Search

When multi-dynamic target UAVs escape, the uncertainty of the formation method and the external environment causes difficulties in rounding them up, so suitable solutions are needed to improve the roundup success rate. However, traditional methods can generally only enable the encirclement of a single target, and when the target is scattered and escaping, this will lead to encirclement failure due to the inability to sufficiently allocate UAVs for encirclement. Therefore, in this paper, a real-time roundup and dynamic allocation algorithm for multiple dynamic targets is proposed. A real-time dynamic obstacle avoidance model is established for the roundup problem, drawing on the artificial potential field function. For the escape problem of the rounding process, an optimal rounding allocation strategy is established by drawing on the linear matching method. The algorithm in this paper simulates the UAV in different obstacle environments to round up dynamic targets with different escape methods. The results show that the algorithm is able to achieve the rounding up of multiple dynamic targets in a UAV and obstacle scenario with random initial positions, and the task UAV, which is able to avoid obstacles, can be used in other algorithms for real-time rounding up and dynamic allocation. The results show that the algorithm is able to achieve the rounding up of multi-dynamic targets in scenarios with a random number of UAVs and obstacles with random locations. It results in a 50% increase in the rounding efficiency and a 10-fold improvement in the formation success rate. And the mission UAV is able to avoid obstacles, which can be used in other algorithms for real-time roundup and dynamic allocation.

The convolutional neural network (CNN) has been widely applied in medical image segmentation due to its outstanding nonlinear expression ability. However, applications of CNN are often limited by the receptive field, preventing it from modeling global dependencies. The recently proposed transformer architecture, which uses a self-attention mechanism to model global context relationships, has achieved promising results. Swin-Unet is a Unet-like simple transformer semantic segmentation network that combines the dominant feature of both the transformer and Unet. Even so, Swin-Unet has some limitations, such as only learning single-scale contextual features, and it lacks inductive bias and effective multi-scale feature selection for processing local information. To solve these problems, the Residual Local induction bias-Unet (RL-Unet) algorithm is proposed in this paper. First, the algorithm introduces a local induction bias module into the RLSwin-Transformer module and changes the multi-layer perceptron (MLP) into a residual multi-layer perceptron (Res-MLP) module to model local and remote dependencies more effectively and reduce feature loss. Second, a new densely connected double up-sampling module is designed, which can further integrate multi-scale features and improve the segmentation accuracy of the target region. Third, a novel loss function is proposed that can significantly enhance the performance of multiple scales segmentation and the segmentation results for small targets. Finally, experiments were conducted using four datasets: Synapse, BraTS2021, ACDC, and BUSI. The results show that the performance of RL-Unet is better than that of Unet, Swin-Unet, R2U-Net, Attention-Unet, and other algorithms. Compared with them, RL-Unet produces significantly a lower Hausdorff Distance at 95% threshold (HD95) and comparable Dice Similarity Coefficient (DSC) results. Additionally, it exhibits higher accuracy in segmenting small targets.

A highly intelligent system often draws lessons from the unique abilities of humans. Current humanlike models, however, mainly focus on biological behavior, and the brain functions of humans are often overlooked. By drawing inspiration from brain science, this article shows how aspects of brain processing such as sensing, preprocessing, cognition, obstacle learning, behavior, strategy learning, pre-action, and action can be melded together in a coherent manner with cognitive control architecture. This work is based on the notion that the anti-collision response is activated in sequence, which starts from obstacle sensing to action. In the process of collision avoidance, cognition and learning modules continuously control the UAV’s repertoire. Furthermore, simulated and experimental results show that the proposed architecture is effective and feasible.

The purpose of knowledge representation learning is to learn the vector representation of research objects projected by a matrix in low-dimensional vector space and explore the relationship between embedded objects in low-dimensional space. However, most methods only consider the triple structure in the knowledge graph and ignore the additional information related to the triple, especially the text description information. In this paper, we propose a knowledge graph representation model with a symmetric architecture called Joint Knowledge Representation Learning of Text Description and Knowledge Graph (JKRL), which models the entity description and relationship description of the triple structure for joint representation learning of knowledge and balances the contribution of the triple structure and text description in the process of vector learning. First, we adopt the TransE model to learn the structural vector representations of entities and relations, and then use a CNN model to encode the entity description to obtain the text representation of the entity. To semantically encode the relation descriptions, we designed an Attention-Bi-LSTM text encoder, which introduces an attention mechanism into the Bi-LSTM model to calculate the semantic relevance between each word in the sentence and different relations. In addition, we also introduce position features into word features in order to better encode word order information. Finally, we define a joint evaluation function to learn the joint representation of structural and textual representations. The experiments show that compared with the baseline methods, our model achieves the best performance on both Mean Rank and Hits@10 metrics. The accuracy of the triple classification task on the FB15K dataset reached 93.2%.

The accuracy of dam deformation prediction is a key issue that needs to be addressed due to the many factors that influence dam deformation. In this paper, a dam deformation prediction model based on IEALL (IGWO-EEMD-ARIMA-LSTM-LSTM) is proposed for a single-point scenario. The IEALL model is based on the idea of a combination model. Firstly, EEMD is used to decompose the dam deformation data, and then the ARIMA and LSTM models are selected for prediction. To address the problem of low prediction accuracy caused by simple linear addition of prediction results from different models in traditional combination models, the LSTM model is used to learn the combination relationship of different model prediction results. The problem of neural network parameters falling into local optima due to random initialization is addressed by using the improved gray wolf optimization (IGWO) to optimize multiple parameters in the IEALL combination model to obtain the optimal parameters. For the multi-point scenario of dam deformation, based on the IEALL model, a dam deformation prediction model based on spatio-temporal correlation and IEALL (STAGCN-IEALL) is proposed. This model introduces graph convolutional neural networks (GCN) to extract spatial features from multi-point sequences, increasing the model’s ability to express spatial dimensions. To address the dynamic correlation between different points in the deformation sequence at any time and the dynamic dependence on different points at any given time, spatio-temporal attention mechanisms are introduced to capture dynamic correlation from both spatial and temporal dimensions. Experimental results showed that compared to ST-GCN, IEALL reduced the RMSE, MAE, and MAPE by 16.06%, 14.72%, and 21.19%. Therefore, the proposed model effectively reduces the prediction error and can more accurately predict the trend of dam deformation changes.

In this study, the dynamic mechanical properties of steel fiber reinforced concrete under the influence of freeze–thaw cycles were studied. The studied parameters include steel fiber content (0%, 1% and 2%), confining pressures (0, 5 and 10 MPa) and strain rates (10−5/s, 10−4/s, 10−3/s and 10−2/s). Performance was also evaluated, including triaxial compressive strength, peak strain, the relationship between stress and strain, failure mode and microstructure. The results show that with the increase in F–T cycles, the compressive strength and energy absorption capacity of concrete gradually decrease. The mechanical properties of concrete increased with the addition of steel fibers during F–T cycles, and the optimum amount of steel fiber to enhance resistance to F–T cycles is 1% within the evaluation range. In this study, the effects of strain rate and confining pressure on the strength and failure mode of concrete after fiber addition are studied. Both the dynamic increase factor and the concrete strength increase linearly with the increase of strain rate, the dynamic increase factor is characterized by an increase in intensity caused by strain rate. When there is no confining, the crack direction of the concrete specimen is parallel to the stress loading direction, and when there is confining, it is manifested as oblique shear failure. The results of scanning electron microscopy analysis of the microstructure demonstrate the performance results at the macroscopic level (compressive strength and peak strain).

With photoplethysmograph (PPG) sensors showing increasing potential in wearable health monitoring, the challenging problem of motion artifact (MA) removal during intensive exercise has become a popular research topic. In this study, a novel method that combines heart rate frequency (HRF) estimation and notch filtering is proposed. The proposed method applies a cascaded adaptive noise cancellation (ANC) based on the least mean squares (LMS)-Newton algorithm for preliminary motion artifacts reduction, and further adopts special heart rate frequency tracking and correction schemes for accurate HRF estimation. Finally, notch filters are employed to restore the PPG signal with estimated HRF based on its quasi-periodicity. On an open source data set that features intensive running exercise, the proposed method achieves a competitive mean average absolute error (AAE) result of 0.92 bpm for HR estimation. The practical experiments are carried out with the PPG evaluation platform developed by ourselves. Under three different intensive motion patterns, a 0.89 bpm average AAE result is achieved with the average correlation coefficient between recovered PPG signal and reference PPG signal reaching 0.86. The experimental results demonstrate the effectiveness of the proposed method for accurate HR estimation and robust MA removal in PPG during intensive exercise.