Aiming at the problem of spatio-temporal coordination constraint in formation penetration route planning under the condition of team support interference, a uni-machine evaluation function based on threat cost and energy consumption cost is constructed, and a comprehensive evaluation function with the timing coordination coefficient is designed to solve the problem of attack timing constraint.The multi-subgroup co-evolutionary ant colony algorithm introduces mutually-repulsive factors and designs dynamic target guidance factors to achieve collision avoidance within the space of the formation and the shortest voyage.The simulation results show that the improved multi-subgroup co-evolutionary ant colony algorithm can effectively solve the problem of spatio-temporal constraint in the formation-coordinated route planning under the interference of team support.

Aiming at infrared time-sensitive weak and small target detection in air-ground background, a practical detection method is proposed.Firstly, the fast adaptive bilateral filter is used to preprocess the infrared image, so as to smooth the edge information and eliminate the noise.Secondly, the high-pass filter with a template of 3×3 is adopted to suppress the large-area background of the infrared image.Then, according to the gradient characteristics, the multi-step eight-direction gradient method is applied to further eliminate the high-frequency background information.Finally, the adaptive threshold segmentation method is used to extract the infrared time-sensitive weak and small targets.The experimental results show that the proposed method has good robustness and performance for infrared time-sensitive weak and small target detection in air-ground background.

Aiming at the problem that the attitude calculation algorithm based on the Inertial Measurement Unit (IMU) is noisy, has low accuracy and cannot accurately calculate the attitude angle variation in the yaw direction, a MARG sensor system attitude calculation method based on the conjugate gradient algorithm is proposed.The accelerometer and magnetometer are used to measure the attitude error, the conjugate gradient algorithm is used to compensate for and correct the gyroscope attitude quaternion, and the data is weighted and fused to obtain the aircraft attitude quaternion.A three-axis turntable is used to verify the stability and accuracy of the algorithm.The results show that the algorithm can effectively suppress measurement noise and improve the accuracy of attitude angle calculation.

Aiming at the task allocation problem when multiple heterogeneous UAVs perform multiple tasks such as reconnaissance and attack, a Windows Consensus-Based Bundle Algorithm (WCBBA) is proposed.The algorithm solves the problem of multi-UAV task allocation based on the constraints of task time window and UAV loading limit.Firstly, the task allocation model of multiple heterogeneous UAVs is established based on the constraints, and the task revenue function and the discount function due to flight distance are designed.Secondly, the algorithm is used to solve the allocation model of reconnaissance and attack tasks of multiple UAVs.The simulation results show that the algorithm can successfully and efficiently resolve the conflict in the process of task allocation and realize the optimal task allocation.It has certain theoretical and practical value for multi-task decomposition in the cooperative operation of multiple heterogeneous UAVs.

Laser decoy jamming equipment lacks reasonable and feasible jamming effect evaluation method in actual combat training.According to the three main problems that laser decoy jamming effect evaluation method needs to solve, that is, corresponding jamming effect evaluation criteria, suitable object to be jammed and feasible jamming effect experiment method, the jamming effect evaluation criteria of deflection angle is proposed.The laser semi-active guided missile simulation system with the laser seeker simulator as the core is determined as the object to be jammed, and the ground simulation jamming effect experiment is improved.Finally, the feasibility of the improved ground simulation evaluation method based on the evaluation criteria of deflection angle is verified through experiments.

In order to solve the problems of collision, communication interruption and external disturbances during formation forming of Multi-Agent Systems (MASs), the robust formation control with collision avoidance and connection preservation for nonlinear second-order MASs is studied.Based on the Artificial Potential Field (APF) method, the collision avoidance algorithm is designed, which can maintain connections among agents, and distributed control law and consensus theory are employed to design the robust formation control law of the MASs, so that accurate formation forming with collision avoidance and connection preservation of the MASs is realized.Lyapunov stability theory and Linear Matrix Inequality (LMI) are used to guarantee the stability of the MASs and the convergence of tracking errors.The effectiveness of the proposed formation control law is demonstrated by numerical simulations.

The target detection system carried on UAV platform often faces many small target detection tasks in practical application.In order to overcome the problems of low detection rate and poor detection accuracy, an improved small target detection algorithm based on YOLOv3 is proposed.Firstly, the K-means clustering algorithm is used to conduct clustering analysis on the remote sensing small target data set from high-altitude perspective, and the number of anchor boxes and corresponding parameters are reset.Then, in the part of feature extraction network, the number of residual blocks after five times of down sampling is reconfigured, and an output scale is introduced in the shallower network, whose feature information is spliced with that in the previous output scale, so as to retain more small target information and edge information.Through the test and analysis of the test data set, the mAP of the improved detection algorithm reaches 92.21%, which is 5.84% higher than that of the original YOLOv3.It effectively solves the problem of YOLOv3 that miss detection is likely to occur on some small targets.

A neural network model with sparsity constraint and a complete moment feature set are proposed for improving the performance of infrared dim target detection.The traditional neural network models employ such activation functions as SoftMax or Logistic regression for classification.A simpler sign function is designed as the activation function of the classification layer, and forward regression is used to estimate the parameters.In order to reduce computational complexity and improve detection performance, norm constraints are added to the objective function, which can maintain the consistency and sparsity of the parameters.Experimental results show that the new method outperforms traditional approaches and realizes real-time implementation.

Remote sensing target detection algorithms based on deep learning have redundant parameters, large computation amount and poor real-time detection performance.To solve the above problems, a real-time remote sensing target detection algorithm based on depth separable convolution is proposed.The datasets are analyzed by anchor box (Anchor) clustering via the K-means++ algorithm, making the anchor box parameters more compliant with the remote sensing detection scenario.In order to reduce the quantity of model parameters and improve the detection speed, feature extraction is performed with the lightweight network MobileNetv3 as the backbone network.In addition, the design of PANet (Path Aggregation Network) structure based on depth separable convolution makes the quantity of network parameters further reduced.The quantity of model parameters after improvement is only 18.3% of the original, and the detection speed is 2.19 times faster than the original.Tests are conducted on three remote sensing datasets, that is, UCAS_AOD, RSOD and DIOR, and the experimental results show that the algorithm is robust and can effectively improve the real-time detection performance while ensuring the model detection accuracy.

The water column signal at the impact point on the sea is an important basis for evaluating the firing effect, so it is of great significance to obtain the situation of water column signal detection quickly and accurately in training and exercise.In view of the characteristics of multiple sizes and shapes of marine water column signals, YOLOv5 algorithm is improved and CAs-YOLOv5s algorithm is proposed.The mixup data augmentation strategy is added to the input to build new training samples and labels by using linear interpolation, which can enrich the sample information without occupying too much storage space.The Coordinate Attention (CA) mechanism is introduced to embed location information into the channels attention, so as to enhance the feature extraction capability of the model.At the same time, the pooling method of Maxpool in Spatial Pyramid Pooling (SPP) module of the original YOLOv5s algorithm is replaced with Softpool to retain more fine-grained feature information and magnify more intense feature activation.Experimental results on the target dataset show that the mAP of CAs-YOLOv5s algorithm is increased by 4.54% to 94.75%, and the speed is up to 23.51 frames per second, which can better complete the task of water column signal detection at the impact point on the sea under the real-time requirements.

With the development of electronic countermeasure technology, the ability of electronic reconnaissance system to detect aircraft has been significantly improved.The existence of aircraft in modern air combat is seriously threatened and its penetration ability is greatly weakened.Radio Frequency (RF) stealth radar is a hot spot of research on aviation equipment at present.It can fundamentally improve the electronic defense capability of radar, and the research focuses on the waveform design of RF stealth radar.Firstly, the development of RF stealth radar is summarized, the advantages and disadvantages of four RF stealth technologies are compared, the main characteristics and technical indexes of electronic reconnaissance system are introduced, and the necessity and urgency of research on RF stealth radar are expounded.Then, three RF stealth waveform design methods based on composite signal, wide random variation of signal parameters and optimization algorithm are analyzed, and the difficulties and scientific exploration in waveform design are summarized.Finally, the development direction of RF stealth waveform design technology is predicted.

The transmitted beam of the phased array only has angle dependence.In contrast, Frequency Diverse Array (FDA) adds an increment of frequency offset between the array element to make it display the characteristics of range-angle two-dimensional dependence, which not only solves the problem of angle pointing during beam transmission, but also solves the problem of distance pointing during beam transmission.By introducing MIMO into FDA, the transmitted information can be integrated into the receiver, thus realizing more flexible utilization of range dimension Degrees of Freedom (DOF) and improving the ability of range-angle two-dimensional joint signal processing, which gives FDA-MIMO unique application advantages in countering range-dimension main lobe deception jamming.In view of the good anti-jamming characteristics of FDA-MIMO, the basic principle and model of FDA-MIMO are introduced, the research progress of FDA-MIMO at home and abroad is summarized from the perspective of anti-jamming, the basic principle and main techniques of FDA-MIMO radar anti-jamming are systematically discussed and summarized, and the main application prospects of FDA-MIMO and other FDA radar technologies in anti-jamming in the future are analyzed.

The search for targets in a large range of sector of airborne fire control radar is usually conducted in four-line scanning mode.In order to discover the targets earlier than the traditional progressive scanning can do, a four-line flyback scanning mode is proposed.Suppose that the target may appear in every position of the searched sector, and the coordinates conform to a two-dimensional uniform distribution.Based on the assumption of discovery probability model in a single scanning, the expected discovery time of the two scanning modes is calculated by using the Monte Carlo method.The simulation results show that the expected discovery time of the flyback scanning mode is less than that of the progressive scanning mode in one scanning period of radar.That is to say, the flyback scanning mode can realize target discovery faster than the traditional progressive scanning mode in statistical sense.

Prognostics and Health Management (PHM) is an essential operation in improving the reliability of UAVs.Aiming at the contradiction between data quantity and transmission efficiency in airborne PHM communication of BeiDou short message, a kind of BDS priority-based packet transmission strategy for airborne PHM communication is developed.Based on the idea of transmission information priority, a packet transmission strategy is designed to balance the contradiction between communication frequency and insufficient capacity.Then, the reliability of communication is improved by using the automatic repeat-request mechanism, and the visual communication application is realized.The experimental results show that the proposed strategy can reliably transmit PHM information within the specified time, which verifies the reliability of the developed BDS priority-based packet transmission strategy in terms of airborne PHM information transmission.

Infrared searching and tracking (IRST) system has an important position in many fields such as situational awareness and environmental safety monitoring.Image motion compensation is one of the key technologies for fast scanning of IRST.A large area array IRST optical system with image motion compensation under the conditions of long integration time and fast circle scanning is designed, and image stabilization method of parallel optical path is adopted in the optical system.The optical system has the technical characteristics of 4 ms integration time, 360 (°)/s circle scanning rate, and 1k area array scale.The principle of parallel optical path image stabilization is introduced, and the calculation and matching of the parameters are proposed.The results of lens design and circle-scanning imaging tests on optical prototype indicate that the optical system has excellent imaging detection capabilities, and the relevant indexes of the optical system have reached domestic leading level.

To solve the problems of low accuracy and poor robustness of traditional video anomaly event detection algorithms, a video anomaly event detection algorithm based on two-stream residual network is proposed.The algorithm uses a combination of deep residual networks, temporal segmentation networks and convolutional fusion strategies.Based on the traditional two-stream network, the algorithm extracts motion information and temporal behavior from single-frame images and multi-frame optical flow images respectively.The network’s depth is further deepened to extend the motion information modeling capability.The temporal features are fully extracted by using the segmented network construction to enhance the effect of processing long-time videos.High-dimensional spatio-temporal features are fused in the middle layer of the network by convolutional fusion method to fully explore the spatio-temporal correlations in videos and obtain final detection results.Experimental results of training and validation on publicly available UCF-Crime and XD-Violence datasets show that the proposed video anomaly event detection algorithm based on two-stream residual network has approximately 10% improvement in accuracy over methods that only use uni-modal network(spatial stream network).The accuracy is improved by 3.2% and 6.1% respectively in comparison with traditional two-stream networks.

Firstly, the coherent efficiency with mismatching angle is deduced theoretically.Secondly, the relationship between the optimal relative aperture and coherent efficiency and the relationship between coherent efficiency and mismatching angle are simulated.Finally, the influence on coherent efficiency and receiving field when local oscillator is out of focus is evaluated.The results show that the receiving field of coherent detection is inversely proportional to antenna aperture in ideal conditions, and the overall performance of receiving antenna when local oscillator is out of focus is no better than that of small-aperture antenna when the effective coherent efficiency is identical.The results can serve as a theoretical basis of matching design between coherent detection system and optoelectronics system, and provide guidance on the antenna design, adjustment and measurement of coherent detection.

In order to improve the transient performance of turbofan engine and provide limit protection, min-max fuel selection strategy is often used.The complex part of this strategy is the tuning process of controller gain.A controller gain tuning method based on Particle Swarm Optimization (PSO) is proposed.The tuning process of the gain parameters of the min-max fuel selection controller is transformed into a numerical optimization problem.The main objects of optimization are the gain parameters of the transient state controller and the limit protection controller.In this numerical optimization problem, the objective function is designed as the sum of weighted transient response time and weighted fuel consumption.The simulation results show that the proposed method not only improves the response speed of transient state, but also provides limit protection, which prove the effectiveness and superiority of the method.Besides, the effect of the weighting coefficient in the objective function on the engine response time and fuel consumption is analyzed, and the balance diagram between the three factors is drawn,which can serve as a reference when selecting the weighting coefficient in the design of engine.

In order to meet the needs of high-speed parallel processing of UWB signals in future electronic information systems, a coherent Optical Frequency Comb (OFC) generation method is proposed, which can be applied to UWB channel-processing.Based on nonlinear broadening effect, coherent OFC generation with high repetition rate and multiple combs can be realized, and the repetition rate and repetition rate difference can be tuned.The cascaded electro-optical modulator is used to generate the initial coherent optical comb.The seed comb selected by Wavelength Division Multiplexing (WDM) is injected and locked.The locked comb is replicated and broadened by the combination module of nonlinear, single-mode and dispersion flattened optical fiber.The coherent OFC with the repetition rate of 80 GHz and 79.4 GHz, the flatness of less than 3 dB and 20 combs is obtained, which can receive and process the UWB signals in Q/V band, and support the reception and processing of the RF signals with wide-band coverage and large instantaneous bandwidth in such applications as RF integration and large-capacity satellite communication.

In order to solve the problems of poor adaptation to infrared image noise,poor accuracy of segmentation results and complicated parameter determination in image segmentation by Pulse Coupled Neural Network (PCNN), an adaptive segmentation algorithm (AGK-PCNN) in combination with multi-scale Gaussian kernel and Particle Swarm Optimization (PSO) is proposed.Firstly, based on the simplified PCNN, global coupling and pulse synchronization of input images are conducted.The weight matrix is designed by using the characteristics of anisotropic Gaussian kernel, so as to effectively suppress infrared noise.The model key parameters of different images are adaptively adjusted to achieve the optimal segmentation effect by using PSO algorithm.Finally, as for visual effect, a comparison with the maximum inter-class variance method,the adaptive Gaussian thresholding segmentation method and SCM segmentation method is conducted, and a quantitative comparison of the segmented images is conducted by using IoU and Dice score, etc.The results show that the segmentation effect is better than that of other methods in terms of both subjective visual effect and objective indexes.

To solve the problem of low detection and recognition rate of small-sized pedestrian objects in images, a pedestrian detection algorithm based on the improved multi-scale feature fusion is proposed.Firstly, the BN layer is fused into the convolution layer based on the original YOLOv3 model.Secondly, a detection layer is added, and the features of the high and low layers are combined and predicted by referring to the idea of feature pyramid.Finally, K-means clustering algorithm based on linear scaling is used to optimize the anchor box, so as to improve the detection effect of small-sized pedestrian.The experimental results on INRIA pedestrian dataset show that the accuracy of the improved algorithm reaches 91.4%, which is 3.4% higher than that of the YOLOv3 algorithm.The effectiveness of the proposed algorithm is also proved in complex monitoring environment.