In terms of computer vision, image classification plays a crucial role. With the progress in deep learning technology, there has been a significant improvement in image classification performance. Due to their capability of self-learning, deep learning models are applicable to automatically extract features from a vast amount of data and effectively address the complexity in image classification by navigating the nonlinear relationship inherent in high-dimensional data. In this paper, a novel approach to image classification is presented on the basis of an enhanced VGG network, with the GELU activation function incorporated into the VGG network to address gradient vanishing for deep learning. Furthermore, the Efficient Multi-scale Attention Module (EMA) is introduced for the network to focus more on the retention of information in each channel. This is achieved by dividing the channels into subgroups and balancing the spatial semantic features within the feature groups. Through cross-dimensional interaction, this network can be applied to encode the global information and enhance the features. Finally, an experiment is conducted on the CIFAR-100 classification dataset to demonstrate the superiority of the proposed method in image classification.
In the current most popular field of deep learning, activation function refers to a mathematical function applied to perform non-linear transformation on the input. At present, it is commonly used in the hidden and output layers of feature extraction networks. When activation functions are introduced, the non-linear capabilities of neural networks are enhanced, Meanwhile, the model performs better in terms of expressive power and fitting ability. Inspired by trigonometric functions, an activation function is proposed in this study. Based on the inverse tangent trigonometric function -, namely ATLU, which is the arctangent-based nonlinear activation unit, it is expressed as f(x) = x(arctan(1.09x) + π/2). This activation function can be applied to address the gradient vanishing and exploding in deep neural networks. It is validated on the cifar100 dataset with the VGG11 network as the benchmark.
Drop point measurement precision is one of the core indexes to evaluate the combat effectiveness of weapons. With the development of experimental equipment, the experimental training venue has been expanded to the far sea. Due to the little known data in the far sea area and the measuring area, the measurement methods are limited. In response to the characteristics of the high seas, this paper proposes a method of mounts an optoelectronic pod on a drone and utilizes two drones for collaborative intersection measurement, achieving high-precision landing point measurement and high reliable data acquisition rate. This paper provides a detailed comparison between the traditional H-E-A single station angle measurement and distance measurement methods, the collinear equation based non ranging information positioning method, and the dual aircraft intersection positioning measurement principle combined with RLS filtering algorithm. At the same time, this paper analyzed various factors that affect the accuracy of positioning measurement. Through actual measurement verification of simulated targets, this method achieved a drop point measurement accuracy of 2m within a range of 3Km and a measurement accuracy of over 95%, which is significantly improved compared to traditional methods. The method provides data support for evaluating weapon effectiveness and obtaining field situation, and can also serve as auxiliary means for personnel search and rescue, debris search, etc., greatly improving the fusion ability of multidimensional data and enhancing the independent innovation and support ability of far sea measurement equipment.
The control accuracy of Single Gimbal Control Moment Gyroscope (SGCMG) is limited by the existence of interference factors such as dynamic imbalances. This paper analyzes the output torque disturbance of SGCMG from a dynamic perspective and verifies the influence weights of disturbance factors through simulations, providing a reference for the control of torque output stability in SGCMG. Firstly, considering factors such as static and dynamic imbalance, installation errors of the high-speed rotor system using the Newton-Euler method, a preliminary analysis of each subsystem is conducted. Secondly, considering disturbance factors such as frame speed, a dynamic model analysis is performed. Finally, based on actual parameter values, the characteristics of the system's output torque are analyzed through simulations. Experimental results show that the torque output process of the system is stable under ideal conditions. However, in the presence of disturbances such as dynamic imbalance, the stability of SGCMG's torque output process is influenced to varying degrees. The analysis of the influence weights of various disturbance factors is accomplished using principal component analysis.
KEYWORDS: Computer simulations, Detection and tracking algorithms, Control systems, Telecommunications, Error analysis, Systems modeling, Control systems design, Precision optics, Precision mechanics, Sensors
Consensus problem is a hot area of multi-agent cooperative control, and has produced many research results.Design consensus algorithm is the focus of multi-agent problem research.However, for high precision situations, multi-agent cooperative control needs more effective consensus algorithm.Since most of the dynamic models in reality are second-order systems, this paper provides a consensus algorithm of second-order multi-agent system with integral, and compares it with the traditional consensus algorithms. The algorithm has higher response rate and consensus accuracy.In order to illustrate the effectiveness of the proposed algorithm, a set of simulation results is provided.
In the application of space satellite turntable, the design of balance wheel is very necessary. To solve the acquisition precision of Brushless DC motor speed is low, and the encoder is also more complex, this paper improves the original hall signal measurement methods. Using the logic device to achieve the six frequency multiplication of hall signal, the signal is used as speed feedback to achieve speed closed-loop control and improve the speed stability. At the same time, in order to prevent the E.M.F of BLDC motor to raise the voltage of the bus bar when reversing or braking, and affect the normal operation of other circuit modules, the analog circuit is used to protect the bus bar voltage by the way of energy consumption braking. The experimental results are consistent with the theoretical design, and the rationality and feasibility of the frequency multiplication scheme and bus voltage protection scheme are verified.
The optical frequencies band is used as information carrier to realize laser communication between two low-orbit micro-satellites in space which equipped with inter-satellite laser communication terminals, optical switches, space routers and other payload. The laser communication terminal adopts a two-dimensional turntable with a single mirror structure. In this paper, the perturbation model of satellite platform is established in this paper. The relationship between the coupling and coordinate transformation of satellite disturbance is analyzed and the laser pointing vector is deduced. Using the tracking differentiator to speed up the circular grating angle information constitute speed loop feedback, which avoids the problem of error amplification caused by the high frequency of the conventional difference algorithm. Finally, the suppression ability of the satellite platform disturbance and the tracking accuracy of the tracking system are simulated and analyzed. The results show that the tracking accuracy of the whole system is 10μrad in the case of satellite vibration, which provides the basis for the optimization of the performance of the space-borne laser communication control system.
When the star navigation system working during the day, the strong sky background radiation lead to a result that the detect target light is too weak, in the field of view, because of the limitation on the number of the navigation star, usually choose the single star navigation work mode. In order to improve the reliability of the airborne SINS/CNS integrated navigation system, meet the demand of the long-endurance and high precision navigation, use the tight combination way, single star patrol algorithm to get the position and attitude. There exists filtering divergence problem because of the model error and the system measurement noise is uncertain, put forward a new fuzzy adaptive kalman filtering algorithm. Adjust the size of measurement noise to prevent the filter divergence; the positioning accuracy of integrated navigation system can be improved through BeiDou satellite. Without the information of BeiDou satellite, based on the level of the virtual reference, the navigation precision of integrated navigation system can be ensured over a period of time.
High precision tracking platform of celestial navigation with control mirror servo structure form, to solve the disadvantages of big volume and rotational inertia, slow response speed, and so on. It improved the stability and tracking accuracy of platform. Due to optical sensor and mirror are installed on the middle-gimbal, stiffness and resonant frequency requirement for high. Based on the application of finite element modality analysis theory, doing Research on dynamic characteristics of the middle-gimbal, and ANSYS was used for the finite element dynamic emulator analysis. According to the result of the computer to find out the weak links of the structure, and Put forward improvement suggestions and reanalysis. The lowest resonant frequency of optimization middle-gimbal avoid the bandwidth of the platform servo mechanism, and much higher than the disturbance frequency of carrier aircraft, and reduces mechanical resonance of the framework. Reaching provides a theoretical basis for the whole machine structure optimization design of high-precision of autonomous Celestial navigation tracking mirror system.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.