检索结果-内蒙古大学图书馆

Contour-texture preservation transformer for face super-resolution

NEUROCOMPUTING 2025年 626卷

作者： Wu, Ziyi Zhang, Yanduo Lu, Tao Zhao, Kanghui Wang, Jiaming Wuhan Inst Technol Hubei Key Lab Intelligent Robot Wuhan Peoples R China

In the field of computer vision, face super-resolution (FSR) technology is an important tool for enhancing the performance of basic tasks such as face recognition and video surveillance. However, when faced with complex face images, existing FSR methods often rely on the Transformer model to improve image quality through its powerful global modeling capabilities. Yet, they tend to be slightly insufficient in local feature extraction due to a lack of adequate local detail capture capabilities. To alleviate these problems, we propose a novel contour- texture preservation Transformer (CTP) method for FSR. This method consists of two key components: the multi-scale attention enhancement block (MSAEB), which captures and fuses image features of different scales to improve the detail level of feature representation, and provides high-quality input for the contour-texture Transformer enhancement block (CTTEB). Additionally, CTTEB integrates convolution operations to enhance local feature extraction and improve feature expression. The feed-forward network (FFN) we introduced ensures the full fusion of global and local information. By combining MSAEB and CTTEB into a residual progressive attention group (RPAG), the network gradually extracts and fuses multi-scale features, ultimately achieving dual preservation of contour structure and texture details. Experiments show that our method achieves the best results on LFW, FFHQ, CelebA, and Helen, with a 0.32 dB increase in PSNR, a 0.0126 increase in SSIM, and a 0.0056 increase in FSIM over the second-best model. Experiments on real-world datasets SCface and Chokepoint confirm that the CTP method excels in both FSR reconstruction and face recognition, verifying its effectiveness.

关键词： Face super-resolution Contour structure Texture details Face recognition

来源：评论

学校读者我要写书评

暂无评论

RO-BEV: Towards Robust BEV Feature Enhancement for Surround-View Instance Prediction

引用

ELECTRONICS LETTERS 2025年第1期61卷

作者： Yang, Zhi Lu, Tao Wang, Jiaming Lang, Xiujuan Fu, Shichang Han, Jifeng Wuhan Inst Technol Hubei Key Lab Intelligent Robot Wuhan Peoples R China

The most existing instance prediction methods often lack practical understanding and enhancement of bird's-eye view (BEV) spatial-temporal features, particularly distant features. To alleviate the above problems, we propose a robust BEV feature enhancement network (RO-BEV). The proposed RO-BEV includes two major components: the novel adaptive linearly increasing dividing (ALID) strategy and the hierarchical cross-scale fusion (HCSF) module. Specifically, the ALID strategy can mitigate instance loss caused by insufficient features by providing a linearly increasing 3D representation to build a robust BEV feature. Then, the proposed HCSF module enhances spatial-temporal BEV features in the latent space to predict vehicle instances by sampling from feature distributions at different scales and timestamps. Experimental results on the nuScenes dataset show that our RO-BEV outperforms existing state-of-the-art methods.

关键词： autonomous driving Bird's-eye view computer vision image processing instance prediction

来源：评论

学校读者我要写书评

暂无评论

Interfacial Microstructure and Mechanical Properties of TiC-Ni Cermet Joint Brazed with Ag-Cu28 Alloy

引用

JOURNAL OF MATERIALS ENGINEERING AND PERFORMANCE 2025年 1-10页

作者： Zheng, Qingze Lei, Min Liu, Haichun Wei, Zixiong Li, Yulong Nanchang Univ Sch Adv Mfg Jiangxi Key Lab Intelligent Robot Nanchang 330031 Peoples R China

A surface Ni/Cu coating was applied to the TiC-Ni cermet using the resistive evaporation method. Subsequently, the coated cermet was joined to itself by vacuum brazing using the Ag-Cu28 brazing alloy. The study investigated the formation mechanism of the joints and the impacts of heating temperature and holding time on the joint's interfacial microstructure and mechanical properties. The results showed that during the heat treatment, the atoms within the Ni/Cu coating interdiffused, forming a continuous (Cu, Ni) layer adjacent to the cermet interface, which could promote the wettability of the liquid alloy and increase the shear strength of joints. The typical microstructure of the joints was TiC-Ni/(Cu, Ni)/Ag(s,s) + Cu(s,s)/(Cu, Ni)/TiC-Ni. Meanwhile, the joint shear strength achieved a maximum value of 146 MPa with a heating temperature of 800 degrees C for 10 min. And the joint fracture occurred in the interfacial region between the base material and the (Cu, Ni) reaction layer. The reaction layer became discontinuous with an increase in temperature or time. With increasing temperature, the shear strength tended to increase first and then decrease. With the extension of time, the shear strength gradually decreased.

关键词： brazing coating interfacial microstructure mechanical properties TiC-Ni cermet

来源：评论

学校读者我要写书评

暂无评论

MPOC-SLAM: An RGB-D SLAM System With Motion Probability and Object Category in High Dynamic Environments

引用

IEEE-ASME TRANSACTIONS ON MECHATRONICS 2025年第2期30卷 1061-1071页

作者： Wu, Songyang Zhang, Xuebo Zhang, Shiyong Song, Zhixing Wang, Runhua Yuan, Jing Nankai Univ Inst PLXSPACE Robot & Automat Informat Syst Coll Artificial Intelligence Tianjin Key Lab Intelligent Robot Tianjin 300350 Peoples R China

The performance of simultaneous localization and mapping (SLAM) is significantly impacted by dynamic objects. Existing dynamic SLAM methods still suffer from some major issues, such as insufficient use of semantic information, losing track when dynamic objects occupy the field of view, and inability to achieve real-time processing. In this article, we present SLAM with motion probability and object category (MPOC-SLAM), an RGB-D SLAM system capable of high-performance, real-time work in high dynamic environments. Considering intensive semantic information, our method introduces the concepts of motion probability (MP) and object category (OC), and tightly integrates them within the whole SLAM system. First, a priority feature extraction strategy is proposed based on MP, and then a dual-constraint feature matching method is designed with the adaptive probability threshold and category consistency. Both methods reduce the influence of dynamic objects in the extraction and association process. Second, to speed up the process of calculating reprojection errors, a semantic-guided geometric method with the partitioning hypothetico-deductive method is presented. The semantic-guided geometric method and instance segmentation are then fused through the variance weighted fusion with their confidences. Finally, as a more important part of SLAM, dynamic-aware weights are added to the original pose optimization, which assigns larger weights to static feature points and map points to reduce their alteration during optimization. We perform evaluations on the TUM public dataset and real-world environments. Compared with ORB-SLAM2, the percentage reduction of root mean square error of absolute trajectory error is as high as 98.72%. Results show that MPOC-SLAM realizes a significant improvement in accuracy as well as a more robust and real-time performance compared with other state-of-the-art dynamic SLAM methods.

关键词： Dynamic environments instance segmentation motion probability (MP) object category (OC) visual simultaneous localization and mapping (SLAM)

来源：评论

学校读者我要写书评

暂无评论

A four-beam piezoelectric transducer for energy harvesting and tissue hardness detection

引用

MECHANICAL SYSTEMS AND SIGNAL PROCESSING 2025年 224卷

作者： Zhang, Jiantao Cong, Xingang Yu, Xiangfu Luo, Yi Li, Yanlong Shanghai Univ Sch Mechatron Engn & Automat Shanghai Key Lab Intelligent Mfg & Robot Shanghai Peoples R China

This paper develops a piezoelectric device that can be used for both vibration energy harvesting and soft tissue hardness detection. The piezoelectric transducer allows for simultaneous energy harvesting and hardness testing. When the piezoelectric transducer is used as an energy harvester, the four-beam piezoelectric vibration energy harvester has a good voltage output when the angle alpha between each beam of the harvester and the base plane is 0 degrees. The output power is maximum when the external load resistance is 100 k Omega. When the piezoelectric transducer is used for hardness detection, the feasibility of the hardness measurement scheme is verified by simulation analysis. By simulating the results of contact materials with different hardness levels, a calibration curve for the sensor's modal frequency and the elastic modulus of the tested object was fitted. Finally, a prototype four-beam piezoelectric sensor for energy harvesting and tissue hardness detection was fabricated. An experimental platform for vibration energy harvesting and soft tissue hardness detection was developed and different experiments were conducted. When the piezoelectric transducer is used as an energy harvester, the experimental results of the output characteristic change curve are basically consistent with the simulation results. The maximum rms output voltage of the harvester is about 423 mV and the maximum output power is about 1.79 mu W. When piezoelectric sensors are used as hardness detection sensors, it has high accuracy within the range of 0 similar to 800 kPa. The sensor can effectively detect the location of hard objects within soft samples and identify the hardness of the tested samples.

关键词： Piezoelectric Energy harvester Tactile sensor Hardness detection Theoretical model Finite element analysis

来源：评论

学校读者我要写书评

暂无评论

A percussion-type piezoelectric vibration energy harvester

引用

MECHANICS BASED DESIGN OF STRUCTURES AND MACHINES 2025年

作者： Zhang, Jiantao Li, Yingjie Ning, Yiwen Yu, Xiangfu Shanghai Univ Sch Mechatron Engn & Automat Shanghai Key Lab Intelligent Mfg & Robot Shanghai Peoples R China

Nonlinear piezoelectric energy harvesters utilize actively designed structural dynamics to introduce nonlinear restoring forces, which enable large-amplitude oscillations over a low-frequency broadband range. This approach can enhance the efficiency of ambient vibrational energy harvesting under specific excitation conditions, compared to traditional linear systems. In this study, a novel percussive piezoelectric vibration energy harvester is developed, which, due to its symmetric cantilever beam structure, generates a knocking force under external excitation causing the harvester to switch between two states. The electromechanical coupling model of the percussive piezoelectric vibration energy harvester is established. Numerical simulations are performed to analyze the effects of different design parameters on its output performance. Three configurations of harvester prototypes are fabricated and tested. When the excitation frequency is 12 Hz and the axial compression distance of the PVDF beam is 0.2 mm, the prototype 1 achieves the maximum RMS output voltage of 3.61 V. When the excitation frequency is 12 Hz, the axial compression distance of the PVDF beam is 0.2 mm, and the load resistance is 0.2 M Omega, the maximum output power of prototype 1 reaches 32.7 mu W. The validity of the proposed model is validated by comparing simulation results with experimental data. In addition, the experimental results of the three prototypes show that the structural parameters of the harvester can be effectively designed based on the vibration characteristics of the environmental vibration source.

关键词： Piezoelectric energy harvester electromechanical coupling model percussion response numerical simulations nonlinear vibrations

来源：评论

学校读者我要写书评

暂无评论

A Differential-Mechanism-Based Leg Configuration Balances the Load and Dynamic Contribution for All Actuators of the Quadruped robot

引用

IEEE TRANSACTIONS ON robotICS 2025年 41卷 2014-2030页

作者： Wang, Zeyu Jia, Wenchuan Sun, Yi Bao, Tianxu Ding, Zihan Chen, Qi Shanghai Univ Sch Mechatron Engn & Automat Shanghai Key Lab Intelligent Mfg & Robot Shanghai 200444 Peoples R China

Kinematic performance of a quadruped robot is determined by the mechanical structure. This article presents a novel leg structure for legged robots that integrates a differential mechanism into the conventional design. This approach enables all actuators to be positioned within the robot's torso at fixed locations, significantly reducing the leg's inertia. Furthermore, the new structure introduces a parallel transmission system that balances motion and torque distribution among the joint actuators, effectively reducing torque peaks and enhancing the drive capability during dynamic motions. A family of configurations of differential leg structures is constructed, and their mapping to the classic serial leg structure is dissected in kinematic and mathematic. Simulations of various single-leg models are conducted to validate the performance of the new configuration under typical gait conditions. Subsequently, a leg prototype is designed, manufactured, and tested in experiments involving tasks, such as trajectory tracking, weighted squats, and squat jumps. The development of a prototype quadruped robot featuring this novel leg structure is also presented.

关键词： Legged locomotion Actuators robots Gears Knee Couplings Hip Dynamics Thigh Quadrupedal robots Differential mechanism kinematics legged robots mechanism design

来源：评论

学校读者我要写书评

暂无评论

A modular multi-step forecasting method for offshore wind power clusters

引用

APPLIED ENERGY 2025年 380卷

作者： Fang, Lei He, Bin Yu, Sheng Shanghai Univ Sch Mechatron Engn & Automat Shanghai Key Lab Intelligent Mfg & Robot Shanghai 200444 Peoples R China

Offshore wind farm clusters, driven by economies of scale, are emerging as a prevalent trend. However, the intermittency and volatility of offshore wind power due to wind resource uncertainties pose significant challenges for forecasting. Existing research on offshore wind farm cluster power forecasting remains limited. This paper addresses this gap by proposing a modular and decoupled multi-step forecasting method for offshore wind farm clusters. The modular design enables adaptability to various forecasting scenarios, particularly with and without Numerical Weather Prediction (NWP) data, providing a flexible framework for future research and applications. The method leverages the spatiotemporal information of all wind farms within the cluster by first preprocessing the historical power output series using signal processing techniques, including Fast Fourier Transform (FFT) and Singular Value Decomposition (SVD), to decompose and denoise the data. Spatiotemporal feature extraction is then achieved through an Enhanced Long Short-Term Memory (LSTM) network, combining two-dimensional convolution and LSTM layers. Subsequently, incorporating NWP data, a Light Gradient Boosting Machine (LightGBM) model is employed for final forecasting. Finally, the proposed method is validated on a wind farm cluster in the eastern coastal region of China, demonstrating its effectiveness, accuracy, and generalizability at both individual wind farm and cluster levels.

关键词： Modular Multi-step forecasting Enhanced LSTM LightGBM Offshore wind power cluster

来源：评论

学校读者我要写书评

暂无评论

Multi-step short-term forecasting of photovoltaic power utilizing TimesNet with enhanced feature extraction and a novel loss function

引用

APPLIED ENERGY 2025年 388卷

作者： Yu, Sheng He, Bin Fang, Lei Shanghai Univ Sch Mechatron Engn & Automation Shanghai Key Lab Intelligent Mfg & Robot Shanghai 200444 Peoples R China

The instability of weather conditions often causes photovoltaic power generation to exhibit randomness and volatility, making accurate and reliable photovoltaic power forecasting crucial for the stable scheduling of integrated energy systems. Multi-step forecasting remains a challenge due to the difficulty in capturing temporal dependencies among neighboring discrete time points, which is attributable to the limited expressiveness of time-series features using one-dimensional modeling methods. Hence, this paper proposes a methodological framework tailored for multi-step short-term forecasting of photovoltaic power generation. The framework is based on the TimesNet architecture, which models meteorological features in two dimensions to enhance feature expressiveness. Additionally, a new feature extraction module is introduced to replace the Inception module in the original TimesNet, mitigating issues of feature redundancy and convolution kernel sharing associated with standard convolution. This enhancement aims to improve TimesNet's ability to recognize critical information. Considering the inevitable presence of outliers in datasets and the drawbacks of traditional loss functions, which are sensitive to outliers or struggle to fit nonlinear relationships, this paper proposes a novel loss function to overcome these limitations. To validate the performance of the proposed method, it was tested on three datasets across four prediction horizons (1 h, 3 h, 6 h, and 12 h ahead). Compared to the original TimesNet, it reduces the average RMSE and MAPE by 3.21 % and 9.36 % for the 12-h prediction. Compared to LightTS, Informer, and DLinear, it reduces the average MAE by 16.45 %, 24.62 %, and 11.41 % for the 12-h prediction, respectively. The proposed loss function also outperforms traditional loss functions (MAE, MSE, Huber, Log-Cosh) with an optimal metrics rate averaging 77 %. These results demonstrate that the proposed model and loss function achieve excellent accuracy in

关键词： Multi-step forecasting Photovoltaic power TimesNet Feature extraction Loss functions

来源：评论

学校读者我要写书评

暂无评论

An improved cross-machine transfer strategy based on multi-source domain knowledge for abnormal sample recognition

引用

RELIABILITY ENGINEERING & SYSTEM SAFETY 2025年 257卷

作者： Yan, Zhenhao Zhou, Bingqiang Gao, Zenggui Nong, Weiping Liu, Lilan Sun, Yanning Shanghai Univ Sch Mechatron Engn & Automat Shanghai Key Lab Intelligent Mfg & Robot Shanghai 200072 Peoples R China

Cross-machine transfer has garnered significant attention owing to its capacity to transfer diagnostic knowledge across different machines and address unforeseen operating conditions. Nevertheless, the personalized sample biases arising from industrial-specific conditions reduce the generalization performance of traditional crossmachine diagnostic methods. To address this, an enhanced cross-machine transfer strategy with multi-source domain knowledge (CMMK) is proposed for bearing fault diagnosis. Specifically, targeted training of model parameters is conducted to address the task challenges encountered in cross-device diagnosis. Multiple sets of source domain data are introduced for collaborative training, effectively mitigating feature discrepancies between samples from different distributions. To address ambiguous fault samples at class boundaries, adversarial training between independent task classifiers is incorporated, enabling precise fault identification under consistent working conditions. Furthermore, we introduce the custom threshold module and propose a novel residual block structure, which makes each residual block generate its own adversarial mechanism. Note that as the training progresses, the network parameters gradually evolve in a direction that aligns with the requirements of cross-device diagnosis. Finally, comprehensive experiments on extensive bearing fault datasets validate the superior diagnostic accuracy and generalization ability of the proposed CMMK compared to state-of-the-art methods.

关键词： Cross-machine fault diagnosis Residual neural network Domain adaptation Multi-source domain

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：