Search

In this paper, the fast integral terminal sliding mode control for spacecraft formation flying system is concerned subjected to external disturbances and parametric uncertainties. First, a novel modified fast integral terminal sliding mode control law is designed, which contains the advantages of fast terminal sliding mode. Then, for estimating the external disturbances in the system, the control law based on the reduced-order disturbance observer is proposed which not only reduce the order of the observer but also guarantee the tracking errors converge in finite time. Further, to identify the unknown parameters accurately, by introducing an adaptive parametric updating law, the adaptive fast integral terminal sliding mode control law is proposed, which has better robustness against external disturbances and parameter uncertainties simultaneously. Finally, the effectiveness of the proposed control laws is demonstrated by the simulation results.

As a popular distributed machine learning based on stochastic gradient decent (SGD), federated learning enables edge devices to compute stochastic gradients and then upload them to an edge server for global AI-model updating. However, there are two problems to be faced in federated learning. First, the transmission of high -dimensional stochastic gradients from edge devices to edge server causes serious communication bottleneck. Second, the fading and noise of wireless channels may distort the transmitted stochastic gradients. To address these issues, we adopt scaler quantization for SGD by taking dynamics of wireless channels into account and analyze the influence of wireless channels and quantization on the convergence rate. Then, the quantization error and channel effects are characterized in received signal-to-noise ratio (SNR). We formulate and then solve an optimization problem of sum-SNR maximization by quantization resources allocation, for the goal of making the quantization and transmission as reliable as possible. Simulations show that our scheme can outperform the benchmark schemes.

Decision-making is a fundamental task in both brain functions and AI applications. In this study, we develop a brain-inspired decision-making network for spatio-temporal pattern recognition. We analyze the mechanism and properties of the proposed model using synthetic data and demonstrate the potential application of the model in speech recognition.

In this paper, the vibration signal reproduction of a flexible beam actuated by multiple exciters with parameter uncertainties is investigated. By combining Barrier function with PD control, the parameters of the PD controller can be updated in real-time and the output constraint of the flexible beam system can be achieved. Consider the possible saturation of the controller, an improved Barrier based PD controller is proposed to avoid the singularity while updating the gain of the PD controller. Through the simulation analysis of an example that the flexible beam actuated by three exciters, the effectiveness of the proposed Barrier based PD controller is demonstrated.

Multi-stage weapon target assignment (MSWTA) problem is usually treated as a classical constrained multi-objective optimization problem, which tends to balance the attack efficiency and economic loss for weapons after adding engagement resource constraints. In the context of multi-stage confrontations, the MSWTA becomes more complex and challenging to solve. To handle this constrained problem, this paper develops the advantageous two-archive evolutionary algorithm (C-TAEA) to combine the density estimation mechanism with a fast non-dominated sorting method to obtain solutions to this problem efficiently, while maintaining the convergence-oriented archive (CA) to speed up the generation of solutions and the diversity-oriented archive (DA) to enhance the diversity of populations. Extensive experiments are conducted to compare the proposed C-TAEA with the traditional NSGA-II according to the average of IGD and HV values. The numerical simulation results demonstrate that our method intuitively shows the performance improvement after adding the feasibility and can find the approximate Pareto Front more effectively.

In this paper, the two order quasi-newton iterative learning control algorithm for the multi-actuators vibration test system is investigated. First, a vibration coupling system with multi-actuators is established based on the frequency response function of the test beam. Second, a proportional-resonant filter is employed to filtrate the measurement disturbance in the system. Third, a novel two order iterative learning control algorithm based on he quasi-newton method is proposed to remove the actuate force coupling. The proposed control algorithm not only resolve the coupling phenomena effectively, but also guarantee the tracking errors of the system convergence with parameter uncertainty. Finally, simulations with eight actuators mathematical model with uncertainty are used to demonstrate the effectiveness of the proposed control algorithm.

The 3-th Generation Partnership Project (3GPP) Long Term Evolution-Advanced (LTE-A) has recently been investigating heterogeneous network (HetNet) deployments to improve the system throughput However, it is a great challenge to coordinate the transmission of each node in the HetNet scenario to increase the spectral efficiency. In this paper, we investigate the system-level performance of coordinated multi-point transmission for HetNets in LTE-A systems. We focus on the coordinated multi-point transmission between the Macro base station (Macro BS) and the low-power node, the low-power node can either shares the same cell ID with the Macro BS or have a separate cell ID. We evaluate the cell average and cell edge user's performance of those two HetNet scenarios on our system-level simulation platform. Moreover, we also apply a power consumption model to evaluate the energy efficiency.

We present CARAS, a web server that allows the automatic annotation of a chloroplast genome sequence, and the visualization and editing of the annotation results interactively and in real-time. CARAS accepts a complete chloroplast genome sequence as input. First, it accurately predicts protein-coding sequences and exon-intron structures by combining the results from two types of annotation approaches: ab initio prediction algorithms and similarity based methods. Second, tRNA genes and inverted repeats are identified using tRNAscan and vmatch. Using 220 chloroplast genome sequences as test, we show that CARAS outperforms a similar application DOGMA overall. Third, the annotation results are presented as a circular map that includes the name, location, orientation, and length of the various features. A Flex-based module is implemented, allowing the users to add, delete, and edit the features on-line. The results are stored on the server and can be retrieved from a given URL. Users also have options to download the annotation results in GFF3 format for further analyses in other third party software tools or in JPEG format for publication. Finally, CARAS can be used to create a Sequin file for GenBank submission of the annotated genome sequence. CARAS is freely available at http://caras.bicoup.com, and it will significantly facilitate research work involving chloroplast genomes.