ant colony algorithm is proposed to address the problems of the traditional ant colony algorithm,such as being prone to local optimum and poor convergence performance in UAV trajectory planning.Firstly, the initial pheromone concentration is unevenly distributed. A relatively good trajectory is obtained by using A* algorithm,and the blindness of initial search is avoided by making the pheromone concentration on this trajectory higher than that on other trajectories. Secondly,a heuristic factor is introduced to optimize the state transition rules,which improves the convergence rate of the algorithm.Finally,the variation of pheromone volatilization factor is smoothed by using the characteristics of Gaussian function,and the pheromone volatilization factor can be dynamically adjusted according to the distance between the current state and the target,thus avoiding the situation that the algorithm may fall into local optimum.The simulation results show that the proposed algorithm generates trajectories with shorter lengths and converges faster than the traditional ant colony algorithm does in the same environment.

It is difficult to accurately identify the state of deception jamming faced by UAVs during collaborative positioning.To solve the problem,a quantitative identification method that combines WGAN with Trust Model (TM) is proposed.Firstly,based on WGAN,rapid detection of the spoofing state of the UAV formation is realized.Based on this,the TM for collaborative positioning is used to realize accurate identification of the spoofed UAV.The simulation results show that,compared with similar methods,the proposed method has higher accuracy and stability in identifying the spoofed UAV.Moreover,this method does not require a large number of abnormal samples in the training stage,nor does it need a lot of new data interactions in the precise-screening stage.

The assessment of UAV autonomous combat effectiveness is an effective approach to measure the autonomous capability of UAV to complete predetermined mission objectives under specific combat conditions,and it is also the core issue in the demonstration of UAV autonomous combat technology and the development of equipment systems.Starting from the basic issue of UAV combat effectiveness,this paper firstly analyzes the basic concepts of UAV autonomy and autonomous capability as well as autonomous combat capability and autonomous combat effectiveness,and gives the basic process of UAV autonomous combat effectiveness evaluation.Secondly,the research status of unmanned equipment combat effectiveness evaluation index system and evaluation method is summarized from the perspectives of mission and principle respectivelyand the status quo of three types of index system construction and the progress of four types of effectiveness evaluation methods are summarized.Finally,the problems existing in the current evaluation are sorted out,and the development trend of UAV effectiveness evaluation technology is proposed.The relevant research can provide technical reference and direction guidance for the evaluation of UAV autonomous combat effectiveness,which has great military significance.

With the continuous development and maturity of Digital Radio Frequency Memory (DRFM) technology,active jamming to radar is more flexible and effective,so it is important to find a reliable and autonomous recognition algorithm for military applications.According to the general process of radar active jamming recognition,this paper summarizes the research results in recent years,and compares and analyzes the advantages and disadvantages of each algorithm.Firstly,the process and principle of jamming generation are illustrated by introducing the structure of DRFM.Then,17 kinds of common jamming are classified and the signal model is given according to the nature of the action,and the flow of jamming recognition algorithm is comprehensively sorted out based on the two core links of signal characteristics and classifier types.Taking two kinds of suppression jamming as an example,the influence of different characteristics and classifiers on the jamming recognition results is analyzed.Finally,the future development direction of radar active jamming recognition algorithm is predicted.

As a multi-source information fusion tool,D-S evidence theory has been widely used in the field of aerial target identification.Firstly,the D-S evidence theory is summarized.Then,the development of D-S evidence theory in the field of aerial target identification is analyzed,and three key issues that need to be addressed in the application are put forward.Then,regarding the above issues,this paper reviews the application status of Basic Probability Assignment （BPA） acquisition methodsevidence conflict measurement methods and evidence fusion methods in this field.Finally,from the perspective of airspace control,the application prospects of D-S evidence theory in the field of aerial target identification are given.The research can provide reference for theoretical development and engineering application in the field of aerial target identification.

Helicopters are widely used on the battlefield owing to their advantages of vertical take-off and landing,hovering in the air,and flying close to the ground.However,in degraded vision environmentsthe helicopter pilot's vision capability is limited and it is difficult for the pilot to observe the environment outside the cabin,which may lead to the occurrence of accidents.The airborne vision system can effectively solve this problem through the steps of sensor detection,data processing and terminal display.The directions of research and development of the vision system in degraded vision environments are sorted out,the capabilities of the vision system oriented to degraded vision environments are analyzed,and the principles,advantages and problems of different levels of vision systems are illustrated,which have certain reference value for the development of the airborne vision system onboard a helicopter.

The improved Clustering by Fast Search and Find of Density Peaks (CFSFDP) method is used for radar signal Pre-sorting.To solve the problem that the original CFSFDP method is sensitive to the cut-off distance,a method about histogram equalization is proposed to equalize the local density,which improves the robustness of clustering results.In view of the serious overlapping of radar signal parameters and unbalanced density distribution,an improved reachable distance calculation method and an improved clustering category assignment mechanism are proposed to improve the clustering performance on complex data sets.Through experiments on simulation data set and UCI data sets,the clustering results are evaluated by ARI,AMI and F1-measure.The results show that the proposed method can effectively deal with datasets with complex signal parameter distribution,and has better clustering performance compared with the original CFSFDP method and classic clustering methods (DBSCAN,AP,K-means,OPTICS).

In order to meet the high control stability requirements for the output of a tunable laser,a multi-loop control drive system applied to tunable lasers was designed and implemented.Firstly,in order to achieve high-stability current drive,an inner loop for current control was designed based on the principle of deep negative feedback control,and the interference of noise and current fluctuation on the Laser Diode (LD) was reduced by using integral compensation and RC filtering circuit.Secondly,in order to prevent the problem of LD power attenuation under constant current drive,an inner loop for power control was designed by using integral circuit and optical Power Diode (PD),and linear control was realized by using instrument amplifier circuit to eliminate common mode noise interference in the loop.Finally,in order to achieve real-time adjustment of drive current and output powerpositionbased PID control was applied to the outer loop of the system and together with the inner loopsa multi-loop control system was formed.The experimental results show that the stability of the current control is better than 200 ppm,the stability of the power control is better than 0.005 dB/h,and the stability of the laser wavelength is better than 1 pm/h,which can meet the high control stability requirements of the tunable laser.

The traditional single Infrared Search and Tracking (IRST) system lacks full coverage of observation,so it cannot provide all-dimensional situation information of the target.To solve the problem,an adaptive multi-IRST collaborative localization technology based on Geometric Dilution of Precision (GDOP) is proposed.Its core idea is to perform dual-aircraft pairing combination on multiple IRSTs,and to construct an information function oriented to localization accuracy based on the GDOP value of the pairing combination,and then to determine the federal filter information assignment coefficient.Finally,the problem of global information fusion of local estimation of dual-IRST pairing combination is solved based on the adaptive UKF federal filter algorithm,and the accurate location information of the target is obtained.The simulation results have proved that the proposed algorithm has high positioning accuracy and strong reliability.Meanwhile,it can ensure the accuracy and stability of the global positioning result.

To address the problem that existing image deraining algorithms cannot well preserve the image background details,a multi-stage image deraining network based on highly efficient channel attention is proposed.Firstly,the network uses 3×3 convolution to extract shallow features of the rain map and passes them on to the Highly Efficient Channel Attention Block (HECAB),assigning different weights to different feature channels.Then,it is transfered to three parallel stages.In the first two stages,the encoder-decoder is used for multiscale feature extraction to reduce rain pattern information loss,where the Transformer block is used to suppress useless information transfer.Finally,in the third stage,the initial resolution block is used to replace the encoder-decoder,thus preserving the fine features of the output image.The experimental results show that：1） The structural similarities of the proposed algorithm on the public test sets of Rain800,Rain12,Rain100L and Rain100H are 0.830,0.968,0.960 and 0.944,and the peak signal-to-noise ratios are 27.33 dB,35.27 dB,36.79 dB and 28.94 dB；and 2） Compared with classical and novel image deraining algorithms,the proposed algorithm has better results in removing rain patterns and recovering background details.

To solve the problem of insufficient identification performance in the research of intelligent individual radar emitter identification by using traditional convolutional neural network in low signal-to-noise ratio environment,a method of individual radar emitter identification based on Short-Time Fourier Transform (STFT) and EfficientNet is proposed.Firstly,the STFT is applied to the radar signal to extract time-frequency features,and then multiple MBconv modules in the EfficientNet are used to superimpose different time-frequency feature images to mine the more complex and abstract deep-level time-frequency features hidden in the signal image,including the distribution of signal intensity,time-frequency mode,periodic changesetc.,to complete individual classification and identification.Three parameters of network depth,width and image resolution of the EfficientNet can be simultaneously changed,which solves the problems of gradient disappearance and gradient explosion.The experimental results show that,in low signal-to-noise ratio environment,the method of individual radar emitter identification based on STFT and EfficientNet has better identification performance than traditional convolutional neural network.

An improved ADC assessment model is proposed for operational effectiveness assessment of the reconnaissance Unmanned Aircraft System (UAS).Based on the operational mission and operational process of reconnaissance UAS,a reconnaissance UAS operational effectiveness assessment index system is constructed. Based on the traditional ADC method,taking into account the factors of combatant protection and battlefield environment,the operational effectiveness of reconnaissance UAS is evaluated by using the hierarchical analysis method and the fuzzy comprehensive evaluation method.Finally,the feasibility and effectiveness of the model are verified by arithmetic examples,and the evaluation results show that the improved ADC method makes the reconnaissance UAS operational effectiveness assessment more relevant to the actual combat.