Target assignment and path planning are crucial for the cooperativity of multiple unmanned aerial vehicles (UAV) systems. However, it is a challenge considering dynamics environments partial observability UAVs. In this article, problem multi-UAV target formulated as partially observable Markov decision process (POMDP), novel deep reinforcement learning (DRL)-based algorithm proposed to address it. Specifically, network introduced into twin-delayed deterministic policy gradient (TD3) solve...

The objective of this research is to enhance pneumonia detection in chest X-rays by leveraging a novel hybrid deep learning model that combines Convolutional Neural Networks (CNNs) with modified Swin Transformer blocks. This study aims significantly improve diagnostic accuracy, reduce misclassifications, and provide robust, deployable solution for underdeveloped regions where access conventional diagnostics treatment limited. developed architecture integrating CNNs blocks work seamlessly...

The remaining useful life (RUL) prediction of Lithium-ion batteries is closely related to the capacity degeneration trajectories. Due self-charging and regeneration, trajectories have property multimodality. Traditional models such as support vector machines (SVM) or Gaussian Process regression (GPR) cannot accurately characterize this This paper proposes a novel RUL method based on Mixture (GPM). It can process multimodality by fitting different segments with GPR separately, that tiny...

An important step in the construction of a support vector machine (SVM) is to select optimal hyperparameters. This paper proposes novel method for tuning hyperparameters by maximizing distance between two classes (DBTC) feature space. With normalized kernel function, we find that DBTC can be used as class separability criterion since between-class separation and within-class data distribution are implicitly taken into account. Employing an objective develop gradient-based algorithm search...

Cloud segmentation is one of the hot tasks in field weather forecast, environmental monitoring, site selection for observatory, and other areas. In this letter, we proposed a new deep convolutional neural network architecture called CloudU-Net daytime nighttime cloud images' segmentation. The net consists dilated convolution, activation, batch normalization (BN), max pooling, upsampling, skip connection, fully connected conditional random (CRF) layers. benefits are four aspects: First,...

A new voice activity detector (VAD) algorithm using support vector machines (SVM) is proposed in the paper, and VAD effectiveness validated. The sequential minimal optimization (SMO) for fast training adopted. via SVM (SVM-VAD) also uses characteristic parameters set used by G.729 Annex B (G.729B) VAD. Comparing SVM-VAD with G729B shows that it effective applying to integrated G.729B instead of VAD, informal listening tests show speech coding system has a little better efficiency over perceptivity.

Abstract. Cloud segmentation plays a very important role in astronomical observatory site selection. At present, few researchers segment cloud nocturnal all-sky imager (ASI) images. This paper proposes new automatic algorithm that utilizes the advantages of deep-learning fully convolutional networks (FCNs) to pixels from diurnal and ASI images; it is called enhancement network (EFCN). Firstly, all images data set Key Laboratory Optical Astronomy at National Astronomical Observatories Chinese...

Obtaining accurate cloudage information through ground-based cloud observation is of great significance to astronomical telescope observatory site selection. This paper proposes a residual attention-based encoder–decoder network (CloudRAEDNet) for image segmentation in nychthemeron. CloudRAEDNet uses ImageNet pre-trained ResNet50 as the encoder backbone network, which reduces number training. The decoder introduces modules solve problem degradation caused by increase layers. connects and...

On the basis of short-time energy speech signals and efficient method noise statistics adaptation estimation proposed by Sohn et al.(1998), a new highly robust voice activity detection (VAD) rule for any kind environmental is in this paper. The accurate recognition rate about five percent higher than that Sohn's on average, also has same merit tracking spectrum properly as method. Simulation experiments show an detector.

To address the limitation and obtain position of drone even when relative poses intrinsics camera are unknown, a visual positioning algorithm based on image retrieval called AGCosPlace, which leverages Transformer architecture to achieve improved performance, is proposed. Our approach involves subjecting feature map backbone an encoding operation that incorporates attention mechanisms, multi-layer perceptron coding, graph network module. This allows for better aggregation context information...

In wireless sensor networks (WSNs), data recovery is an indispensable operation for loss or energy constrained WSNs using sparse sampling. However, the accuracy not satisfying with various types due to neglect of correlation among multi-attribute data. this paper, we propose a novel method joint sparsity and low-rank constraints based on tensor completion in WSNs. The proposed represents high-dimensional as tensors effectively exploit that exists utilization spatiotemporal signal emphasized...

In recent years, Rwanda’s rapid economic development has created the “Rwanda Africa Wonder”, but it also led to a substantial increase in energy consumption with ambitious goal of reaching universal access by 2024. Meanwhile, on basis and dynamic connection new households, there is uncertainty about generating, importing, exporting whichever imposes significant barrier. Long-Term Load Forecasting (LTLF) will be key country’s utility plan examine electrical load demand growth patterns...

The accurate, rapid, and stable prediction of electrical energy consumption is essential for decision-making, management, efficient planning, reliable power system operation. Errors in forecasting can lead to electricity shortages, wasted resources, supply interruptions, even grid failures. Accurate enables timely decisions secure management. However, predicting future challenging due the variable behavior customers, requiring flexible models that capture random complex patterns. Forecasting...