检索结果-内蒙古大学图书馆

2023 China automation Congress, CAC 2023

作者： Zhou, You Huang, Zihao Zeng, Deyu Wu, Zongze School of Automation Guangdong University of Technology Guangzhou China College of Mechatronics and Control Engineering Shenzhen University Shenzhen China

ISBN: (纸本)9798350303759

Unsupervised image anomaly detection and segmentation algorithms hold significant importance in enhancing product quality in real-world industrial manufacturing. Anomaly detection methods based on the knowledge distillation framework perform detection by characterizing differences between anomalous samples. However, knowledge distillation-based anomaly detection methods often employ similar or identical architectures for constructing both teacher and student models. This practice has led to the issue of vanishing anomaly activation values, resulting in detection failures. Hence, the proposed Multi-Channel Knowledge Distillation model (MCKD) consists of a pair of student networks proficient at effectively capturing defect-related information. One branch of the student network enhances its diverse representation of anomalies by restoring the input's multi-scale representations. The other branch enriches the model's capacity to represent detailed defect information by extracting features from different scales. Additionally, our design of the Residual Features Aggregation Module (RFAM) ensures better retention of normal pattern information within the student network while disregarding information from abnormal patterns, resulting in enhanced modeling of positive samples. The proposed framework achieves state-of-the-art (SOTA) on the well-known MVTec AD dataset, with an area under the receiver operating characteristic curve of 97.1% for anomaly detection across all 15 categories and an area under the anomaly segmentation of 98.14%. © 2023 IEEE.

关键词： Distillation

来源：评论

学校读者我要写书评

暂无评论

controller Dynamic Linearization Based Data-Driven Adaptive control for a Vapor-Compression Refrigeration System 13

Controller Dynamic Linearization Based Data-Driven Adaptive ...

引用

13th IEEE Data Driven control and Learning Systems Conference, DDCLS 2024

作者： Fang, Jingyu Yu, Xian Xiong, Shuangshuang College of Mechatronics and Control Engineering Shenzhen University Shenzhen518060 China School of Automation Beijing Information Science and Technology University Beijing100192 China

ISBN: (纸本)9798350361674

This paper applies a data-driven adaptive control method to a benchmark of vapor compression refrigeration system based on the controller dynamic linearization method. Two controllers, namely, a single-input single-output controller and an extended multiple-input multiple-output controller, are designed to track the exit temperature of the evaporator secondary flux and the refrigerant superheat at the evaporator exit, by controlling the valve opening of the evaporator, and the compressor speed. Both the controllers entirely rely on input-output data of the controlled system in the controller design. Comparative results between the applied data-driven adaptive control and the default PID control provided by the benchmark demonstrate the effectiveness of the two designed controllers. © 2024 IEEE.

关键词： Vapor compression refrigeration

来源：评论

学校读者我要写书评

暂无评论

Real-Time Implementation of MEKF using VDPL Localization Algorithm by Utilizing MATLAB & ROS Communication

Real-Time Implementation of MEKF using VDPL Localization Alg...

引用

2024 Innovations in Intelligent Systems and Applications Conference, ASYU 2024

作者： Altınpınar, Ozan Vahit Contarlı, Emre Can Sezer, Volkan Department of Mechatronics Engineering Smart and Autonomous Systems Laboratory Istanbul Technical University Istanbul Turkey Department of Control and Automation Engineering Smart and Autonomous Systems Laboratory Istanbul Technical University Istanbul Turkey

ISBN: (纸本)9798350379433

Mobile robots, which are produced to fulfill certain tasks to make our lives easier, are becoming more and more common in our environment day by day. In the long run, these robots can be more cost-effective, faster, and more efficient than humans in accomplishing tasks. In order for mobile robots to fulfill the tasks, they need to be able to safely navigate to specific destinations without hitting obstacles. Safely reaching a given goal most accurately and shortly is the problem of autonomy. This autonomy problem is divided into sub-problems and algorithms are needed to solve each sub-problem. The main sub-problems in the field of autonomy are localization, mapping, local planning, and global planning. The main purpose of this paper is to describe how the position information computed by a MEKF (Modified Extended Kalman Filter) localization algorithm using the VDPL (Virtual Dynamic Point Landmark) algorithm, which was previously designed in MATLAB, is published to the ROS environment via a topic. This position information can be used together with the local and global planning algorithms in the navigation stack to realize fully autonomous driving capability in a real-time environment. Using these algorithms, full-autonomous driving tests were conducted on the Turtlebot3 mobile robot in a real-time environment according to a predefined scenario. The performance results of both simulation and real-world tests show that the MEKF using the VDPL algorithm outperforms the AMCL algorithm in the navigation stack in both cases. © 2024 IEEE.

关键词： Extended Kalman filters

来源：评论

学校读者我要写书评

暂无评论

Reward Drops in Learning-based control with an Experimental Validation on Microdrones 63

Reward Drops in Learning-based Control with an Experimental ...

引用

63rd IEEE Conference on Decision and control, CDC 2024

作者： Tukenmez, Nejat Fotiadis, Filippos Junior, Jose M. Magalhaes Vamvoudakis, Kyriakos G. Bogosyan, Seta O. Istanbul Technical University Mechatronics Engineering Department Istanbul34469 Turkey Georgia Institute of Technology School of Aerospace Engineering AtlantaGA United States Istanbul Technical University Control and Automation Engineering Department Istanbul34469 Turkey

ISBN: (纸本)9798350316339

In this paper, we consider a computationally efficient learning-based control mechanism dealing with dense reward processing for a zero-sum game. The problem has been formulated as the online learning of the Nash equilibrium without requiring any information on the system dynamics. It has firstly been constructed as an infinite horizon optimal control problem, then as an online model-free Q-learning framework, which is composed of critic and actor networks (i.e., for the control and disturbance input). The closed-loop system is also proved to have a stable equilibrium point even in the presence of reward drops. The efficacy of the learning-based controller has been validated through simulations and experiments on micro drones. © 2024 IEEE.

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Design of A Novel Primary Permanent Magnet Linear Machine with Low Thrust Ripple 7

Design of A Novel Primary Permanent Magnet Linear Machine wi...

引用

7th IEEE Student Conference on Electric Machines and Systems, SCEMS 2024

作者： Chai, Jiawei Xu, Jiajia Gui, Xianguo Gao, Qiang School of Electrical Engineering & Automation Harbin Institute of Technology Harbin China Innovation Center for Control Actuators Beijing Institute of Precision Mechatronics and Controls Beijing China

ISBN: (纸本)9798350366945

In order to reduce thrust ripple, this paper proposes a new primary permanent magnet linear machine with uneven pole (UP-PPMLM). Firstly, the proposed structure and advantages compared to conventional structure are introduced. Then, the expressions of air-gap flux density by permanent magnets (PMs) and armature reaction are derived. Finally, by using finite element method (FEM), the new topology reasonability is proved. Meanwhile, the performances of proposed machines are compared to conventional topology, and the result confirms the electromagnetic performances of the proposed machine have been greatly improved. © 2024 IEEE.

关键词： Permanent magnets

来源：评论

学校读者我要写书评

暂无评论

System for Determining the Shape of the Wavefront of a Nonsinusoidal Periodic Acoustic Wave 17

System for Determining the Shape of the Wavefront of a Nonsi...

引用

17th International Conference on engineering of Modern Electric Systems, EMES 2023

作者： Codoban, Adrian Coroiu, Laura Gergely, Eugen Sarosi, Jozsef University of Oradea Department Control Systems Engineering and Management Oradea Romania University of Szeged Department of Mechatronics and Automation Szeged Hungary

ISBN: (纸本)9798350310634

The shape of the wavefront is important for the most realistic reproduction of the acoustic wave. In acoustics, the wave front (or wave surface) is defined as the totality of points in the space of the propagating medium that have the same instantaneous elongation amplitude. In the ideal case, in a homogeneous and isotropic propagation medium without other disturbances, the wave has a spherical symmetry, i.e. the shape of the wavefront is circular and at large distances from the wave source, due to dispersion, the shape of the wavefront can be considered plan. In practice, however, due to the multiple component harmonics, reflections and possible summations of the primary acoustic wave, the elongation has a non-sinusoidal shape, and the wavefront may have an irregular random shape. Moreover, if the source of the wave or bodies from which it is reflected are in motion, the shape of the wavefront may even be variable in time. The paper presents a method for determining the shape of the wave front of a non-sinusoidal periodic plane acoustic wave, resulting from the addition of several harmonic components, reflected waves, etc. It uses an array of omnidirectional capacitive microphones arranged in line, at equal distances between them. At each of the microphones, the acoustic wave is captured out of phase depending on the position of the microphone in space and the shape of the wave front. These phase shifts placed next to each other give an image of the shape of the wavefront captured by the microphone system. The method presented in this paper reproduces sounds very realistically and can be used in Active Noise Cancel systems, in open spaces, with one or more mobile noise sources. One of the benefits is the increase in hearing comfort in noisy environments. Another is to record sounds with high fidelity. © 2023 IEEE.

关键词： Wavefronts

来源：评论

学校读者我要写书评

暂无评论

Self-Tuning Neural Network controller Based on Fuzzy Logic for Multiple Positions Tracking of a Pneumatic Driven Soft Endoscope Actuator 10

Self-Tuning Neural Network Controller Based on Fuzzy Logic f...

引用

10th International Conference on automation, Robotics, and Applications, ICARA 2024

作者： Acosta, Renzo Tafur, Julio Canahuire, Ruth Pontificia Universidad Catolica del Peru Control and Automation Engineering Master Degree Program Enginnering Department Lima Peru Universidad de Ingenieria y Tecnologia - UTEC Department of Electrical and Mechatronics Engineering Lima Peru

ISBN: (纸本)9798350394245

This paper presents the dynamic modelling and end effector position control of a soft endoscope. Soft endoscope system under study consists mainly of a pneumatic driven soft actuator (PDSA) with four independently chambers. Recurrent neural network (RNN) and dynamic back propagation (DBP) training algorithm are used to obtain PDSA dynamic model. PDSA position controller is based on a feedforward neural network (FNN) and it is trained using PDSA closed loop system (CLS) and DBP training algorithm. CLS stability analysis concludes that it is not possible control PDSA end effector position using one position controller for different end effector initial and desired positions. Fuzzy logic methodology is used to integrate several positions controllers in a one controller valid for all operation range. The controller implementation and close loop system simulation are performed in MATLAB for different end effector desired position to validate system performance in all range of operation. © 2024 IEEE.

关键词： Pneumatics

来源：评论

学校读者我要写书评

暂无评论

Robust control Performance in Uncertain Environments: A Semi-Heuristic Gradient Descent Approach 8

Robust Control Performance in Uncertain Environments: A Semi...

引用

8th International Conference on Instrumentation, control, and automation, ICA 2023

作者： Ahmed Hassan, Mohammed Elbadri Irham Jambak, Ahmad Bayezit, Ismail Soylemez, Mehmet Turan Istanbul Technical University Department of Control and Automation Engineering Istanbul Turkey Istanbul Technical University Department of Mechatronics Engineering Istanbul Turkey Istanbul Technical University Department of Aeronautical Engineering Istanbul Turkey

ISBN: (纸本)9798350301274

Designing a robust controller for systems with parameter uncertainties is a complex and demanding task. Traditional deterministic and probabilistic approaches may fall short in providing efficient and satisfactory solutions. To address this challenge, we propose a semi-heuristic approach that exploits the Kharitonov theorem to establish an initial point for the gradient descent algorithm. Through an iterative optimization process that incorporates user-defined performance criteria, our approach provide a robust controller with respect to presence of system uncertainties. Numerical simulations validate the effectiveness of our proposed method and highlight its superiority in addressing robust control problems compared with auto-tuned ΡΠ) controller. © 2023 IEEE.

关键词： Optimization

来源：评论

学校读者我要写书评

暂无评论

Balancing Precision and Speed: Introducing The Performance Efficiency Evaluation Ratio (PEER) in Visual Odometry

引用

IEEE Access 2025年 13卷 89874-89904页

作者： Atilgan, Cem Mercimek, Muharrem Kırklareli University Department of Mechatronics Engineering Kırklareli Turkey Yildiz Technical University Department of Control and Automation Engineering Istanbul Turkey

Visual odometry is extensively utilized in robotics, autonomous navigation, and augmented reality to estimate motion by analyzing sequential images. Provides high precision without relying on external sensors such as GPS. However, real-time visual odometry systems must strike a balance between computational correctness and speed, a critical yet often overlooked trade-off in existing evaluation methods. Traditional metrics such as ATE, RPE, RMSE, FPS, and RTF typically assess precision or efficiency in isolation and lack adaptability to real-time constraints. To address this limitation, we propose the Performance Efficiency Evaluation Ratio (PEER), a novel, adaptive, and lightweight metric that jointly evaluates algorithm performance based on both fidelity and computation time. PEER incorporates a tunable weighting parameter to prioritize performance, speed, or a balanced trade-off, and employs normalization techniques to ensure comparability across different algorithms and systems. We evaluated PEER using the KITTI dataset with various feature extraction algorithms (SIFT, ORB, BRISK, KAZE, AKAZE) and matching algorithms (BF, FLANN). The findings indicate that PEER not only distinguishes performance effectively across various scenarios but also shows strong alignment with established MCDM methods such as TOPSIS and VIKOR, all while maintaining lower computational complexity and greater suitability for real-time deployment. Overall, PEER provides a robust and flexible framework for optimizing visual odometry systems. © 2013 IEEE.

关键词： Augmented reality

来源：评论

学校读者我要写书评

暂无评论

Reinforcement Learning Based Optimization of Autonomous Aircraft Performance 11

Reinforcement Learning Based Optimization of Autonomous Airc...

引用

11th International Conference on mechatronics and Robotics engineering, ICMRE 2025

作者： Sari, Metin Caliskan, Fikret Istanbul Technical University Department of Mechatronics Engineering Istanbul Turkey Turkish Aerospace Industries Ankara Turkey Istanbul Technical University Department of Control and Automation Istanbul Turkey

ISBN: (纸本)9798331509293

This research introduces an advanced flight control system for optimizing autonomous aircraft performance, leveraging deep reinforcement learning (DRL) to address the complexities of nonlinear flight dynamics. Using a six-degree-of-freedom (6-DoF) rigid aircraft flight dynamics model, we develop a Deep Deterministic Policy Gradient (DDPG) controller tailored for waypoint navigation and attitude stabilization tasks. A custom reward framework and extensive hyperparameter tuning enable effective training within a high-fidelity MATLAB/Simulink environment, achieving high rewards and precise control. Although computationally intensive, the simulations demonstrate robust performance across diverse flight conditions, with potential for real-world applications and future extensions to multi-agent scenarios. © 2025 IEEE.

关键词： Deep reinforcement learning

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：