An improved ESKF algorithm is proposed to solve the inaccuracy problem of navigation results caused by ignoring the high order in the error equation in the traditional filtering algorithm,which combines the filtering algorithm based on the error state modeling with the EKF.Aiming at the robustness and adaptability of integrated navigation,a filtering algorithm combining Kalman filtering algorithm and adaptive filtering algorithm is proposed and verified by simulation.The simulation results show that when the integrated navigation system is disturbed greatly,the accuracy of longitude and latitude of the integrated navigation system is improved by 49.8% and 34.21% respectively compared with that of the traditional filtering algorithm,and the accuracy and robustness of the integrated navigation system are improved effectively.

Based on the situation between UAV and target,the modeling and solution of unbalanced target task assignment are studied.Firstly,the situation between UAV and target in air combat battlefield is analyzed,a situation assessment method at a certain moment is proposed,and an unbalanced target assignment model for multiple UAVs is established.Then,the Hungarian-genetic algorithm is adopted to solve the problem.Finally,a simulation example is built for experimental analysis.The simulation results verify the effectiveness of the algorithm and also show that the Hungarian-genetic algorithm can improve the solving efficiency of the model.

Aiming at the defects of traditional detection methods in detecting and tracking micro-satellite that if the detection threshold is set too high,the target may be lost,while if it is set too low,a large number of false alarms will arise,this paper proposes a Track-Before-Detect(TBD) algorithm for micro-satellite by applying the basic concept of TBD to specific scenes.The observation model of infrared sensor is established,the TBD algorithm is deduced,and the specific implementation steps of particle filter are designed.Finally,simulation experiments are carried out to verify the effectiveness of the model and algorithm.

Aiming at the autonomous maneuver decision-making problem of UAV air combat,a lateral maneuver decision-making algorithm is designed.By adding heuristic factors and double Q-table alternating learning mechanism,the shortcomings of traditional Q-Learning algorithm,such as slow learning speed and many ineffective learning,are overcome.Through path planning simulation and the comparison of data,it is verified that the improved Q-Learning algorithm has better stability and solving ability.A dynamic grid planning environment is designed,which can make the UAV adjust the grid size adaptively according to the changing of air combat situation,and has no impact on the solution rate.Based on the Q-Learning algorithm,the lateral maneuver decision-making model of UAV air combat is constructed,and it is verified that the improved Q-Learning algorithm can play a significant role in improving the winning / losing ratio of UAV air combat through the exchange of weapon platforms.

Aiming at the problems of low brightness,weak contrast,high noise,and lack of details in low-light aerial images,a low-light aerial image enhancement algorithm(MARNet) based on Retinex theory and multi-attention mechanism is proposed.Firstly,the low-light aerial image is decomposed into light map and reflection map,and the CBAM attention mechanism is introduced into the noise adjustment network,so that the network pays more attention to the high-noise area and removes lots of the noises in reflection map.Secondly,a lighting adjustment network is designed by using up-and down-sampling structure.The channel attention mechanism is introduced to improve the brightness of the light map,and the regional loss function is added to improve the detail contrast.Finally,a set of low-light aerial images dataset is established by using the low-light image synthesis method with real noises.The experimental results show that the proposed method can enhance brightnes,reduce noise and restore details,and PSNR,SSIM,NIQE and visual perception are improved.

To optimize the trajectory of multi-UAV cooperative search for multiple moving targets,an information environment model based on search probability graph is established,and an Artificial Potential Field based Adaptive Particle Swarm Optimization (APF-APSO) search algorithm with adaptive parameter adjustment is proposed for dynamic target search in uncertain environment.In artificial potential field,the virtual repulsive force between the UAV and mountains and between UAVs is used to avoid obstacles effectively,and the virtual attraction force between the UAV and the target is used to speed up the search for targets.Through the nonlinear exponential function parameter adjustment method,the particle swarm parameters are adjusted,and the search probability graph is updated in real time according to the raster cell information certainty and target existence probability obtained by UAV in the search process,so as to guide the UAV to search the target.The simulation results show that compared with other algorithms,the proposed algorithm has great advantages in searching the target and shortens the length of the path.It avoids falling into local optimal solution and has good convergence.It can effectively realize the cooperative search among multiple UAVs and improve the search efficiency.

Aiming at the problem that UAV image matching takes long time,an improved UAV image matching algorithm based on SIFT is proposed.By introducing the FAST corner point detection algorithm,which can quickly judge whether it is a corner point by comparing the gray value of the center pixel and the pixel in a certain field,thus improving the speed of the algorithm.At the same time,in order to make up for the shortcomings of the FAST algorithm in searching on the Gaussian difference pyramid,an algorithm based on Ostu and GA is used to segment the image,the Gaussian pyramid is constructed for the segmented image,and the feature points are searched on the Gaussian pyramid.Experimental results show that compared with the traditional SIFT algorithm,the new algorithm improves the speed and accuracy of UAV image matching.

In SAR image target detection based on deep learning,in order to reduce the interference of noise and other characteristics in SAR images on feature learning and improve the interactivity of identifying and locating tasks in detection methods,a joint task detection method is adopted.The method makes use of the joint task network,so that the identification and location tasks share features as much as possible while retaining their own particularities,thus improving the supervision ability of the two types of tasks on feature learning.In addition,the method also uses the joint task learning method,and considers the reliability of identifying and locating tasks in the selection of anchor frame and the calculation of loss function,thus improving the training effect.The experimental results on public dataset prove the effectiveness of the method.

At present,most object detection rely on large-scale annotation datasets to ensure the accuracy of detection.In the actual scene,it is very difficult to obtain a large amount of data,and it also takes a lot of manpower and material resources to annotate the data.To solve the problem,a Few-Shot Object Detection algorithm based on Faster RCNN,CA-FSOD,is proposed,which detects the object samples when there are only a few annotated samples in the object category.In order to improve the detection performance,a CBAM-Attention-RPN module is proposed to reduce the number of irrelevant candidate regions.Secondly,a global-local relation detector module is proposed to obtain candidate regions that are more related to the object category by associating the features of a small number of annotated samples and samples to be detected;Finally,a classifier based on cosine Softmax loss is proposed as the classification branch of object detection, which can effectively aggregate the same category features,reduce the intra-class variance and improve the detection accuracy.In order to verify the proposed algorithm,it is trained and tested on MS COCO dataset.The experimental results show that the AP50 is 21.9% for this method,which is better than that of some existing few-shot object detection algorithms.

Aiming at the problems that the exponentially variable gain Iterative Learning Control (ILC) algorithm is difficult to be further improved and lacks optimization theory,a control gain optimization method of exponentially variable gain Iterative Learning Control (ILC) algorithm in Linear Time-Invariant (LTI) systems is proposed.Firstly,the necessary and sufficient conditions for convergence in Single-Input Single-Output (SISO) discrete LTI systems are obtained from the Toeplitz matrix properties and matrix iteration theory,and the convergence of the algorithm is proved.Secondly,the monotonic convergence condition of the algorithm is obtained from the optimization theory.Finally,the exact solution of the optimal control gain is obtained,and the relationship between the exponential gain and the optimal control gain is obtained.The method obtains the optimal control strategy according to the state equation of the system,and can calculate the precise optimal control value,which further improves the system convergence speed.The simulation results show that the method can effectively improve the learning speed of the algorithm and has good control performance.

With the rapid development of communication technology,optical communication system also has higher requirements for transmission distance,channel capacity and transmission rate.In order to solve these problems,it is necessary to study FEC code scheme with better performance.Concatenated code is a code types which has good performance in correcting burst and random errors,and can maintain low redundancy, so it has become the main research object.Based on the concatenated code of ITU G.975.1 standard,we proposed two new RS-BCH codes,made comparison and analysis in theory,and carried out corresponding modeling and simulation.One was the concatenated code of RS (255,239) code and BCH (31,16) code,and the other was the concatenated code of RS (255,239) code and BCH (511,448) code.The error correction performance of the two concatenated codes has been greatly improved compared with that of the single RS (255,239) code and the original code,and the latter has moderate redundancy and is easy to implement,which is more suitable for high-speed optical communication systems.

In recent years,with the rapid development of UAV technology,various successful cases of “UAV+” replacing manual work emerge one after another.The air-based intelligent EOD system designed integrates technologies such as UAV autonomous navigation,deep learning and YOLOv5 target detection, which combines hardware devices such as EOD actuators,dual-light integrated pan-tilt pods,and real-time image transmission systems.Experimental results show that the system has the advantages of wide visual search range,high target recognition accuracy,strong operability,and rapid EOD action.In the actual operation process,the operator controls the system to grab,transfer and destroy the unexploded bomb,and the work position of personnel is far away from effective radius of the unexploded bomb killing,which solves potential casualty problem during the Unexploded Explosive Ordnance(UXO) removal process.

Aiming at the formation control problem of single-leader multi-UAV systems with a communication delay,a controller is designed based on the consistency control theory,so that the system can still make each system state consistent under a certain communication delay,and form the expected formation.Firstly,the dynamic model of each UAV in the system is established,some basic assumptions are made,and the controller with undetermined parameters are designed according to the consistency control theory.Secondly,a new system is obtained through variable substitution and mathematical deformation,which transforms the formation control problem of the original system into the asymptotic stability problem of the low-order system.Finally,the Lyapunov-Krasovskii function is designed,and the sufficient conditions for the system to achieve consistency are derived by combining linear matrix inequalities.The simulation results show that the system can form the expected time-varying formation under the condition of time delay when the given conditions are met.

In order to solve the problems of unreasonable task assignment and low efficiency of multi-unmanned vehicle collaborative autonomous exploration in unknown environment,a distributed multi-unmanned vehicle collaborative exploration method based on wave-front algorithm is proposed.Firstly,a priori decision function based on the task assignment method of market mechanism is designed,and the boundary points are preprocessed by maximizing similarity with wavefront algorithm,so that the boundary guiding points are separated from the local optimization,and the min-max evaluation function is introduced to evaluate the boundary points.Secondly,the Dijkstra path planning algorithm is improved through the wave-front algorithm,which makes the moving trajectory of the unmanned vehicle more accurate and reduces the path inflection points.Finally,the method in this paper is compared with the method in literature [12] in various simulation environments,and the average exploration time of this method is reduced by 35.69% when the environment is fully covered.The experimental results show that the method can effectively improve the exploration coverage,reduce repeated paths and shorten exploration time,and improve the efficiency of multi-unmanned vehicle collaborative exploration.

Since the received signals suffer from mutual interference between the target echo and communication signals in UAV Radar-Communication integration(RadCom) system.In order to solve the problem,the standard OFDM signal is improved,an integrated signal model using interleaved OFDM is designed,in which each of the UAV is allocated a non-overlapping set of subcarriers within the signal bandwidth.A modulation symbol-based processing method is used to achieve the high resolution range-velocity image,no deterioration of the achievable range resolution.Simulation results show that the proposed method is superior to the existing method in the performance of target imaging and computational complexity,where the peak side lobe ratio of the range profile is improved by about 9 dB in low signal-to-interference ratio,and it can realize efficient data communication at megabits per second.

Aiming at the reduction of LOS stability accuracy of the airborne optoelectronic stabilization platform due to the disturbance,based on the disturbance estimation analysis and bandwidth limitation analysis of the LESO,the fast power reaching is modified,and then a fast power reaching law sliding mode active disturbance rejection controller is proposed.The system stability analysis based on Lyapunov method is carried out.The simulation results show that,under the disturbance and parameter perturbation,compared with the traditional sliding mode controller and PID,the system designed with the new controller has higher stability accuracy and remarkable chattering suppression effect.Compared with the controller when the reaching law is not modified,the new controller can maintain the high-precision output of the system.Furthermore,the design of the new controller hardly depends on the system model information.Therefore,the proposed method can improve the performance of the control system of airborne optoelectronic stabilization platform.

This paper mainly studies the unilateral signal detection method for resonant fiber-optic gyroscope,discusses the influence of thermally induced polarization noise on the resonance curve,and designs a signal detection method.The application effects of this detection method and traditional detection method are compared and analyzed,and the research results show that the proposed method shall suppress the influence of polarization noise on gyro accuracy.According to the simulation results,when the temperature changes by 0.003 ℃,the normalized amplitude error is significantly decreased from 0.2308 to 0.0298,thus verifies the effectiveness of the designed unilateral detection method,which effectively suppresses the adverse impact of polarization fluctuation noise on detection accuracy and improves the accuracy of detection results.

An improved algorithm based on bidirectional A* is proposed to solve the problem of high-efficiency UAV trajectory planning in 3D complex environment.A sector search area is designed to reduce computation overhead.In order to improve search efficiency,the dynamic weighted bidirectional A* algorithm is developed to optimize cost function.The key node screening strategy is introduced to eliminate redundant points in trajectory and generate global optimal static trajectory.Aiming at the problem of dynamic obstacle avoidance in trajectory,an optimization objective function of the shortest dynamic obstacle avoidance trajectory is designed,and the dynamic obstacle avoidance algorithm based on the calculus of variations with trajectory tolerance constraints is introduced,which realizes online adjustment and optimization of local trajectory.Simulation results show that the designed trajectory planning algorithm can not only plan an expected trajectory of UAV with high efficiency in complex environment,but also dynamically avoid random obstacles.

Accurate ship target detection technology can improve the omni-directional perception ability of weapons and equipment.Aiming at the serious problem of false alarm and missing alarm in SAR ship target detection in complex environment,a ship target detection algorithm 3S-YOLO based on YOLOv5 in lightweight SAR image is proposed.Firstly,3S-YOLO algorithm reconstructs the network structure,adjusts the relationship between receptive field and multi-scale fusion,and realizes the lightweight processing of feature extraction network and feature fusion network.Then,the network is pruned,and compressed by FPGM pruning algorithm to speed up the reasoning.Finally,the network is trained with varifocal loss to make IACS regression.The results show that the accuracy of the algorithm can be improved to 99.1% after optimization.After pruning,the volume of the model is greatly reduced,which can be compressed to 190 kiB,a decrease of 98.6%.The reasoning speed of the algorithm is increased by 4 times,and the reasoning time is reduced to less than 3 ms.Compared with the current mainstream algorithms,3S-YOLO has achieved good results in all aspects,which can meet the real-time ship target detection in SAR images.

Aiming at the problem of low tracking accuracy caused by the solidification of model probability in the traditional interactive multi-model filtering algorithm,a target tracking waveform optimization algorithm based on improved interactive multi-model filtering is proposed.Firstly,the probability transition matrix is weighted by combining the model probabilities of the two moments to improve the interactive multi-model filtering.Then the bald eagle search algorithm is improved based on tent mapping and Gaussian perturbation.Finally,in the multi-target scenario,according to the maximum mutual information criterion,the improved interactive multi-model filter is used to establish the objective function,and the improved bald eagle search algorithm is adopted to design the optimal transmission waveform.The simulation results show that the tracking error can be significantly reduced by using the algorithm.