Aiming at the appearance model updating problem of visual tracking,an adaptive model updating strategy was proposed based on particle information criterion updating in the framework of the particle filter visual tracking.By mining the related information between the particles, analysis was made to the changes of tracking status and appearance model, and then a fuzzy rule table for model updating was built up.The method was compared with the model updating method of classical particle filter visual tracking by using the famous visual tracking evaluation data sets.The experimental results demonstrate the effectiveness of the algorithm.

The autonomous collision avoidance problem of Unmanned Aerial Vehicle (UAV) against non-cooperative dynamic obstacles is studied, including the decision making and control.First of all, the autonomous collision avoidance problem is described, and it is divided into two parts of collision prediction and avoidance control.Then, the rules of the collision prediction are defined, and a three-dimensional algorithm is proposed for dynamic collision prediction based on the two-dimensional collision prediction.The collision avoidance strategies are designed finally by using the optimization theory.Simulation results demonstrate the validity of the proposed method.

In order to solve the problem that the Target Birth Intensity (TBI) is unknown in multi-target tracking, a measure-driven PHD algorithm is proposed for TBI estimation.By using augmented state space method, PHD multi-target tracking is achieved in the augmented state space composed of the real-target state space and the spurious-target state space (clutter space).By constructing the newborn target measure set, the TBI is estimated with the measure-driven method, which avoids both the dependence on the prior knowledge of TBI and the interference of unknown clutter to the real-target intensity estimation.Simulation results show that the proposed algorithm is sensitive to the change of target number and can reduce the computational complexity, which improves the accuracy of tracking obviously.

In order to preserve the feature of salient edge for pre-processing of infrared images, an edge-preserving smoothing algorithm fused with L0 norm of global gradient and local saliency is proposed.First, Laplace transform is made to the infrared image, local edge saliency is estimated rapidly by the nested rectangles, and the weight function is designed by using mean and variance.Secondly, significance of each pixel in the approximation term is adjusted based on the weights, to restrain L0 norm of global gradient and build the optimization model.Finally, the model is solved by alternating direction method of multipliers, the infrared image is smoothed while preserving the local salient edges.Experimental result shows that:Comparing with other edge-preserving algorithms, our algorithm can smooth infrared images while preserve local salient edges effectively, which supplies a theoretical support for further application.

The object tracking method based on Circular Structure with Kernels (CSK) is only suitable for tracking objects with fixed scale.To solve the problem, a new patch-based scale adaptive CSK tracking method was proposed.First, the relationship between the position of patches and the scale of object was derived from spatial structure of each patch.Then, the tracking confidence level of each patch was evaluated, and those with high confidence levels were selected for integration.Therefore, effective estimation was made to the scale of the object through independent tracking of each patch and the integration.The robustness of the method under interference was improved.Experiment was made by using the typical vision sequences.The result showed that the algorithm can not only realize real-time tracking of the change of object scale,but also has higher accuracy and robustness compared with the original CSK method.

Aiming at the problem that the complementary filtering in the integrated navigation system could not adapt to the motion state change and the magnetic field disturbance, a multi-sensor complementary filtering algorithm with variable parameters is designed.Firstly, the sensor noise is analyzed in frequency domain, and the complementary filter is designed according to the spectral characteristics.After that, the motion state and magnetic field are monitored through accelerometer and electronic compass, according which to set the filter parameters to decrease the estimation error.Finally, with FPGA as the core processor, an integrated navigation system of MEMS gyroscope, accelerometer and electronic compass is developed to verify the algorithm.The experimental results show that:Compared with the commercial AHRS, the algorithm can keep a higher system output accuracy, and reduce the influence of the motion state change and the magnetic field disturbance effectively.

A fusion algorithm of pose estimation based on hierarchical cooperation objective is studied for visual landing of the rotor UAVs. First, the target area is determined by using the properties of the cooperation objectives that they have the same mass center and the fixed area ratio. Then, the LSD line extraction method is used to get the coordinates of the feature points. The Tsai algorithm is used together with fast four-point estimate algorithm to estimate the position and attitude of the UAV. Experimental results show that the hierarchical objective can increase the application range of position estimation and improve the precision of pose estimation.

In civil aviation and other application fields, the availability of satellite navigation system is the key performance indicator for whether it can be taken as the only or the main navigation system.This paper mainly evaluates and analyzes the availability of Space-in-Signal (SIS) of GPS satellite navigation system.A model of single satellite availability algorithm based on Markov process is presented, and the model for evaluating availability of navigation satellite constellation is studied.In addition, based on the relevant data and the availability index of the latest GPS Standard Positioning Service Performance Standards (GPS SPS PS), it is verified that the model is effective and conforms with the availability index of GPS SIS.The result has great reference value to the study of the availability of BeiDou navigation system.

The characteristics of high grey-scale and integrality of the airport runways are used for detecting them in remote sensing image.Firstly, the image was pre-processed by using fuzzy enhancement.Secondly, binary image was obtained though threshold segmentation, and pixel labeling method was used for connected area extraction, thus the largest four areas were chosen as suspected airport areas, which were then located.Then, within these four suspected airport areas, edges were detected on the image by using Canny edge detection operator, and straight line segments were extracted by using Hough transform.Finally, the area with the longest parallel lines was recognized as the airport area.Experimental results show that by using the airport runway detection algorithm, the airport runways in remote sensing image can be detected efficiently and accurately.

To solve the problem that the new track can not be linked to the preceding track when the track breaks off in ballistic missile's unpowered phase, a method based on fuzzy theory and evidence combination is presented for interrupted track association.Firstly, the tracks before breaking are predicted according to missile prediction model, and those after breaking are filtered by missile filtering algorithm.The new tracks above are regarded as targets waited association, the tracks in the same time segment are regarded as association samples.Then, fuzzy processing is made to characteristic distances of sample points at the same moment from different tracks in order to obtain the association membership degrees, which can be transferred into Basic Probability Assignments (BPAs) in evidence theory, so the evidence combination result in each sample instant can be obtained.According to the combination results during a continuous time, the whole association result is obtained.The simulation results show the effectiveness of this approach.

The purpose of this paper is to improve the stabilization accuracy of airborne electro-optical stabilized platform.Analysis is made to the working principle of the stabilized platform, based on which a stabilized control loop is modeled.The factors that have effect on the stability accuracy and the deficiencies of classical control algorithm are analyzed briefly.An improved algorithm is then presented based on Active Disturbance-Rejection Control (ADRC).At first, the model of motor is simplified.Then, an extended state observer is designed to estimate the disturbance in real time and a feedback control law with disturbance compensation is designed.Eventually, simulation is made to test the ability of active disturbance rejection controller with comparison with the existing classical control algorithms.The result proves that:1) The ADRC algorithm can reject the nonlinear disturbance effectively; and 2) Compared with the classical algorithm, it has stronger anti-jamming capability, and higher velocity/LOS stabilization accuracy.

Aimed at the aerial threats in systematic combat of multi-platform, multi-target under complicated environment, the design philosophy of multi-missile cooperative operation is proposed with its key technologies.First, analysis is made to the application and development of the multi-missile cooperative guidance in the world.Then the combat system and operation tactics of multi-missile cooperative combat for air defense are discussed.Based on which, the design concepts of the key technologies of multi-missile cooperation for air defense are proposed, including such technologies of overall technology of multi-missile cooperation, flight control of multi-missile formation, data chaining, and confirmatory experimentation.Some ideas about the development of multi-missile cooperative guidance are given in the end.

In order to effectively suppress the possible common mode failure of computer system software and hardware, analysis is made to dissimilar redundant technology and its redundant architecture and the number of redundancy based on the redundancy technology.The design and implementation of a dissimilar dual redundant alarm computer are discussed from such aspects of the hardware/software architecture design, hot-standby switching logic and so on.The fault tree analysis method is used for analyzing and calculating the reliability of the system based on the experimental data.The results show that: The use of dissimilar dual redundant technology can effectively suppress the common mode failures of software and hardware system, and improve the reliability of the airborne warning computer system to 0.999 97.

To the servo system of stabilized platform working in a severe environment,Fault Tree Analysis (FTA) method is used to determine the fault classification and the logical relation.Firstly, based on rough set theory, the original fault decision table is constructed.Discernibility matrix and genetic algorithm are used together for reduction of it.Then, the table after reduction is used as learning samples for training Elman neural network to generate fault diagnosis model.Finally, test samples are used to verify fault diagnosis model.The correct rate of fault diagnosis reaches 98%.It shows that this method is feasible, and it has a certain guiding significance to the fault diagnosis of servo system with complex fault model.

To study the impacts of environment luminance, chromaticity of background and interferences (including specular reflection, diffuse reflection and glare) on the readability of Head-Up Display (HUD), a readability evaluation model was established based on Perceptible Just Noticeable Difference (PJND) theory by taking the characteristics of HUD and its environment into consideration.Simulations and experiments result shows that:1) The value of PJND can be used for evaluating the readability of HUD;2) Low-luminance environment has a higher luminance contrast requirement than high-luminance environment;To satisfy the readability, the HUD needs to lower the luminance contrast when the ambient luminance increases;3) The bigger the difference between background chromaticity and object chromaticity, the lower the requirement to luminance contrast;and 4)When the environment luminance is constant and the interferences increase, a higher luminance contrast can be used to satisfy the readability;when the interference luminance is beyond the maximum, physical occlusion is a way to satisfy the readability.

In orderto study the visual effect of the external environment color on the information presented on Head-Up Display (HUD), a simulation system is developed to evaluate the performance of color scheme.At first, the IData software is used to model the flight navigation display interface, and the color character block of HUD is set up.Then, the HUD picture generated by the model is introduced into the simulated scene for flight simulation test.JSBSim model is used for flight operation data calculation.In the test process, different color of HUD character block is used for different visual scenes to inspect the simulated operating performance.The OSG visual simulation software package for high fidelity visual simulation development is used in the simulation system, thus the system development cycle and difficulty are reduced greatly.A number of pilot flight simulation tests show that:1) The displayed picture has no fluttering and no distortion; and 2) There is no obvious visual delay in the color changing process, which can meet the needs for test.

Time-frequency spectrum analysis can provide the signal features in time and frequency domains.However, the method is difficult to make a good tradeoff between high time-resolution and high frequency-resolution in the process of signal identification.Therefore, a new algorithm is proposed, which has a good time-frequency concentration.Firstly, the signal is decomposed into different frequency bands by harmonic wavelet packet.Then, Morlet wavelets with high time-resolution, and short-time Fourier transform with high frequency-resolution, are used to analyze the time-frequency distribution of each frequency band.Therefore, Morlet wavelets scale matrix and short-time Fourier time-frequency matrix are obtained.At last, these two matrices are fused into a new matrix with a specific algorithm.Analysis is made to accelerometer dynamic signal under simulation and dynamic condition.The result shows that the fusion algorithm can not only extract the weak dynamic change, but also has high time-frequency resolution, which is an intuitive, comprehensive, and accurate signal identification algorithm.

To solve the problem that the helmet-mounted display is not so comfortable for wearing, we put forward a method to replace the cable with wireless communication technology.Taking the small cockpit environment as the object for study, we analyzed the transmission characteristics of the wireless signal in the cockpit space, and proposed a new Doppler shift expression based on Jakes model together with the head movement regularity of the pilot.Simulation was made with Matlab to analyze the performance of the improved Jakes model.The simulation results show that:the autocorrelation function curve and cross-correlation function curve of the improved Jakes model is more close to the theoretical value, which has better second-order statistics, and is applicable for modeling of the wireless channel.

The establishment of taxiing model for the Unmanned Aerial Vehicle (UAV) is the basis for control law design and real-time simulation system development of the UAV's taxiing in take-off and landing.We established the landing gear model and tire model of a wheeled UAV.The quality of UAV was divided into elastic support quality and rigid support quality.Analysis was made to the dynamic characteristics of the two parts in the UAV's taxiing phase, and the taxiing model of a wheeled UAV was established.Simulation tests of accelerated taxiing (take-off) and decelerated taxiing (landing) were made to the model, and the result show that the model is correct.

There are many industrial systems that may experience operating state switches between storage and usage throughout their whole life-cycle, and it has been proved that the operation switches have great influence on equipment's degradation process.However, the issue of remaining useful life estimation for such kind of systems is seldom founded in the literatures.In this paper, some key problems are systematically analyzed, such as the stochastic characteristics of the operation state switches, the lasting time of each state, and the difference in the degradation rates in each state.Particularly, a two-state Continuous-Time Markov Chain (CTMC) is presented to approximate the switches between the states of usage and storage.Then the CTMC is integrated into the random coefficient regression model to realize the remaining life estimation.A simulation experiment shows that the presented approach can effectively realize the system's remaining life estimation under the condition of the operation state switches.

The conventional residual chi-square test can not accurately detemine the fault duration and thus is easy to cause false alarm and alarm leakage.Aimed at this problem, a fault detection and system reconstruction method based on hierarchical filtering is proposed.The residual chi-square test is used for fault detection of main and sub filters, and the specific failure satellite can be identified by comparing the detection results.Then, the filtering results of the sub filter uncontaining the failure satellite are feedback to the main filter and other filters and replace the results of them to reconstruct the system.The results show that: Under the condition of single satellite failure, the novel algorithm can reduce the fault tracing function, determine the failure satellite and fault duration more accurately, and improve the integrity monitoring performance significantly.