Abstract ：

Group activity recognition can remarkably improve the understanding of video content by analyzing human behaviors and activities in videos. We propose a random walk graph convolutional network (RWGCN) for group activity recognition. (1) Considering the limitation of the convolutional structure to the visual information of group activities, the position feature extraction module is used to compensate for the loss of visual information. (2) A graph convolutional network (GCN) with distance-adaptive edge relations is constructed using individuals as graph nodes to identify the intrinsic relationships among the individuals in the group activities. (3) A Levy flight random walk mechanism is introduced into the GCN to obtain information from different nodes and integrate the previous position information to recognize group activity. Extensive experiments on the publicly available CAD, CAE datasets, and self-built BJUT-GAD dataset show that our RWGCN achieves MPCA of 95.49%, 94.82%, and 96.02%, respectively, which provides a better competitiveness in group activity recognition compared to other methods.

Keyword ：

Random walk Graph convolutional network Group activity recognition Levy flight Position information

Cite：

Copy from the list or Export to your reference management。

Abstract ：

Rack-and-pinion gear sets with variable transmission ratios have gradually become a standard core component in steering systems, as the harmony between portability and sensitivity in vehicle steering can be achieved with such applications. When the conventional meshing theory is adopted to calculate point clouds of variable-ratio tooth surfaces, drawbacks will occur with respect to unevenness in point cloud density, inaccessible tooth surface boundaries, and point cloud fusion at the interface between the working tooth surface and fillet. Thus, the generated point cloud of variable ratio tooth surfaces is inapplicable for computer-aided engineering (CAE) and plastic forming. In this study, the point cloud reconstruction of pinion tooth surfaces with variable transmission ratio was investigated. Meanwhile, based on the analysis of transient length in contact lines that occur during gear meshing, various criteria were proposed to determine the validity of the variable-ratio rack-and-pinion design in terms of whether continuous and reverse driving can be achieved. By adopting the proposed methods, the designed variable-ratio pinion tooth surfaces demonstrate point clouds with equidistant distributions and clear boundaries. CAE analysis results also indicate a good quality in gear meshing. The real-time gear meshing status of variable-ratio rack-and-pinion sets is characterized by the transient length of contact lines. The measured variation curves for the transmission ratios of a prototype demonstrate significant consistency with the theoretical curves, indicated by attaining transmission ratio errors within a reasonable range. The validity of the design methods proposed in this study has been verified by the aforementioned results.

Keyword ：

Steering box variable transmission ratio tooth surface pinion and rack tooth contact analysis non-involute gears

Cite：

Copy from the list or Export to your reference management。

Abstract ：

Using machine learning methods to estimate brain effective connectivity networks from functional magnetic resonance imaging (fMRI) data has gradually become one of the hot subjects in the fields of neuroscience. In particular, the encoder–decoder based methods can effectively extract the connections in fMRI time series, which have achieved promising performance. However, these methods generally use Granger causality model, which may identify false directions due to the non-stationary characteristic of fMRI data. Additionally, fMRI datasets have limited sample sizes, which significantly constrains the development of these methods. In this paper, we propose a novel brain effective connectivity estimation method based on causal autoencoder with meta-knowledge transfer, called MetaCAE. The proposed approach employs a causal autoencoder (CAE) to extract causal dependencies from non-stationary fMRI time series, and leverages meta-knowledge transfer to improve the estimation accuracy on small-sample data. More specifically, MetaCAE first employs a temporal convolutional encoder to extract non-stationary temporal information from fMRI time series. Then it uses a structural equation model-based decoder to decode causal relationships between brain regions. Finally, it utilizes a model-agnostic meta-learning method to learn the meta-knowledge of the shared brain effective connectivity among different subjects, and transfers the meta-knowledge to the CAE to enhance its estimation ability on small-sample fMRI data. Comprehensive experiments on both simulated and real-world data demonstrate the efficacy of MetaCAE in estimating brain effective connectivity. © 2024 Elsevier Ltd

Keyword ：

Small-sample fMRI data Causal autoencoder Brain effective connectivity Meta-knowledge transfer Meta-learning

Cite：

Copy from the list or Export to your reference management。

Abstract ：

With the powerful learning ability of deep convolutional networks, deep clustering methods can extract the most discriminative information from individual data and produce more satisfactory clustering results. However, existing deep clustering methods usually ignore the relationship between the data. Fortunately, the graph convolutional network can handle such relationships, opening a new research direction for deep clustering. In this paper, we propose a cross-attention based deep clustering framework, named Cross-Attention Fusion based Enhanced Graph Convolutional Network (CaEGCN), which contains four main modules: the cross-attention fusion module which innovatively concatenates the Content Auto-encoder module (CAE) relating to the individual data and Graph Convolutional Auto-encoder module (GAE) relating to the relationship between the data in a layer-by-layer manner, and the self-supervised model that highlights the discriminative information for clustering tasks. While the cross-attention fusion module fuses two kinds of heterogeneous representation, the CAE module supplements the content information for the GAE module, which avoids the over-smoothing problem of GCN. In the GAE module, two novel loss functions are proposed that reconstruct the content and relationship between the data, respectively. Finally, the self-supervised module constrains the distributions of the middle layer representations of CAE and GAE to be consistent. Experimental results on different types of datasets prove the superiority and robustness of the proposed CaEGCN.

Keyword ：

Image segmentation Clustering methods graph convolutional network Cross-attention fusion mechanism Smoothing methods Task analysis Image reconstruction Deep learning Data mining deep clustering

Cite：

Copy from the list or Export to your reference management。

Abstract ：

The performance of landing gear retraction mechanism in aircraft directly affects its safe operation. Therefore, it is important to analyze and evaluate its comprehensive performance during the design process. Multiple single kinematic and dynamic performance indexes of landing gear retraction mechanism could be solved by CAD/CAE software. The weighting factors of every single performance index are used to distinguish the different effects of comprehensive evaluation, and also achieved by the expert investigation method. Combining the a priori information of the mechanism, the comprehensive performance of landing gear retraction mechanism could be analyzed by Relative Principal Component Analysis (RPCA) method, and the scale of the landing gear retraction mechanism with the best comprehensive performance could be effectively selected. Further, RPCA could also provide a scientific reference basis for the optimization design of landing gear retraction mechanism.

Keyword ：

landing gear retraction mechanism PCA comprehensive performance evaluation Aircraft RPCA

Cite：

Copy from the list or Export to your reference management。

Abstract ：

An air-powered vehicle is a low-cost method to achieve low-pollution transportation, and compressed air engines (CAE) have become a research hotspot for their compact structure, low consumption, and wide working conditions. In this study, a pneumatic motor (PM) test bench is built and tested under different inlet pressures, operation modes, and three driving cycles. On the basis of the data obtained by sensors, power output, compressed air consumption rate, and efficiency are calculated to evaluate the pneumatic motor performances. The results show that with an increase in rotation speed, the output power and efficiency first increase and then decrease, and the compression air consumption rate decreases. With an increase in torque, the rotation speed decreases, and the power output and efficiency first increase and then decrease. With an increase in mass flow rate, the torque increases, the power output and efficiency first increase and then decrease. The pneumatic motor achieves the best performance under a rotation speed of 800–1200 rpm, where power output, efficiency, and compressed air consumption rates are 1498 W, 13.6%, and 10 J/g, respectively. The pneumatic motor achieves the best power output and efficiency under the UDDS driving cycle. © 2023 by the authors.

Keyword ：

compressed air engine operation mode driving cycle pneumatic motor

Cite：

Copy from the list or Export to your reference management。

Abstract ：

Based on MBSE, a TSRFA method was proposed to solve the problem of multi-parameter coupling design. The method is divided into three parts: Digital model analysis, Virtual prototype model, and Physical prototype model. By integrating methods such as qualitative analysis, quantitative analysis and evaluation integration (QQE) to build a unified demand analysis with multiple parameters of coupled agronomy; by integrating industrial design ideas, using CAD modeling and CAE performance simulation. Finally, the feasibility of TSRFA method is verified with the subject case, and the design process is provided for the subsequent development of intelligent agricultural equipment.

Keyword ：

Requirements Analysis Method Simulation Multi-parameter Coupling Design MBSE TSRFA

Cite：

Copy from the list or Export to your reference management。

Abstract ：

In semiconductor etching processes, fault detection monitors the quality of wafers. However, the detailed dynamics in batch data are ignored in many traditional methods. In this paper, sequential image-based data visualization and fault detection, using bi-kernel t-distributed stochastic neighbor embedding (t-SNE), is proposed for semiconductor etching processes. In the proposed method, multi-modals, multi-phases, and abnormal samples in batches are visualized in two-dimensional maps. First, the batch data are restructured into sequential images and input to a convolutional autoencoder (CAE) to learn the abstract representation. Then, bi-kernel t-SNE is applied to visualize the CAE codes in two-dimensional maps. To reduce the computational burden and overcome the out-of-sample projection diffusion in bi-kernel t-SNE, data subsampling is used in the training procedure. Finally, a one-class support vector machine is employed to calculate the visualization control boundary, and a batch-wise index is presented for fault wafer detection. To demonstrate the feasibility and effectiveness of the proposed method, it was applied to two wafer etching datasets. The results indicate that the proposed method outperforms state-of-the-art methods in data visualization and fault detection.

Keyword ：

Etching Transforms Convolutional autoencoder (CAE) Kernel Indexes fault detection Training Data visualization data visualization t-distributed stochastic neighbor embedding (t-SNE) Fault detection semiconductor manufacturing

Cite：

Copy from the list or Export to your reference management。

Abstract ：

Accurate and fast spatial temporal trajectory similarity measure is the foundation of spatial temporal trajectory data mining. Massive data, spatial temporal heterogeneous distribution and data noise bring great challenges for spatial temporal trajectory similarity comparison. With the idea of similar image matching in computer vision, we propose a fast computational method TISM-CAE for trajectory image structure matching based on convolutional auto-encoder. Firstly, we remap the given spatial temporal trajectory slices into a two-dimensional matrix for generating trajectory images. Secondly, we use a convolutional auto-encoder network to obtain the low-dimensional features of trajectory images by unsupervised learning. Finally, the trajectory similarity is equivalent to compare the Euclidean distance between two low-dimensional features. We use real floating vehicle dataset of Shanghai and artificial simulated trajectory dataset for experimental analysis. Final results show that the proposed method improves the accuracy of similar trajectory identification and reduces the time complexity by nearly 50% compared with the currently used similarity measure method as Longest Common Subsequence(LCSS) and Edit Distance on Real sequence(EDR), providing a new direction for trajectory similarity measure. © 2022 ACM.

Keyword ：

Learning systems Convolution Convolutional neural networks Data mining Unsupervised learning Trajectories Network coding

Cite：

Copy from the list or Export to your reference management。

Abstract ：

等几何边界元法(IGABEM)具有边界元法的计算优势(只需在模型边界离散,降低问题维数,计算精度高等),同时可以实现CAD/CAE的无缝集成,提高分析效率。本文采用IGABEM对IGBT的DBC基板进行传热分析,得到其温度分布,同时研究陶瓷层厚度变化对结构中温度的影响。结果表明:固定铜层厚度时,适当减小陶瓷层厚度对结构是有益的。

Keyword ：

温度分布 等几何边界元法 DBC基板

Cite：

Copy from the list or Export to your reference management。