Based on the performance and features of the anti-submarine patrol aircraft and the airborne magnetic detector, the model for patrol search situation analysis, which meets the three requirements, is established. Two types, totally three kinds of patrol search methods are proposed for magnetic detector. Considering the V-H fuzziness of enemy submarines, the adaptability of different patrol search methods is studied through simulation calculation. The detection probability calculation model of patrol search of anti-submarine patrol aircraft using magnetic detector is set up.The study supplies a theoretical and technical support for the patrol search of anti-submarine patrol aircraft with magnetic detector.

Due to the status of the Unmanned Combat Aerial Vehicle (UCAV) itself and the situation of the air combat, there are many uncertainties in UCAV aerial combat, including the maneuvering and choosing of behavioral strategy of the two sides. To solve this problem, the reinforcement learning method is introduced into UCAV air combat, and the UCAV maneuvering model and action set are established.The function of air combat situation assessment is taken as the signal function for reinforcement learning, and the Probabilistic Neural Network (PNN) is used as the unit for enemy maneuver prediction. Under the condition of thorough perception of battlefield information of both sides, the UCAV can grasp the best maneuvering strategy through interacting with the environment by reinforcement learning, thus to implement one-on-one aerial combat.

In the method for the measurement of vibration displacement by way of machine vision, the segmentation of the marked image of vibration is the key point in the processing procedure. The segmenting result directly affects the final measurement accuracy. However, the segmenting results of the traditional Otsu threshold segmentation method and the improved algorithms are not satisfactory because of the small number of pixels of the target in the marked image and the interference factors in measurement.In order to get a better segmentation algorithm, this paper firstly explores the problems existing in the improved algorithms of segmenting the marked image under the interference of impulse noise, strong light and shadows. Then, an improved Otsu algorithm is proposed based on the distribution information of the gray level between the threshold and the target peak.Based on the improved algorithms, the ratio of the difference of the gray values between the threshold and the nearest target peak to the population, and the ratio of the sum of the pixel number of thein gray values to the population, are both taken as the weight. Finally, the segmentation experiment of the marked image and the measurement experiment of vibration displacement are carried out under the condition of multiple and complex interference. Experimental results show that, the proposed algorithm is still able to effectively segment the marked image under interference, and the deviation of the target's position is lower than that of other algorithms in general. Meanwhile, the accurate measurement of vibration displacement is realized.

A strategy based on Markov Decision Process (MDP) was proposed to solve the air-to-ground attack operation planning problem for manned/unmanned aerial vehicle formation. Firstly, according to the characteristics of combat operations of manned/unmanned aerial vehicle formation, the forming method of the state set and action set was proposed, and the method for calculating the state transition probability and reward function of MDP model was designed. Then, with consideration of the resources and time constraints, the MDP optimization model was established to execute the operation planning. After that, we made improvements to the Ant Colony Algorithm (ACO) on the path selection probability and pheromone updating method to solve the MDP optimization model. Lastly, a simulation analysis was carried out. The result showed the proposed method can obtain a good solution to the operation planning problem for manned/unmanned aerial vehicle formation with high calculation speed.

The traditional median filtering algorithm has poor filtering effect on the images with high-density salt and pepper noise. Based on the theory of cyclic iterative processing, an iterative adaptive median filtering algorithm is proposed to remove the salt and pepper noise. Following the traditional filtering approach based on decision-making, the proposed algorithm adaptively adjusts the size of the filtering window and calculates the median of the non-salt and pepper pixels in the filtering window to replace the corrupted pixels.After that, the proposed algorithm adaptively determines the iteration number according to the noise density, so as to completely remove the salt and pepper noise and retrieve the original image.Simulation results demonstrate that, compared with the standard median filtering algorithm and four other improved algorithms, the proposed algorithm can remove the salt and pepper noise with less running time and can efficiently retrieve the details of the original image for the images with the noise density varying from 10% to 99%.

In order to realize accurate and rapid image matching, this paper proposes a corner matching algorithm based on corner point detection and the descriptor. Firstly, in the curvature scale space, the corner points of the image at different scales are detected, and the unstable corner points are removed.Secondly, based on the accuracy of the curvature describing the contour of the image, the neighborhood is divided into 3×4 sub-neighborhoods taking the character points as the center, the 4-dimensional vector features in the sub-neighborhood are calculated, such as the Gaussian weighted curvature of the contour curve points, and a 48-dimensional descriptor is built.Since the CSS corner point detection includes curvature calculation, the quadratic calculation of the curvature is avoided when generating the descriptor, and thus the matching speed is improved. Finally, a binary distance method is proposed to match the descriptors to further optimize the matching speed. Experiments have proved that, with the assurance of accuracy, the CSS fast matching algorithm can shorten the matching time to a great extent and achieve satisfying matching results under rotation and illumination changes.

In order to realize the real-time image stabilization of the UAV aerial image and solve the problem of slow speed in feature detection and the divergence in motion filtering during the process of image stabilization, a real-time image stabilization algorithm is proposed combining the improved AGAST (Adaptive and Generic corner detection based on the Accelerated Segment Test) algorithm with the adaptive Kalman filter. For the real-time aerial video sequences, the image stabilization processing is based on the frame prior to the current frame. The improved AGAST feature detection algorithm rapidly extracts the features of AGAST corners on the basis of scale space, which are described by binary descriptors, and then the feature points are matched by using Hamming distance. For the matched sets of point pairs, the RANSAC principle is used to eliminate the mismatched points and then the motion vector is calculated. Finally, the adaptive Kalman filter is used to extract the active scanning component of the UAV in the motion vector, which performs motion compensation to obtain the result of image stabilization. The standard test atlases and the UAV aerial video collected by researchers serve as the experimental data.The experimental results show that the proposed algorithm is effective and fast in the processing of the continuous video sequenceand can meet the needs of real-time image stabilization.decision tree; scale space; motion filtering; Kalman filtering

In order to solve the problem of high complexity and low accuracy of the current method for Volterra frequency-domain kernel identification, a method for Volterra frequency-domain kernel identification based on neural network is proposed. Firstly, the amplitude of each Volterra frequency-domain kernel is accurately measured after choosing multiple frequency components. Then, we use the characteristics of BP neural network that it can approximate nonlinear functions to design different models for different-order Volterra frequency-domain kernels, so as to identify each kernel. Finally, a nonlinear circuit is adopted for simulation. The results show that this method can directly identify all the Volterra frequency-domain kernels in the frequency range, and the process is simple with high accuracy, which is suitable for engineering realization.

To address the practical problems in the prediction and assessment of cyberspace state, this paper studies the prediction of variables in cyberspace, and presents a state prediction method based on Non-Homogeneous Dynamic Bayesian Networks (NH-DBNs), so as to solve the problems in the existing model of single time-sequence regression and the ignorance of the inter-influence between multiple variables.The presenting form of the network state is given, and the method and process of state prediction are presented. The comparison experiment conducted in the same environment has proved that the algorithm accords with the practical application environment, and can improve the prediction accuracy of the indexes of the cyberspace state. The algorithm has certain value for the conductor to make scientific, reasonable and accurate decisions.

The infrared imaging technology is finding increasing applications in various fields with the rapid development of science and technology. However, the images after infrared imaging are relatively vague with great noises, which seriously affect the target detection and recognition effects.It is necessary to denoise the image effectively. On the basis of studying Double-Tree Complex Wavelet Transform (DT-CWT) and morphological filtering, we proposed an infrared image denoising method based on the MAP estimation of DT-CWT and morphological filtering, and applied it to the infrared image denoising.In comparison, the proposed method is superior to the traditional denoising methods both in terms of visual effects and objective evaluation indexes, and can be used in the field of infrared image processing.

By establishing the model of the submarine, and the models of the fixed-wing anti-submarine aircraft and of its typical flight route for deploying the encirclement arrays, simulation modeling is implemented for the submarine escaping and sonobuoy array deploying in anti-submarine combat. By using the simulation result, the influence of the anti-submarine distance, the submarine speed, and the type of sonobuoy array on the submarine searching efficiency is studied. Based on the analysis to the probability distribution of each escape direction of the submarine in the case of searching failure under high search probability condition, the directional characteristics of different anti-submarine sonobuoy array are obtained.

Convolutional Neural Network(CNN) usually adopts a single network for feature extraction. However, the extracted features are not sufficient, which may result in the poor accuracy in image classification. To solve the problem, it is proposed to use two networks for extracting features simultaneously. Then the two networks are cascaded together to obtain the fused features of the two networks, which makes the extracted features more discriminative.The dual-network cascade uses two network branches for feature extraction. One branch is the traditional CNN. The other branch is the traditional CNN plus the residual operation.Before the next dimensional reduction of the feature map, the two different branches are put together.We use the data sets of 101_food and caltech256 to test the networks. The cascaded network is compared with the two separate branches, and the results are favorable.

To solve the problems that the pixels in the millimeter-wave holographic imaging algorithm have a certain degree of divergence and that the resolution sharpness is not high enough, a Stolt interpolation optimization algorithm is proposed. According to the distance between each data point in the cell and the center, a weighted average value is used for interpolation. Compared with the traditional interpolation algorithm, the algorithm makes each pixel further converge, and the imaging clearer, while the amount of computation will not be greatly changed. Finally, the cross-section simulation experiment proves the effectiveness of the algorithm.

An attitude estimation algorithm is designed by using Gaussian Particle Filtering(GPF）,algorithm.The quaternion discrete equation is taken as the state equation. The algorithm consists of five modules:sampling conditioning particle, sampling particle, weight calculation, mean covariance calculation and Cholesky. The “average” value and covariance matrix of the quaternion are calculated by using non-standardized weights, and the calculation formula of the covariance matrix is rewritten.In this way, the pipeline Gaussian particle filter algorithm is implemented. At the same time, a parallel design scheme is proposed, which can further optimize the running speed by using the remaining resources of FPGA. The simplified particle filter algorithm and Gaussian particle filter algorithm presented here can not only be used in UAV attitude estimation, but also be applied to other nonlinear estimation problems and applications. The simulation results have proved the feasibility and effectiveness of this design.

The theory and application techniques of the stretched chirp-step signal synthesizing are investigated, and the possible problems while applying this new technique to the radar system are studied thoroughly. The corresponding solutions to these problems are provided as well, including RVP compensation and system tracking error compensation. The design requirements of the system are also put forward, such as the design constraints of the system parameter and the echo SNR.On this basis, the processing procedure of stretched chirp-step signal synthesizing is presented. Simulation results demonstrate the efficiency and validity of the algorithm and the processing procedure.So the problems of how to apply the theory of this new technique are solved. This technique can also be used for the generating and processing of wideband/ultra-wideband signal.

An optimization control algorithm based on the combination of inversion adaptive control and classic PID control is proposed for the optimization of the quadrotor attitude control algorithm.The algorithm uses the controllers designed by different control algorithms in combination. The control system is divided into two loops: the attitude angle and height channel are inner loops, and the plane channels x and y are outer loops. The simulation results show that the inversion adaptive PID control algorithm has a very good control performance. Compared with other control algorithms, it has a great improvement on stability, response, anti-jamming ability and accuracy.

In view of the inefficiency of airborne weapon integration, this paper states the status quo of airborne weapon integration at home and abroad, analyzes the principle of Universal Armament Interface (UAI) technology, and proposes a method to improve the integration efficiency based on the functions of Stores Management System (SMS). A weapon control system is developed based on the UAI technology, and a comprehensive demonstration and verification platform is built.This platform verifies that the weapon control system can integrate new weapons and improve integration efficiency without modifying the software of the product, which is important for our country to realize Plug-and-Play (PnP) in the field of airborne weapon integration.

The complexity of the avionics environment makes the traditional data bus unable to meet the transmission requirements of real-time messages. With the excellent performance of the FC protocol, an FC protocol chip is designed based on the FPGA platform, and a logic design is implemented in detail on the main modules.The results from simulations and development board tests in the Windows7+ Modelsim10.1c environment show that the chip can meet the transmission requirements of real-time messages, implement protocol functions, and meet the application requirements of the avionics system.

Full-color holographic waveguide display usually adopts the complex coupling structure of multi-layer gratings or the combined structure of multi-layer waveguide, which results in heavy weight and large volume of the system. To solve this problem, a configuration of dual-color holographic waveguide display with single plate and mono-layer grating is proposed. This configuration adopts three gratings to realize the input and output of the light and the two-dimensional expansion of the exit pupil. To display the red and green color with a better uniformity in the wide field of view, the shape and films of grating are designed and optimized by using the theory of Rigorous Coupled-Wave Analysis (RCWA).

Based on the standard of the new-generation Avionics Digital Video Bus (ADVB), an airborne aeronautical chart assisting video transmission system was designed. In this paper, the ARINC 818 protocol is described in general. Then, the system design principle is presented according to the latest protocol standard. Finally, the detailed scheme for the auxiliary video transmission system of the electronic aeronautical chart is given. The system is designed with the modular and hierarchical design method. In this system, the protocol conversion of ARINC 818 is implemented by using FPGA, and the function of optical-electro, electro-optical conversion is realized through the optical module. The real-time HD avionics video data transmission is implemented with the video frame rate as 4.25 Gibit per second, and the resolution as 1920×1080P@60 Hz, 24 bit.The system design has been successfully applied to the integrated data management system onboard a certain type of modified aircraft, and has an important practical value.

The IMA platform has such characteristics as time correlation, function dependence, multiple states and numerous parts. To solve the difficulty of the IMA platform in reliability analysis, a method for Discrete Event Evolution Tree (DEET) simulation analysis was proposed.In the running process of the IMA platform, the random failure event was transformed into a discrete failure event sequence, so as to drive the constant evolution of the system state. The evolution tree of the failure event was used to analyze the impact of the discrete failure event on the system state. Then, the multi-state sets and large sample observation were set up to evaluate the probability of the IMA platform being in different states at different time. The DEET method can quantitatively analyze the multi-state reliability of the IMA platform and serve as a theoretical data basis for further design.