Aiming at the problem of UAVs real-time mission planning under unknown threats,the mission re-planning is divided into two modes of human-machine cooperative planning and autonomous planning, and a logical selection architecture for UAVs mission re-planning is designed based on human-machine collaborative cognition.The architecture combines the autonomous planning ability of the UAV with the observation, judgment, decision-making and planning abilities of human beings to enable the UAV be adaptable to the complex and changeable battlefield environment.The fuzzy comprehensive evaluation method is used to evaluate autonomous ability of the UAV from four perspectives of human-machine interaction, UAV, environment and mission,and then the mission re-planning method is selected according to the autonomous ability of the UAV system.The result shows that the fuzzy comprehensive evaluation of autonomous ability is reliable and objective.

Aiming at the problems that there is a large difference in the positioning accuracy of the aircrafts in aircraft swarm, and the traditional leader-follower cooperative navigation structure system has low reliability and cannot fully utilize the cooperative navigation information of the swarm system, a hierarchical-structure swarm aircraft cooperative navigation method is proposed.The method stratifies the aircrafts participating in the swarm according to their positioning accuracy, and the positioning accuracy of aircrafts at the same layer is equivalent.An aircraft at low-precision layer can simultaneously receive the cooperative navigation information of multiple aircrafts at high-precision layer.Then, a cooperative navigation model based on relative distance and relative azimuth angle is established.Finally, the Kalman filter is used to correct the airborne navigation information of the low-precision layer aircrafts.The simulation results show that, the structure can effectively improve the positioning accuracy of low-precision layer aircrafts in the swarm flight system and enhance the reliability and fault-tolerant performance of the cooperative navigation system compared with the traditional leader-follower cooperative navigation structure.

The Joint Probabilistic Data Association (JPDA) algorithm needs accurate dynamic model of the target when solving the problem of multi-target tracking.However, the dynamic model mismatch often occurs in tracking multiple maneuvering targets.As one of the effective solutions, the Strong Tracking Filter (STF) is designed for a single maneuvering target in an environment without clutter.In order to improve the tracking accuracy of multiple maneuvering targets in clutter environment, a Joint Probabilistic Data Association and Strong Tracking Filter (JPDA-STF) algorithm is proposed.In order to obtain the fading factor of each target, the algorithm uses the weighted fusion of the measurement which is associated with the target to obtain the target innovation covariance, and the measurement weight is calculated by JPDA.The state prediction covariance can be obtained through the fading factor, and then the target state can be updated under the Kalman filter framework.The experimental results show that this algorithm has higher tracking accuracy than the JPDA.

A trajectory planning method satisfying multiple constraints is proposed for the off-line trajectory optimization in the gliding phase of interceptor projectiles in near space.First, the gliding motion model is obtained by simplifying the midcourse guidance motion model of the interceptor.Then, by using Pontryagins principle of minimum value, the indirect method model that satisfies multiple constraints is derived, and the trajectory optimization problem is transformed into a two-point boundary value problem of guessing the initial value of co-state variables.The adjacent extreme value method is used to select the initial value of the co-state variable, which reduces the difficulty and calculation amount of initial value guessing.The simulation results show that, compared with proportional guidance ballistics and Gaussian pseudo-spectral ballistics, the proposed ballistics can satisfy terminal constraints and multiple process constraints at the same time, and the optimization results have higher accuracy, which can effectively solve the offline ballistic optimization problem.

The trajectories of weakly-constrained, non-cooperative targets are massive, complex, and has real-time feature.Traditional trajectory analysis and processing technology is difficult to play an effective role here, thus we propose a technology for preprocessing the weakly-constrained, non-cooperative target trajectories.The model for cleaning the outliers, missing points and redundant points is studied.A dimensionality reduction method for trajectories is proposed based on the constraints of speed and course change rate.The reduction and expansion methods are adopted to increase the diversity of trajectory data.The simulation results show that:1) The compression rate of the processed trajectory is 64.44%, the accuracy loss is only 26.71%, and the data set is expanded by 3 times;2) High-quality trajectory data is obtained;and 3) The sensitivity of the model to the trajectory characteristics is reduced, and the models generalization capability is improved.

Aiming at the problems of model parameter uncertainty and interference of Hypersonic Vehicles (HSVs) in cruise phase, a robust predictive sliding mode attitude control method is proposed.The sliding mode surface is designed by using the error between the actual output value and the reference trajectory, and the prediction function is introduced to predict the error, thus the sliding mode surface of the future time can be obtained.The speed-loop controller is designed by using the improved sliding mode approaching law.This method ensures the stability of the attitude system and the accurate tracking of the command signal under the condition of uncertain parameters and interference, which has strong robustness and improved control performance.The effectiveness of the method has been verified by simulation.

Traditional submillimeter-wave imaging algorithms can image the target to be detected.However, due to the low-pass characteristics of the imaging system, high-frequency components in the signal are lost, resulting in the loss of details in the picture.On the other hand, the lobes are broadened in the course of signal processing, which may lead to aliasing of image points.Aiming at the shortcomings of the traditional algorithm in terms of poor imaging clarity and low resolution, this paper proposes a submillimeter-wave imaging super-resolution algorithm based on improved Projections Onto Convex Sets (POCS).This algorithm uses bilinear interpolation to obtain a high-resolution initial image.The initial image is subjected to iterative correction of convex set projection.Then, the adaptive correction threshold is introduced, and the correction threshold is dynamically adjusted by using the magnitude of the edge intensity, so as to achieve the purpose of accurate high-resolution restoration.Simulation results show that the algorithm can more accurately correct the pixels of the image, and improve the image resolution and imaging quality.

The light propagation under water is influenced by the absorption of water and particles, which may result in low contrast, noise, and color deviation of underwater images.Aiming at the problem that the enhanced image of the traditional homomorphic filtering algorithm still has foggy blur and dark color, an improved underwater image enhancement algorithm is proposed, which fuses the improved homomorphic filtering with multi-scale Retinex.Firstly, the single-scale Retinex algorithm based on bilateral filtering is used to correct the color of the original image.Then, the modified image is filtered by constructing a corresponding Butterworth notch filter.Finally, the image enhanced by homomorphic filtering and the image after color modification are fused at multiple scales.Experimental results show that this method can effectively suppress color offset and improve image resolution.

Images taken in foggy weather often suffer from low contrast and limited visibility due to the substantial presence of particles in the atmosphere, which absorb and scatter light during the propagation, and the efficiency of the visual system is limited and affected.Aiming at the degraded performance of the visual system in low-visibility weather, the foggy image degradation model is established, and an equation of transmittance is constructed by using the constraint relationship between the color distribution and the transmittance.Then, spatial-temporal continuous constraint to the atmospheric light and the transmittance is realized through online updating of atmospheric optical value and by use of Markov Random Field (MRF) model.The problem of adaptability and robustness of the visual system under complex environment or complex weather conditions is solved.Experimental result shows that, the proposed algorithm can effectively improve the imaging distance under low-visibility condition, and the dehazed video image keeps the color distribution of the original image without chromatic aberration, which can supply a fine image for the subsequent high-level vision mission.

Aiming at the problem that the accurate model parameters are not available for establishing a nonlinear dynamic system model to the data with random noise, a composite identification scheme with data preprocessing based on weighted Least Square-Support Vector Machine (LS-SVM) is proposed.According to the distribution information of the data, and by use of the robustness of weighted LS-SVM to abnormal data, the identification scheme removes anomalism of the data through regression calculation.Furthermore, the preprocessed data is used for data training to compensate for parameters of the fuzzy neural network, and the system model is obtained.The proposed composite identification scheme behaves better in the simulation of system identification with random noise.

For the finite-time robust control of a rigid robot manipulator, a new adaptive non-singular fast terminal sliding mode control method is proposed.This method combines non-singular fast terminal sliding mode control with adaptive law.Firstly, the non-singular fast terminal sliding surface is selected, which is used to accelerate the convergence rate of trajectory tracking error of manipulator and solve singular problems in terminal sliding surface.Then, hyperbolic tangent function replaces sign function to reduce chattering of control input.Moreover, the adaptive law estimates the unknown external disturbance and uncertainties, so as to achieve trajectory tracking with unknown lumped disturbance.It is proved that the robot manipulator system can converge stably in finite time by establishing the Lyapunov function.Finally, the simulation results of a two-DOF robot manipulator are presented to illustrate the effectiveness and robustness of the proposed control method.

Aiming at the problem of floating small target detection under sea clutter background, a new floating small target detection algorithm, EEMDCAN-Robust ICA & SG, is proposed.The denoising algorithm combines EEMDCAN, Robust ICA with Savitzky-Golay filtering algorithm to process sea echo signal, and the improved bat algorithm is used to optimize KELM model for chaotic prediction of reconstructed signal.The experimental results show that:1) The new algorithm greatly suppresses the sea clutter without destroying the chaotic characteristics of the sea clutter;2) Under the same conditions, the training time is shorter and the prediction accuracy is higher compared with that of the traditional chaos prediction algorithms;and 3) The detection performance of the new algorithm is stable, and the detection performance under low signal to clutter ratio is significantly superior to that of the traditional algorithm.It has been proved that the new algorithm can detect small floating targets quickly and effectively.

Hyperspectral imagery has a lot of applications in the national defense and civil fields.Especially, the anomaly target detection does not need any prior information and thus has become one of the key technologies and research hotspots in hyperspectral image processing and information extraction.Through systematic research and analysis, this paper summarizes the existing anomaly target detection algorithms in detail, analyzes and evaluates the key problems involved in anomaly target detection, gives the future development direction of the technology, such as sparse representation,tensor decomposition, and deep learning etc., and presents the existing problems of the algorithms.Some innovative ideas and future research trends are also proposed.

Considering of reducing the cost of test and maintenance, a mathematical model with cost function as the objective function is established, which transforms the problem of Testability Index Allocation (TIA) in the process of avionics system upgrading and refitting into a Nonlinear Programming (NLP) problem.Aiming at this NLP, a hybrid optimization algorithm combining Particle Swarm Optimization (PSO) algorithm with Sequential Quadratic Programming (SQP) algorithm is used to solve the problem, which makes full use of the formers strong global search ability and the latters strong local accurate search ability, and supplies an optimal solution method for the TIA problem in the upgrading and refitting process of avionics system.The results show that, compared with using PSO and SQP alone, the hybrid algorithm can obtain the optimal value faster, which verifies the feasibility of the proposed allocation method.

According to the statistical characteristics of signal and noise of diffuse-reflection target detection, analysis is made to the key parameters such as signal to noise ratio and receiving sensitivity of the coherent lidar.The research results show that:1) The receiving sensitivity is closely related to the requirement of false alarm rate and detection probability, as well as the dynamic range of target velocity and the detection wavelength;and 2) The sensitivity can be significantly improved by the mode of multiple detections.The research result is of great significance for the system design of coherent lidars.

To overcome the problems of limited calculation power and slow detection speed of the small intelligent reconnaissance UAV platforms, an improved target detection algorithm based on YOLOv3 is proposed.First of all, depthwise separable convolution is introduced to improve the backbone network of YOLOv3, which greatly reduces the quantity of parameters and calculation cost of the network, and improves the detection speed of the algorithm.Then, according to the characteristics of the target shape under the perspective of the airborne platform, the initial clustering center of K-means is preset when generating prior box, and CIoU loss function is introduced in the box regression.DIoU is combined with NMS to reduce the missed detections for dense targets.Finally, the improved model is optimized and speeded up by TensorRT, and deployed to the NVIDIA Jetson TX2 airborne computing platform.The experimental results show that the Mean Average Precision (MAP) of the improved algorithm on the verification set reaches 82%, and the detection speed is increased from 3.4 to 16 frames, which can meet the real-time requirements.

When the single-threaded operating mode of the existing radar simulation system is used to process a large amount of data or execute multiple tasks, it will increase the burden of the CPU, result in slow system operation and data loss, and affect the effectiveness of the simulation test.To solve the problems, a radar signal system is proposed based on the GNU Radio software platform by using its multi-threading technology and modular ideas, and the method of transmitting memory addresses is designed and used to establish the radar signal system, and improve the resource utilization of the computer.System operation in multi-threaded mode can not only process a large amount of data, but also increase the speed of data processing.The final test data shows the effectiveness of the radar signal processing system, which achieves the expected design goals, and meets the actual application requirements.

To solve the problems that the MEMS gyroscope is greatly affected by random error and filtering is needed, and the random error model established by time series analysis method cannot be directly applied to the dynamic filtering, a random error dynamic filtering method based on angular velocity estimation model is proposed.Firstly, analysis is made to the MEMS gyroscope random error and the model is established by using time series analysis method.Then, the angular velocity estimation hypothesis model is established as a three-dimensional linear model, which is combined with gyroscope random error model to form a dynamic filtering model.Finally, strong tracking Kalman filter method is used directly to estimate the angular velocity value for realizing random error filtering, and experimental verification is made.The results show that:The angular velocity value estimation accuracy of this filtering method is high under both static and dynamic conditions, and it can effectively reduce the random error of MEMS gyroscope and improve its accuracy.

In order to solve the problem of slow polarization changing of airborne radar, a dipole antenna array is designed, which can change the polarization between pulses.The antenna array uses orthogonal slice electric dipole array unit, vertical strips and reflection board, for realizing feed through coaxial coupling Balun.Then, a 4×4 rectangular-grid rectangular-boundary uniform plane orthogonal polarized antenna array is formed by using the improved elements, and fast polarization changing is realized by RF-MEMS switch.Then, HFSS software is used for parameter optimization and modeling simulation.The results show that, the antenna array has low standing-wave ratio, low sidelobe level, and high gain, which can be used as a reference design for the antenna of the orthogonal polarization phased array radar or as a subarray of the orthogonal polarization array antenna.

The Allan Variance is usually utilized to analyze gyroscope random error.However, when Allan Variance is used, there is a disadvantage that the estimation vibrates greatly in condition of long correlation time.The traditional Total Variance can solve the problem of Allan Variance, but the problem of excessive calculation exists in Total Variance.To deal with this problem, three kinds of sampling methods are introduced in this paper.In addition, an analysis is made to Allan Variance and Total Variance,based on which the Improved Total Variance is proposed, and an analysis is carried out on MEMS gyroscope random error.The analysis result shows that, the accuracy of Not Fully Overlapping Total Variance is nearly the same as that of the Total Variance, but computational time of the former is not more than 1/1000 of the latters, which can greatly improve the data processing efficiency.

According to the flight test requirements of a type of aircraft, it is necessary to adopt photogrammetry in bomb bay, install high-speed cameras at the corresponding positions of the bay, and use binocular, stereo intersection measurement method to test the weapon projection test at different altitudes, thus to detect the separation process, speed, attitude and so on of the weapon in the bomb bay.To estimate the accuracy of the airborne high-speed image testing system, and ensure the preciseness of the flight test, a method is presented for theoretical analysis on measuring precision of the airborne high-speed image measurement system.Through two experiments of freely falling body and precise rotation platform, the proposed theoretical analysis method is verified.The experimental results show that the proposed error analysis method has a guiding effect on the airborne high-speed image testing system of a certain type of aircraft, which provides a theoretical support for the measurement accuracy estimation in the following flight tests.

An airborne video and communication bus processing system is designed based on APU+FPGA for several widely-used airborne communication buses in current avionics system.With APU main processor and the programmable logic device FPGA as the core, one channel of input ARINC818 video is superimposed with the internally generated characters, and the function of outputting one channel of ARINC818 video to the display is realized.At the same time, it can support 7-way RS-422 and 4-way ARINC429 bus communication.In addition, the 22-way input discrete data and 7-way input analog data are processed.The system adopts the low-power, high-performance and highly-integrated APU and FPGA as the core processor, which greatly simplifies the hardware architecture, reduces the power consumption of the system, and improves the reliability.The design has been successfully applied to the display control management system of a certain aircraft.

The functions of Flight Management System (FMS) of performance management and so on depend on repeated invocation and calculation of the onboard performance database.Thus the application of the onboard performance database has become the research focus of the aircraft manufacturers and system suppliers.To convert the raw aircraft performance data into onboard database, first-principle based method for FMS performance database design is proposed.The method establishes a basic performance model inside the FMS, and uses built-in aircraft model and algorithms to calculate performance parameters and fuel consumption.The contents of the airborne database, the table creation and format analysis, the software application and configuration management are studied, which supplies a reference for future onboard software development and airworthiness verification.

Aiming at the problem that different operators are required to complete the drone operation and radio monitoring tasks when using drones for aerial radio monitoring, the radio signal direction finding and positioning function was integrated into the traditional ground station.It was realized that a single operator could implement both UAV control and radio monitoring in the ground station.Finally, air radio monitoring and cross positioning tests were carried out.The test results verified the superiority of air radio monitoring and the feasibility of the ground station system development.This system has certain practical guiding significance for improving the efficiency of civil aviation radio monitoring.