The carrier phase variance components from different receivers and different types of satellite were estimated based on the sigma-ε model. The results implied that the variation of these components was closely related to the types of the receivers and satellites. In order to establish more realistic stochastic model, this paper combined the least square method with MINQUE method to estimate the variance components from different type of satellites in real-time.Experiments were carried out to evaluate its performance.The results showed that: 1) In short baseline case, this method is able to improve the solution accuracy and stability of relative positioning;and 2) Compared with empirical models, the azimuth accuracy and range are improved by about 10%, and the accuracy and range in North, East, and Up directions are also improved by about 5% to 10%.

Aiming at the problem of optimal resource allocation of early warning satellite constellation system for ballistic missiles in China, a multi-target satellite constellation resource optimization configuration scheme based on dynamic programming method is proposed. This method is essentially a nonlinear programming method. The principle can be reduced to a basic recursive relationship. The constellation of the ballistic target observed in a single frame of the early warning satellite constellation is taken as a resource to be allocated, and the constellation scheme evaluation function is established to minimize the possibility of losing the ballistic target, thereby achieving the goal of not losing the target or state information. Experiments show that this method is effective in satellite constellation resource allocation. Compared with other algorithms, this method can make full use of the satellite constellation resources.

With the increasing complexity of military missions, the future combat system is a multi-platform joint operation that requires the integration and sharing of multiple datalink resources. The development of Software Defined Network (SDN) provides a new design idea for the global management and control of multi-datalink network. A multi-datalink network control architecture based on SDN is proposed by fully considering the network resource fusion of multiple links.A Node D emand based Sub-domain Partitioning(NDSP) algorithm is proposed through datalink sub-net mapping, which realizes the rational partitioning of physical sub-domains. Then, the improved multi-objective simulated annealing algorithm (ISA) is used to select the central interactive nodes, and thus the network is globally controlled.The experimental results show that the NDSP algorithm has better applicability for sub-domain partitioning in the multi-datalink fusion network, and the ISA algorithm can significantly decrease the average delay and average failure rate of network interaction path.

Route planning of anti-ship missile in coordinated attacking is one of the specific problems of route planning. The special requirements of anti-ship missile for route planning in coordinated attacking are analyzed, and the constraints of the anti-ship missile itself on route planning are discussed. Considering the above characteristics, a sort of rapid method for anti-ship missile route planning is put forward based on geometric principle. The route planning method is presented here, and simulation results have proved that the proposed method is efficient for route planning of anti-ship missile in coordinated attacking.

On the basis of aerodynamic calculation and accelerometer-based method for estimating the angle of attack of the guided munitions, this paper studies the angle of attack identification of a miniature guided munition. By establishing an aerodynamic model, the aerodynamic and balanced angle of attack data under the typical flight conditions of the guided munitions are calculated out. Based on the aerodynamic characteristics, an angle of attack identification model is established, and a layout scheme of the missile-board inertial measurement units is proposed. Combined with the test data, the angle of attack identification results of the guided munitions are calculated out. The comparison with the simulation results shows that, the angle of attack identification based on flight test data is consistent with the aerodynamic simulation results, which proves that the proposed angle of attack identification method is reasonable and feasible.

The standard Q-Learning algorithm based on Markov decision process in reinforcement learning can obtain an optimum path, but the method has the shortcomings of slow convergence rate and low planning efficiency, and thus can not be directly applied to the real environment. This paper proposes a Q-Learning path planning algorithm for mobile robots based on the potential energy field knowledge.By introducing the potential energy value into the environment as the search heuristic information to initialize the Q value, the rapid convergence of the mobile robot can be guided in the early stage of learning, and the blindness of the traditional reinforcement learning process is avoided, which makes it suitable for direct learning in a real environment. The simulation result shows that: Compared with existing algorithms, the proposed algorithm not only improves the convergence speed, but also shortens the learning time, which can make the mobile robot find a better collision-free path quickly.

Aiming at the contradiction between the capture speed and the capture accuracy of fine acquisition of GPS signal, a GPS fine signal acquisition method based on orthogonal search is proposed. The code phase is firstly removed according to the coarse acquisition result of the GPS signal, and then a set of non-orthogonal sinusoidal candidate functions is constructed according to the coarse capture result of the frequency and the coarse capture resolution. Finally, Gram-Schmidt orthogonalization method is used to achieve the fine acquisition of frequency. Experimental results show that the frequency acquisition performance of the proposed method is better than that of the commonly used zero-filled FFT method, which balances the contradiction between capture speed and capture accuracy and is suitable for current mobile navigation and positioning equipment applications.

In view of the facts that the traditional “true or false” evaluation method is too rough and the missile test data is uncertain, the state of missiles is re-classified into five levels, namely “perfect”, “good”, “usable”, “pseudo-failed” and “failed”. A missile state evaluation method based on cloud model and Bayesian network is proposed.The cloud model is used to obtain the membership relationship between the performance test parameter and each state level. The Bayesian network is then applied to establish the multi-parameter state fusion evaluation model.The DS/AHP method is introduced to carry out the conditional probability assignment for reducing the uncertainty in the expert inference process.

For the conventional GNSS receiver, the carrier tracking loop is prone to lose lock in high dynamic environment. Under this condition, the receiver cannot work properly and provide accurate navigation information. To improve the performance of the carrier tracking loop under high dynamic circumstance, this paper proposes a new carrier tracking algorithm combining Phase-Lock-Loop (PLL) with Frequency-Lock-Loop (FLL) based on the support vector machine. A judge factor is introduced to define the state of the carrier tracking loop. The weights of the PLL and FLL are adjusted in real time according to the value of the judge factor. The support vector machine algorithm is employed to establish the nonlinear function of the judge factor and fusion weights of the loops. The simulation results show that this method can greatly improve the performance of the tracking loop in high dynamic environment.

In order to evaluate the visual fatigue caused by 3D equipment observation more effectively, the dynamic Bayesian network is used to calculate the stereoscopic visual fatigue probability of 3D observers for the first time. In the process of constructing a directed acyclic graph, the relationship between multiple factors in stereo vision and fatigue phenomena is taken into accountthe physiological characteristic nodes and dynamic factors are added for making reasonable estimation. The state of each node is in one-to-one corres- pondence with the state probability of Bayesian network nodes, which provides a systematic scheme for stereoscopic fatigue probability assessment. The results show that: Compared with current MOS method, the scheme of dynamic Bayesian network analyzes the observation more comprehensively, the fatigue state of the observer is more accurate than the subjective result of the observer himself, and is closer to the actual situation.

The overall scheme and control method of baseband full digital TACAN signal simulator based on PXI bus interface are proposed. The FPGA chip is used to complete the baseband signal generation and various interface control circuits by digital method. The PCI9054 interface chip is used to realize the protocol conversion between the PXI bus and the local bus. The system software is developed in the Quartus II integrated development environment using the Verilog language, which is then downloaded to the FPGA chip. The results show that the simulator can meet the requirements of the automatic test of the orientation, distance and sensitivity of the TACAN airborne equipment, and improve the test efficiency. The simulator solves the problems that the traditional simulator is large in volume, low in precision, difficult to integrate and low in automatic test efficiency.

In order to solve the problems of low automation level, and low efficiency, accuracy and reliability of the existing manual operation mode and mine-laying aided decision-making system in the formulation of mine-laying scheme, a new method for fast automatic generation of mine-laying scheme is proposed.This method designs a divergence model of port entrance and main channel, with consideration of the impact of various marine environmental factors and hostile anti-mine threats.Combined with the theory of solution of geodetic problem and water depth interpolation and extraction technology, an accurate mine-laying scheme for port blockade is produced quickly and automatically. The simulation experiment results show that the mine-laying scheme generated by this method conforms to the actual situation of sea area and has preferable accuracy, reliability and practicability.

Considering that the off-line computation of explicit model predictive control for complex fast multi-dimensional constrained systems has the shortages of large amount of computationslow processing speed and high memory usage, we proposed an efficient explicit model predictive control algorithm based on separating saturated region. By combining polyhedron, piecewise affine related concepts and Polya theorem and using the feedback saturation characteristic of explicit controllers, the saturated region was removed and the unsaturated critical region was retained, so as to reduce the number of state partitions and computational complexity. Then, based on the state space simplified model of two subsystems of the decoupled quad-rotor aircraft, we realized rapid and accurate adjustment of attitude angles. The effectiveness and superiority of the proposed algorithm in attitude control of the quad-rotor aircraft have been verified by numerical simulation and hardware-in-the-loop simulation experiments.

Aiming at the target extracting, matching and relative pose/position calculation of cooperative target vehicle, the information of gray-image and the depth-image collected by Time-of-Flight (TOF) camera is used for target detection and key-point selection, based on which the feature point matching is implemented and the relative pose/position of the cooperative target is determined. First, the gray information is used for target detection. Then, the critical feature points are selected based on the depth information and morphology of the target, and the deterministic annealing algorithm based on MRF is used for feature point matching. Finally, the Singular Value Decomposition (SVD) algorithm is used to estimate the relative pose and position between the target vehicle and the tracking vehicle. The experimental results demonstrate the effectiveness and the robustness of the proposed procedure.

When Electro-Optical Tracking System (EOTS) is working in a low-speed tracking state, the nonlinear friction torque between shafts of the EOTS may cause “flat-top” or “creeping” phenomenon, and also decrease the angular resolution and repeatability precision, which has a great influence on tracking precision. A friction modeling and compensation method based on Genetic Algorithm(GA) is proposed to solve the problem. Firstly, GA is adopted to model a Stribeck nonlinear friction of the EOTS to obtain a higher degree of model fitting and reduce the fitting error of the friction model. Then, a feedforward compensation is made to the identified friction, and thus to reduce the effect of the nonlinear friction torque on the EOTS.Results in simulation and physical experiments show that the proposed scheme is effective in friction compensation, which can improve the control performance of the EOTS.

To the characteristics of large error in sea target positioning, the method of Multiple Rounds Simultaneous Impact (MRSI) of guided projectile can effectively improve the hitting accuracy and firepower density. The shipborne gun with controllable muzzle velocity can adjust the muzzle velocity of guided projectiles, correct the trajectory of the projectile, and attack the target by using the preplanned guidance law. The models of projectile exterior ballistic and the optimum guidance law with angle constraints are established, and the method for calculating the predicted point of sea target is proposed. According to the guided projectile trajectory characteristics, an analysis is made to the fire control method of the shipborne gun with fixed muzzle velocity, based on which the fire control method of the shipborne gun with controllable muzzle velocity is studied. The simulation results show that the controllable muzzle velocity shipborne gun can greatly increase the number of simultaneous projectiles, and enhance the firepower density.

Rotary-wing flight robot is a novel aerial robot system containing a rotorcraft and a manipulator. To implement the operation when interacting with the environment, an object operation control strategy with the ability of disturbance rejection in task space is proposed. Firstly, the kinematical model and dynamical model of the rotary-wing flight robot are established in the generalized coordinates. And the relation between inputs and outputs is determined. Then, the robot system is divided into three control loops of position, attitude and manipulator loop.Linear Active Disturbance Rejection Control (ADRC) algorithm is employed to design the controller for each control loop. In the control structure, the linear extended state observer and PD control law are used to estimate and compensate the lumped disturbance. Moreover, the stability analysis and parameter tuning are conducted for the control algorithm. Finally, the object operational control experiment is executed on the rotary-wing flight robot system. All the results show that the proposed control algorithm has strong ability of disturbance rejection and fast response speed, which can keep the stability of position/attitude of the rotary-wing flight robot in operation.

Aiming at data imbalance in SAR image recognitiona two-stage transfer learning method based on intra-batch balanced sampling and model fine tuning is proposed. First, the training set of balance data between classes is obtained by using the intra-batch balanced sampling method. Then the training data set is used for pre-training of the model. Finally, the training of imbalanced data and the test is completed by transfer learning and model fine tuning. Aiming at the problem of redundant parameters in the processing of single channel SAR images by using general tri-channel CNN model, a lightweight CNN model is designed for SAR image recognition. Through the three strategies of single-channel convolution kernel, deep separable convolution and using global average pooling instead of full connected layer, the parameters of the model are reduced greatly. The results of experiments conducted on the open data set OpenSARShip show that: 1) The proposed method effectively improves the recognition accuracy of minority classes and reduces the effect of data imbalance on recognition results;and 2) The proposed lightweight CNN model can reduce the size and the single-iteration time of the traditional tri-channel CNN model by about 58.86% and 63.62% respectively, under the premise that the recognition accuracy is basically unchanged.

Considering the problems of strong coupling, underactuation, nonlinearity and parameter perturbations in the system of quadrotor Unmanned Aerial Vehicles (UAVs), the Cerebellar Model Articulation Controller (CMAC) is introduced into the system and combined with the traditional PD controller to improve the static and dynamic performance of quadrotor UAV. The traditional PD controller is used to achieve feedback control of the height, and CMAC is for the feedforward control to realize inverse control of height mathematic model. The simulation results prove that: Compared with the traditional PID controller, our control method has less overshoot, faster response, better stability and stronger anti-jamming ability.

Aiming at the problem of difficulty in updating the algorithm software of embedded atmospheric sensing system, a data solving card is designed, which can update program in application.With DSP+FPGA structure as the core, the data solving card ensures the I2C transmission process is not interrupted by use of special interrupt control logic. The sector structure of Flash is used for reliable implementation of in-application programming. The simulation results show that the data solving card can realize the in-application programming of the calculating software in a simple and quick way, and meet the running requirements of common embedded sensor system algorithm software.

To solve the problems of poor efficiency of production rule in big data and hard weight distribution in the case of heterogeneous information fusion, a knowledge representation method of production rule based on echelon inference and benefit principle was designed taking the data and voice as the information source for fusion. A grouping & ranking inference strategy was proposed on the basis of different sensitivity between mapped knowledge elements and unmapped knowledge elements. The weight of knowledge element was specified by the benefit value obtained from mutual information and information gain. We simulated the engine fault by using the flight data and the synchronous cockpit voice as the information source for making failure prediction. The result shows that our knowledge representation method can provide higher efficiency, stronger intensity and more accurate diagnosis result, which can be used in practical fault diagnosis and fault prediction.