Driven by the technological advancements and military demands， unmanned aerial vehicle （UAV） swarm warfare has emerged as a disruptive new style of warfare that can change the rules of engagement. Fierce competition has unfolded among major military powers around "swarm technologies and tactics". The concept and advantages of UAV swarm operation are introduced， and the development of the swarm operation types abroad is summarized. Based on the shortages and restrictions of conventional military equipment， the anti-swarm operation and explores the development direction of anti-swarm equipment and technology is analyzed. A dynamic kill chain construction strategy is put forward to solve the problem of inadequate application of adequate anti-swarm approaches and theories， and achieve systematic confrontation by comprehensively using a variety of anti-UAV swarm equipment in the future，laying the solid foundation for the development of the systematized anti-swarm operation.

Without a unified top-level design of the system of systems （SoS） construction， the elements of the SoS will fight against each other， and hence the SoS will inevitably face problems such as inconsistent indicators， mismatched functions， and irregular integration. In order to solve these problems， it is necessary to adopt SoS engineering method and carry out forward design. The types， characteristics and the construction requirements of the SoS in the military domain are analyzed. Aiming at realizing the combat ability， the model-based SoS engineering construction method is given， according to the four stages of “SoS design， SoS construction， SoS application， SoS evaluation”. Employing digital means to support SoS research and application is introduced. By using digitization， SoS engineering， processes and tools are integrated to create a new SoS engineering construction mode.

Adaptive nulling anti-jamming technology can greatly improve the anti-jamming ability of satellite navigation receiver. It is widely used in precision guided weapons. It is the focus of current research on satellite navigation counter-measure. This paper simulates the anti-jamming performance of multi-element adaptive nulling receiver in the face of jamming sources with different numbers， different incident azimuth angles and different incident pitch angles， quantitatively analyzes the improvement of the anti-jamming ability of the receiver， and puts forward targeted counter-measure schemes. Based on the experimental results and data， it is found that by adjusting the number of interference sources and the setting angle， the anti-jamming ability of the receiver can be effectively suppressed， so as to achieve a reasonable allocation of jamming resources. The conclusion has a strong practical guiding significance for the further study of the optimization strategy of jamming source deployment.

In order to achieve the high accuracy and long endurance of underwater vehicle， it is essential to apply the technique of error suppression and correction based on strap-down inertial navigation system/ Doppler velocity log （SINS/DVL） integrated navigation. This paper introduces different alternative methods when Doppler data fails， and discusses SINS/DVL data fusion filter techniques. It illustrates the damping algorithm， terrain assisted correction， gravity matching assisted positioning， long baseline system assisted positioning and gravity disturbance compensation to further suppress or correct the navigation errors accumulated over time. This paper provides reference for underwater autonomous navigation with high accuracy and long voyage time.

Based on the analysis of the concept of military meta universe and related technologies，this paper proposes the idea of building a military ecosystem by applying high and new technologies and major projects such as virtual/reality fusion interaction，network information system，etc. The advantages of building cloud services in the military ecosystem are analyzed， which can provide platform and integrated processing and shared transmission services for battlefield information support. In order to adapt to future combat scenarios and missions， the air combat force has both the operational requirements of connecting the network and building the cloud， and the advantages of networking and building the cloud. This paper analyzes the basic concepts and main content of battlefield situation information， focusing its role in all aspects of air combat， and points out that air combat based on real-time battlefield situation information can achieve the goal of Find and Destroy. The air combat cloud kill pattern in the military ecosystem is designed and the method of closing the air combat kill chain based on battlefield situation information is proposed， so as to provide reference for advancing the development of the modern joint operation theory of our army and improving the joint air combat capability.

In order to improve the association accuracy between radar target and IFF plot in complex environment， a method based on gray theory is proposed. In different scenarios， it can be judge whether there is competition among multiple radar targets when it is associated with IFF plot by calculating the gray correlation degree. If there is， the historical information of association between radar target and IFF is used to modify the gray correlation degree. Finally， the associated decision is completed based on the recognition of maximum gray correlation degree to form the identification of the target’s friend and foe attributes and the identification confidence estimation. Simulation results in typical scenarios show that， through multiple interrogations and association with the target， it can improve the association accuracy in different environment and the confidence degree of cooperative target attribute determination.

Compared with single sensors， the multi-faceted and multi-type information collected by multi-source sensors is more valuable for air target classification. In view of the current problems of air target recognition methods having single features， inability to cross-validate classification results and low recognition accuracy， an attention-based centralized air target recognition method with dynamic fusion of multi-source sensor features is proposed. This method uses deep learning models to extract the photoelectric image， motion trajectory， RCS and electromagnetic features of the target vehicle. Considering that the importance of each sensor feature will change dynamically in the real environment， the attention mechanism and the distance parameter are used to dynamically allocate the feature weights. The experimental results on the simulation dataset show that compared with the single-sensor model， the centralized method improves the recognition accuracy by 12.89% on average， which is a significant improvement in recognition effect； compared with the distributed model based on hierarchical analysis voting， the centralized method is more robust and better adapted to complex environments while fusing multi-source features more effectively.

In the unified S-band （USB） tracking telemetering and control （TT&C） system，the ground station extracts the range information through the phase delay of the ranging tone transmitted by the satellite TT&C transponder. Considering the influence of the ranging tone extraction filter and channel noise on the downlink modulation system， the modulation index of the downlink ranging tone will fluctuate unexpectedly. So the system efficiency and communication distance of the whole TT&C system are affected. This paper introduces a ranging sound power decoupling method for a miniaturized USB TT&C transponder， which simplifies the design of the uplink receiver while eliminating the adverse effect of the uplink received signal strength on the modulation system through the normalisation of the received power， so as to guarantee the power efficiency of the modulation signal of the measurement and control transponders. The feasibility and effect of the method are verified by product measurement.

To improve the low-elevation detection performance of ship-borne phased array radar， the low angle multi-path model， sea surface reflection model and elevation estimation model of phased array radar （PAR） are established. The relationship between low-elevation angle accuracy of ship-borne PAR and angle estimation method， radar frequency， beam direction， sea state and beam width is systematically analyzed through simulation. The results show that sum-sum angle estimation method， lower beam width by improved beam weight， higher beam direction of angle estimation sum beam， multi-frequency smoothing are benefit for low-elevation angle accuracy improvement. A low-elevation angle estimation method which takes advantage of all the factors above is proposed， which has achieved excellent real-world application results and significant engineering practical prospects.

With the rapid development of deep learning， the object detection network model hsa achieved great success in ship detection from synthetic aperture radar （SAR） images. In order to achieve better results， a large detection network is usually adopted， which requires more computing resources and slower inference. Knowledge distillation can effectively compress the network， but most of them are for image classification. Considering the difference between ships and background in SAR images， this paper proposes a combining local and global distillation method for ship detection. Considering the scattering characteristics of ships in SAR images， slice preprocessing in the amplitude direction is performed on the ship sample data constructing a separate data channel with more significant scattering characteristics， which improves the quality of input data fed to the network. Experimental results based on SAR ship detection dataset （SSDD） show that the proposed approach can effectively reduce the price of the network model and improve the detection performance， achieving 90.7% mAP.

In the testability verification test of electronic equipment， due to the high degree of circuit integration， the fault mode may be caused by one or more failure mechanisms， and traditional means such as thermal failure analysis can no longer effectively analyze the deep fault mode of electronic equipment， which leads to the reduction of the credibility of test samples. An improved fault mode analysis method based on the combination of failure physics and fault tree is proposed. Taking the overvoltage protection circuit in the power module of an electronic equipment as an example， the failure physics model is used to calculate the failure rate and criticality from the device level failure mode analysis， and the FMECA results of the overvoltage protection circuit are obtained. Guided by the fault tree analysis method， the FMECA results are verified by comparing the priority of the two according to the similarity between the definition of the probability importance of the bottom event and the criticality of the electronic components. According to this method， the fault mode analysis results of electronic equipment are finally obtained with high accuracy， which effectively improves the reliability of testability verification test samples.

In view of the requirements that flight vehicle ignition control software is not suitable for purchasing commercial components and needs customized development， the ignition control principles of flight vehicles are studied. The ignition control processes of various flight vehicles are abstracted， and an ignition control software component based on parameters and actions is designed. These subcomponents can cooperate with each other， and the component needs parameters which reflect the characteristics of control process. This component can adapt to the ignition requirements of different systems and realizes the core function of ignition control software， which can reduce the software development cost and improve software quality efficiently in the meantime.

To choose a right scheme for calibrating the spin correction coefficient of an accelerometer used in a high speed rotation environment， two calibration schemes are analyzed founded on the definition of the spin correction coefficient and the model of the accelerometer. At first， the principle of calibrating spin correction coefficient is summarized according to military standards. Then calibration models are established corresponding to the two calibration schemes， and compared with the expression of spin correction coefficient on the base of its definition. The analysis results show that if calibration is carried out based on the definition of the spin correction coefficient， the two schemes proposed in the national military standards are not equivalent. The scheme which needs the accelerometer rotates around its input axis cannot be employed to calibrate the spin correction coefficient while the other scheme， which uses the accelerometer rotates about the angular bisector of the input axis and pendulous axis can do the work.

Simulation research on triggering current is the kernel of electromagnetic railgun investigation， based on its equivalent electric circuit， the models of pulsed power， rail， armature， sliding electrical contact between armature and rail are established， and the electromagnetic railgun simulation model is established by integrating the component models together. By means of Dommel-EMTP（electromagnetic transient program）algorithm， the triggering current and voltage of every pulse forming unit are simulated， then the total discharging current， velocity and displacement of armature are solved. Compared with the tested date， the relative error of simulation results of total discharging current and armature muzzle velocity is less than 4% ， which indicates the effectiveness and accuracy of the model and provides basic method for further research on electromagnetic railgun.

Aiming at the problems of the Informed-RRT（rapidly-exploring random tree）* algorithm， such as slow convergence speed， low optimization efficiency and inability to execute the generated path， path planning research based on MI-RRT* （modified informed-RRT*） algorithm is carried out. The method of greedy sampling and adaptive step size is introduced to improve the convergence rate of the algorithm， reduce the path generation time and the memory usage. Minimum Snap curve is used to make the path smooth and the power change smoothly， so as to achieve the effect of saving energy and the path to generate the executable. Several sets of different experimental environments are run to show that the MI-RRT* algorithm is more advantageous than the original algorithm， the resulting planned path is smooth executable and it can reduce the number of iterations by 20% and the search time by 25%. Compared with the open and dense environment， the MI-RRT* algorithm has obvious advantages over the Informed-RRT* ，RRT* algorithms.