The multi-task area reconnaissance decision-making problem in an uncertain environment is considered, and the entire task-performing process is divided into two stages.According to the information of reconnaissance task areas and the performance of the UAV itself, a discrete Cuckoo Search Algorithm (CSA) is adopted to solve the reconnaissance route optimizing problem, so that the reconnaissance path of the entire task area is the shortest.Then, according to the characteristics of the task load and the reconnaissance task area, and under the premise of ensuring the reconnaissance of the entire task area and the minimum reconnaissance gains, the improved CSA is used to allocate the optimal task reconnaissance time for each reconnaissance task area, so that the information gains in the entire reconnaissance process is the maximum.Simulation results have proved the effectiveness and feasibility of the proposed algorithm.Compared with classic genetic algorithms, the improved CSA has a higher operating efficiency on such reconnaissance decision-making problems.It provides a reference for the scientific decision-making of UAV multi-task area reconnaissance optimizing problem.

The combination of satellite application equipment is the multiplier for improving operational capabilities. Traditionally, only the number of satellite application equipment is increased according to the operational needs, rather than optimizing the combination of satellite application equipment, which may cause repetitive configuration of satellite application equipment.This paper puts forward an optimizing algorithm for satellite application equipment combination to address operational needs.To achieve the purpose of avoiding the waste of resources, the combination of satellite application equipment is processed and optimized, its operational effectiveness is evaluated, and whether it is optimal is determined. Through a case analysis, the combination-optimizing algorithm of satellite application equipment is verified, and the optimal combination scheme of satellite application equipment is constructed.

To solve the problem of box particle redundancy in existing Box Particle Probability Hypothesis Density (BOX-PHD) filter, a new multi-target tracking algorithm based on box particle dividing PHD filter is proposed.Before the update stage of the target state estimation, each box particle obtained from prediction is divided into several sub-box particles to obtain the equivalent box particle subsets.Then, by using interval measurement, the weights of these box particle subsets are updated to estimate the state and number of the targets.The interval measurement information can be better used because the box particle is divided smaller.Also, the biased estimation caused by the insufficient compression of box particles can be effectively avoided.Simulation results show that the proposed method can effectively improve the target tracking performance.

In order to solve the problem of real-time attitude measurement of quad-rotor UAVs in flight and improve the accuracy of attitude estimation, a gradient-descent attitude estimation system based on STM32 is designed.The system takes STM32F427 microprocessor as the main controller, MPU6000 and LSM303D as the attitude sensor, and realizes data acquisition of the attitude sensor and attitude calculation.The accelerometer and magnetometer are pre-processed by the system, and the error quaternion is estimated by using the method of gradient descent filtering, and thus the integral error of the gyroscope is corrected and the accurate estimation of the attitude angle is realized.Finally, a comparison is made with the attitude angle of Pixhawk flight control system by using the established flight experiment platform.The experimental results show that the attitude estimation system using the gradient descent method can estimate the attitude of UAVs effectively and stably, which can meet the flight requirements of quad-rotor UAVs.

To solve the problem in path planning for multiple UAVsa hybrid Particle Swarm Optimization (PSO) algorithm is adoptedand the sum total of the cost function of each UAVs path is guaranteed to a minimum.The city environmentincluding buildings and other obstacles and threatening areas with radar interferenceis modeled.The method of setting up a number of waypoints and then inserting the split point is used.Thenthe hybrid PSO algorithm with differential evolution operations and adaptive inertia weight strategies is used for path planning of multiple UAVs.Finallythe validity of the algorithm is verified by simulations.

To solve the problem in path planning for multiple UAVs, a hybrid Particle Swarm Optimization (PSO) algorithm is adopted, and the sum total of the cost function of each UAV's path is guaranteed to a minimum. The city environment, including buildings and other obstacles and threatening areas with radar interference, is modeled. The method of setting up a number of waypoints and then inserting the split point is used. Then, the hybrid PSO algorithm with differential evolution operations and adaptive inertia weight strategies is used for path planning of multiple UAVs. Finally, the validity of the algorithm is verified by simulations.

The model-free adaptive algorithm is a data-driven methodwhich does not require an accurate model of the systemhas small calculation cost and is easy to implement.The existing model-free adaptive control algorithm often adopts the cut-and-trial method or uses a fixed constant when selecting the penalty factorwhich is difficult to achieve a satisfactory control performance.To solve this problemthis paper proposes a method to online optimize the penalty factor of the control law and the pseudo partial derivative.By using the idea of iterative optimization in the steepest descent methodthe penalty factor is optimizedthe convergence speed is increased significantlyand better system performance parameters are obtained.Hencethe optimal performance of the system can be achieved.On this basisthe stability of the closed-loop system is strictly proved.FinallyMatlab simulation results show that the proposed method has better control quality and stronger anti-disturbance abilities than the existing model-free adaptive control method.

The model-free adaptive algorithm is a data-driven method, which does not require an accurate model of the system, has small calculation cost and is easy to implement. The existing model-free adaptive control algorithm often adopts the cut-and-trial method or uses a fixed constant when selecting the penalty factor, which is difficult to achieve a satisfactory control performance. To solve this problem, this paper proposes a method to online optimize the penalty factor of the control law and the pseudo partial derivative. By using the idea of iterative optimization in the steepest descent method, the penalty factor is optimized, the convergence speed is increased significantly, and better system performance parameters are obtained. Hence, the optimal performance of the system can be achieved. On this basis, the stability of the closed-loop system is strictly proved. Finally, Matlab simulation results show that the proposed method has better control quality and stronger anti-disturbance abilities than the existing model-free adaptive control method.

To address the problem in reconnoitering the field communication network, a UAV path planning method is put forward for a full-coverage search of the deployment area of the field communication network by using crossed paths. Firstly, according to the structural characteristics of the field communication network and the node antenna beam, the maximum reconnaissance distance and the blind reconnaissance distance of UAV reconnoitering the nodes are obtained through a qualitative analysis. Then the static model and the dynamic model of UAV reconnoitering the nodes are obtained as well. Then, based on the reconnoitering model, the UAV crossed searching path model with exhaustive searching abilities is built and optimized. This method can find out all the nodes of the field communication network and improve the search efficiency. Lastly, a simulation of UAV crossed searching path is carried out. The simulation results indicate that the method can find out all the nodes of the field communication network, and the cost of UAV crossed searching path is calculated.

To address the problem in reconnoitering the field communication networka UAV path planning method is put forward for a full-coverage search of the deployment area of the field communication network by using crossed paths.Firstlyaccording to the structural characteristics of the field communication network and the node antenna beamthe maximum reconnaissance distance and the blind reconnaissance distance of UAV reconnoitering the nodes are obtained through a qualitative analysis.Then the static model and the dynamic model of UAV reconnoitering the nodes are obtained as well.Thenbased on the reconnoitering modelthe UAV crossed searching path model with exhaustive searching abilities is built and optimized.This method can find out all the nodes of the field communication network and improve the search efficiency.Lastlya simulation of UAV crossed searching path is carried out.The simulation results indicate that the method can find out all the nodes of the field communication networkand the cost of UAV crossed searching path is calculated.

To solve the problem in tactical decision-making of Unmanned Combat Aerial Vehicles (UCAVs) in air combatand inspired by the dual-process theory in cognitive psychologywe proposed a double-layer tactical decision-making strategy based on Case Based Reasoning (CBR) and MAX-MIN cloud reasoning.Firstlythe experience of pilots was converted into cases to construct a case libraryand the first-layer heuristic decision-making based on CBR was achieved by matching the degree of similarity between problem situations and original cases from case libraries.If the degree of similarity was lower than the thresholdit would turn to the second layerthe decision-making based on MAX-MIN cloud reasoning.The characteristic attributes concerning air combating were converted into cloud drops of attributesthen the cloud drops of combating rules was derived with the help of MAX-MIN cloud reasoning algorithmand furthermore the tactical decision scheme was obtained.The simulation result showed that the double-layer method can implement decision-making effectivelyand the time cost for decision-making satisfies the real-time request.

To solve the problem in tactical decision-making of Unmanned Combat Aerial Vehicles (UCAVs) in air combat, and inspired by the dual-process theory in cognitive psychology, we proposed a double-layer tactical decision-making strategy based on Case Based Reasoning (CBR) and MAX-MIN cloud reasoning. Firstly, the experience of pilots was converted into cases to construct a case library, and the first-layer heuristic decision-making based on CBR was achieved by matching the degree of similarity between problem situations and original cases from case libraries. If the degree of similarity was lower than the threshold, it would turn to the second layer, the decision-making based on MAX-MIN cloud reasoning. The characteristic attributes concerning air combating were converted into cloud drops of attributes, then the cloud drops of combating rules was derived with the help of MAX-MIN cloud reasoning algorithm, and furthermore the tactical decision scheme was obtained. The simulation result showed that the double-layer method can implement decision-making effectively, and the time cost for decision-making satisfies the real-time request.

To improve the efficiency of networked radar interferencean improved genetic algorithm was designed and applied to interference resource allocation.As to the three allocation models (one-by-onemore-by-lessand less-by-more)the allocation plans were made and the interference benefit value was calculated respectively.Compared with the results of other algorithmsthe improved genetic algorithm not only gets better allocation results and benefit valuebut also needs fewer iterative timeswhich verify the high efficiency of this algorithm.

To improve the efficiency of networked radar interference, an improved genetic algorithm was designed and applied to interference resource allocation. As to the three allocation models (one-by-one, more-by-less, and less-by-more), the allocation plans were made and the interference benefit value was calculated respectively. Compared with the results of other algorithms, the improved genetic algorithm not only gets better allocation results and benefit value, but also needs fewer iterative times, which verify the high efficiency of this algorithm.

Image edge detection has been widely used in practical applications, but the details of images are lost in the detection result. To solve the problem, a new image edge detection algorithm was proposed. Firstly, the two-dimensional binary wavelet transform was used to pre-process the image. Then, the edge points of the image were detected by using a new adaptive dual-threshold algorithm. Finally, an improved mathematical morphological gradient detection algorithm was used to further detect the edge information of the image. The simulation results show that: 1) The new algorithm can detect more edge information of the image, and the extracted edge of the image is more clear and delicate;and 2) Compared with single morphological algorithms, this algorithm greatly decreases the mean-square error value of the image and increases the peak signal-to-noise ratio by 2. 3 dB.

Target recognition is the key link of SAR image interpretation. The recognition rate of the existing SAR image target recognition method using sparse representation is not high enough. Therefore, we proposed a SAR image target recognition method using sparse representation and stretch transformation based on the analysis of the factors affecting the recognition rate, and according to the characteristics of the target region and the shadow area. This method may generate a new training sample image by stretching the training sample image, and construct a sparse dictionary by using the existing and new training sample image. By solving the joint sparse representation of the target region and the shadow area, the SAR image target recognition is completed according to the criterion of the minimum reconstruction error. The proposed method of target recognition was tested by using MSTAR SAR image. The results show that the recognition rate of the proposed method is higher than that of the existing method, and thus the validity of the method is verified.

In order to solve the problem of directional instability of the target drone, the stability augmentation control law is designed as the basis for subsequent control law designs. Then, according to the nonlinear dynamic characteristics of S maneuver, a hybrid control method consisting of robust servo mechanism linear quadratic regulator and adaptive dynamic offsets is used to design the tracking control law of lateral S maneuver command, which can make the roll angle and flight-path angle respond quickly as well as eliminate the sideslip angle rapidly. In order to decrease the coupling effect among channels in the process of S maneuver, the decoupling offsets control law is designed. The S maneuver control command is designed by analyzing the motion relationship between physical quantities in the process of S maneuver. A numerical simulation under Matlab/Simulink environment is carried out. The results show that the S maneuver control law can overcome the external disturbances effectively and track commands well.

Wavelet Neural Network (WNN) uses the gradient descent method to adjust the connecting weight and the scales for expanding/contracting and translating, which has the shortcomings of slow convergence speed and the local extremum. A method for assessing the operational effectiveness of electronic warfare UAVs based on Genetic Algorithm WNN (GA-WNN) is proposed. Based on WNN, the evaluation model uses GA to find the initial optimal WNN connecting weights, scaling parameters, and translating parameters. It avoids the blindness of artificial parameter setting. Simulation results show that this model can accurately and effectively assess the operational effectiveness of electronic warfare UAVs.

To recognize aircraft targets in airport remote sensing images quickly and accurately, a recognition algorithm combining the deep convolutional neural network with the edge contour feature extraction technique is proposed. The depth features of the aircrafts in the airport remote sensing image are extracted by using the deep convolutional neural network. To solve the shadow problem in aircraft parking positions, the target contour is obtained by using the optimized Canny operator, and then the aircrafts are classified by using Support Vector Machine (SVM). The the algorithm consists of the following two stages. The first stage is the training phase, which mainly trains the deep convolutional neural network and normalizes the obtained features. Then the edge features are obtained by using Canny operator and the major axis is obtained by using Principal Component Analysis (PCA) method. The Euclidean distance between the edge points along the two sides of the spindle is extracted as the eigenvector, and finally SVM classifier training is implemented. The second stage is the testing phase, in which the algorithm is verified and its accuracy is tested. Experimental results show that the recognition rate of the method can reach 94. 39%, which can effectively recognize the aircraft targets.

On the basis of studying the theories of classical ITTI visual attention models, the defects of traditional visual models applied to sea-surface SAR images are summarized according to the characteristics of the background and the target of sea-surface SAR images. A visual attention model design algorithm for sea-surface SAR images is proposed. Firstly, the model uses the basic framework of the classical ITTI model, selects and extracts the texture and shape features that can describe the SAR image well. Then the corresponding saliency map of features is obtained. Secondly, the new integration mechanism of the saliency map of features is adopted to replace the linear-adding mechanism of the classical model for fusing the saliency maps and obtaining the overall saliency map. Finally, the gray features of the attention focus of all the saliency maps are integrated to select the optimal significance characterization. By using the multi-scale competitive strategy, the filtering and threshold segmentation are completed to realize the accurate screening of significant areas. Therefore, the detection of the significant areas of SAR images is completed. Experiments were carried out by using Terra-SAR-X and other satellite data, and their results verified the good significance-detection effects of the model. The model can better meet the demands of the detection of high-resolution image targets. By carrying out further comparative analysis with the classical visual model, it is discovered that the proposed algorithm can not only reduce the impact of the false alarm caused by speckle noise and uneven sea-clutter background on the detection result, but also greatly improve the detection speed by 25% to 45%.

In order to realize the accurate navigation of visual odometers in the large-scale complex environment and eliminate the cumulative error of pose estimation, this paper presents a method of visual loop closure detection and pose optimization based on geometric constraints.Firstly, a Bag of Visual Words(BoVW) based on LATCH descriptor is set up, and the loop closure detection method is adopted with the images described by the visual word vector and the similarity degree normalized.Then, the loop closure between the candidate keyframe and the current keyframe is checked based on RANSAC-HORN motion estimation.Finally, the local map points of the loop closure keyframe are projected to the current frame, and the re-projection error is minimized to optimize the pose.The experimental results show that the loop closure detection and pose optimization algorithm proposed in this paper can effectively and accurately detect and verify the loop closure, carry out the loop closure pose optimization of the error accumulation of the visual odometer and improve the accuracy of visual navigation.

The qualitative differential game theory is used to study the pursuit-evasion game of Unmanned Aerial Vehicle (UAV) and Unmanned Ground Vehicle (UGV). The process of determining the dividing barrier between the escape area and the capture area in the presence of obstacles in the environment is studied. The scene of the game is set, and the kinematic equations on both sides of the game are established. The outcome of the game is analyzed according to the kinetic characteristics of both sides in order to obtain the game strategy. Then the impact of the obstacles in such environment and the conditions of game termination are analyzed. Finally the barrier is obtained, and an example illustration is given.

In large envelope ranges, the dynamic characteristics of UAVs have obvious perturbation. Only if the control law is designed according to UAV's different dynamic flight characteristics, can the high-quality control of the UAV be realized. Based on the clustering partition technology and according to the similarity and continuity of the dynamic stability and maneuverability of UAV's envelopes in the regional scope, the envelope is divided into sub-envelope regions with similar characteristics. This approach provides a basis for the design and scheduling of UAV's full-envelope control laws.

To overcome the difficulty of all-range and all-azimuth automatic track initiation in the dense clutter environment, the concept of grid connection is adopted for grid partitioning of the observation space to form an independent connected area based on certain criteria, according to the spatial distribution characteristics of clutters and targets. The trace points are preprocessed based on the analysis of the connected area. After the pre-processing, the track-initiation processing is implemented by using the measurements of targets. The calculation cost is reduced by means of decreasing the number of observation points used for judging track initiation, and thus the automatic track initiation is implemented rapidly. The feasibility of the method in engineering applications has been verified by simulations and the actual data.

In order to improve the operational effectiveness of the laser weapon system, an automatic laser beam-combining device was designed based on the deflection mirror by analyzing the current situation of laser beam-combining technologies. Tests were made on the focusing, beam-combining and damaging effectiveness of two lasers. The results indicated that this method can achieve the expected focusing effects and beam-combining performance. Compared with the single high-energy laser, the time for damaging UAVs by two lasers combined beam is reduced by 58%. It provides a basis for multi-laser beam-combining research in the future.

To solve the problems of large computational complexity and difficult hardware implementation of particle filters, the paper presents a simplified particle filter algorithm for bearings-only tracking, and uses Xilinx System Generator for its implementation on FPGA. Firstly, the generic particle filter algorithm is simplified to reduce calculation quantity and make it easy to implement. Secondly, the modular design is adopted, and the state machine is used to synthesize and realize the sequential control of each module. Finally, the algorithm is converted to hardware language to complete the hardware simulation. The simulation results show that: each module of the simplified particle filter algorithm functions properly and the algorithm has a good tracking accuracy and running speed, which can be used in the implementation of the particle filter in non-linear and non-Gaussian systems. It provides a reference for the hardware implementation of particle filters.

The basic concept and process of information fusion are introduced based on the JDL information fusion model. Then, a method for airborne multi-sensor information fusion composed of a whole set of algorithms is given, including the algorithm for space-time alignment, data association, composite tracking, and state estimation. Suggestions are presented for the engineering application of the algorithms. The simulation results show that the proposed fusion method can guarantee the timeliness of multi-sensor information fusion, and has high correct association rate and state estimation precision, which can satisfy the requirements of engineering applications.

Helmet-Mounted Displays (HMDs) are playing an increasingly important role in modern aerial combats. New binocular digital HMDs have large visual range and high resolution, which are more adaptive to the needs of flight and combat. But the lens of binocular HMD has large off-axis angle, which may lead to serious distortion of images that can change the original image size and shape and influence the display quality and precision. The traditional distortion correction technology of monocular display can't completely meet the needs of real-time distortion correcting. Therefore, this paper introduces a realization method of binocular real-time distortion correction system for the use of binocular digital HMDs. A scheme using pixels cross storage of binocular spatial index coordinates and storage optimization of DDR3 is proposed, and by using the adaptive smoothing filter, the sawtooth problem of the image edges displayed by the distortion correction system is solved. Therefore, a miniaturized real-time distortion correction system with low power consumption is implemented.

A fault diagnosis method is designed for the distributed flight control computer of UAVs. A method combining hardware redundancy with the model's analytical redundancy is proposed for the fault diagnosis of the computer's external sensors, external actuators and internal functional modules. By designing the diagnosis architecture, the fault detection and isolation for the computer's external sensors, external actuators and the internal digital controller of the flight control computer can be effectively realized. It solves the problem that the traditional single fault observer cannot diagnose the computer's internal faults, and improves the coverage rate of fault diagnosis and the reliability of the flight control system.