检索结果-内蒙古大学图书馆

Influencing factor analysis of interception probability and classification-regression neural network based estimation

引用

Journal of Systems engineering and Electronics 2023年第4期34卷 992-1006页

作者： NAN Yi YI Guoxing HU Lei WANG Changhong TU Zhenbiao Department of Control Science and Engineering Harbin Institute of TechnologyHarbin 150001China Beijing Electro-mechanical Engineering Institute Beijing 100074China

The interception probability of a single missile is the basis for combat plan design and weapon performance evaluation,while its influencing factors are complex and mutually *** calculation methods have very limited analysis of the influence mechanism of influencing factors,and none of them has analyzed the influence of the guidance *** paper considers the influencing factors of both the interceptor and the target more *** parameters include speed,guidance law,guidance error,fuze error,and fragment killing ability,while target performance includes speed,maneuverability,and *** this paper,an interception model is established,Monte Carlo simulation is carried out,and the influence mechanism of each factor is analyzed based on the model and simulation ***,this paper proposes a classification-regression neural network to quickly estimate the interception probability based on the value of influencing *** proposed method reduces the interference of invalid interception data to valid data,so its prediction accuracy is significantly better than that of pure regression neural networks.

关键词： interception probability simulation modeling analysis of influencing factors probability estimation neural networks

来源：评论

学校读者我要写书评

暂无评论

Event-based sliding mode control under denial-of-service attacks

引用

Science China(Information Sciences) 2022年第6期65卷 125-141页

作者： Yingxin TIAN Xiang LI Bo DONG Yabin GAO Ligang WU Department of Control Science and Engineering Harbin Institute of Technology School of Software Harbin Institute of Technology

This paper deals with the problem of estimator-based sliding mode control against denial-of-service(DoS) attacks and discrete events via a time-delay approach. A networked system is considered an uncertain dynamical system with matched and mismatched perturbations and exogenous disturbance in the network environment. A network-resource-aware event-triggering mechanism is designed with aperiodically releasing system measurements. Furthermore, to describe the DoS attack duration and inter-event time, a time-delay modeling approach considers the DOS attack duration and inter-event time as a “time delay” of the measurements between the sensor and controller over the network is proposed. Consequently, an intervaltime-delay system with uncertainties is formulated. A state-observer-based sliding mode controller, by which the ideal sliding mode can be achieved, is proposed against the DoS attacks. The resulting sliding motion is proved to be robust and stable with an Lgain performance. Finally, the effectiveness and applicability of the present sliding mode control are validated in a simulated pendulum system.

关键词： sliding mode control time delay uncertain system state observer event trigger

来源：评论

学校读者我要写书评

暂无评论

Facial Expression Recognition via Closed-Loop Transcription 50

Facial Expression Recognition via Closed-Loop Transcription

引用

50th Annual Conference of the IEEE Industrial Electronics Society, IECON 2024

作者： Liu, Xuan Ma, Jiachen Wang, Qiang Department of Control Science and Engineering Harbin Institute of Technology Harbin China Department of Control Science and Engineering Harbin Institute of Technology Weihai China

ISBN: (纸本)9781665464543

Convolutional Neural Networks (CNNs) are the most widely used and successful approaches for accomplishing the Facial Expression Recognition (FER) task. However, these technologies appear to have reached a bottleneck since the networks are becoming increasingly complex and heavier, yet the performance improvements remain minimal. The "black box" nature of CNNs makes it challenging to identify and address this crux. This paper aims to address the issue by utilizing the principles of data compression and discriminative representation within a plausible theoretical framework. We are employing an innovative computational framework, drawing inspiration from feedback loop control systems, to develop an interpretable closed-loop transcription model to generate a Linear Discriminative Representation (LDR) for multi-class real-world datasets. This model, comprised of an encoder and decoder that facilitates data mapping between the data space and the feature space, employs the output of the transcription system as feedback or guidance, with the objective of improving the accuracy and efficiency of the transcription process. The optimal LDR is learned jointly for all data classes through a two-player minimax game between the encoder and decoder for the learned representation over a single rate reduction-based objective. The studies conducted on the FER2013 dataset show promising results for the proposed closed-loop formulation. The visual quality of the learned decoder and the classification performance of the encoder are comparable to those of large and complex networks. An accuracy of 70.97% was attained on the FER2013 dataset by applying Principal Component Analysis (PCA) to the learned representation. © 2024 IEEE.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Contrastive Augmented Graph2Graph Memory Interaction for Few Shot Continual Learning

引用

IEEE Transactions on Circuits and Systems for Video technology 2025年第6期35卷 5805-5820页

作者： Qi, Biqing Gao, Junqi Chen, Xinquan Li, Dong Liu, Jianxing Wu, Ligang Zhou, Bowen Tsinghua University Department of Electronic Engineering Beijing China Shanghai Artificial Intelligence Laboratory Shanghai China Harbin Institute of Technology Department of Control Science and Engineering Harbin China Harbin Institute of Technology School of Mathematics Harbin China

Few-Shot Class-Incremental Learning (FSCIL) has gained considerable attention in recent years for its pivotal role in addressing continuously arriving classes. However, it encounters additional challenges. The scarcity of samples in new sessions intensifies overfitting, causing incompatibility between the output features of new and old classes, thereby escalating catastrophic forgetting. A prevalent strategy involves mitigating catastrophic forgetting through the Explicit Memory (EM), which comprise of class prototypes. However, current EM-based methods retrieves memory globally by performing Vector-to-Vector (V2V) interaction between features corresponding to the input and prototypes stored in EM, neglecting the geometric structure of local features. This hinders the accurate modeling of their positional relationships. To incorporate information of local geometric structure, we extend the V2V interaction to Graph-to-Graph (G2G) interaction. For enhancing local structures for better G2G alignment and the prevention of local feature collapse, we propose the Local Graph Preservation (LGP) mechanism. Additionally, to address sample scarcity in classes from new sessions, the Contrast-Augmented G2G (CAG2G) is introduced to promote the aggregation of same class features thus helps few-shot learning. Extensive comparisons on CIFAR100, CUB200, and the challenging ImageNet-R dataset demonstrate the superiority of our method over existing methods. © 1991-2012 IEEE.

关键词： Prototypes Adaptation models Power capacitors Feature extraction Data models Accuracy Vectors Training Incremental learning Overfitting

来源：评论

学校读者我要写书评

暂无评论

Dual Spatial-Temporal Feature Pyramid With Decoupled Temporal Mining for Video Semantic Segmentation

IEEE Transactions on Intelligent Vehicles

引用

IEEE Transactions on Intelligent Vehicles 2024年 1-12页

作者： Hong, Yuanduo Pan, Huihui Sun, Weichao Wang, Yulei Bian, Ning Gao, Huijun Research Institute of Intelligent Control and Systems Harbin Institute of Technology Harbin China Department of Control Science and Engineering Tongji University Shanghai China Dongfeng Motor Corporation Wuhan China

Semantic segmentation plays a pivotal role in environmental perception for autonomous driving. Video semantic segmentation (VSS) further takes temporal information into consideration for better scene parsing and temporal consistency. Previous research on VSS is mostly dedicated to developing new techniques (e.g. optical flows, attention) to better mine temporal information. In this work, we contribute from a different angle by efficiently incorporating multi-scale temporal information. The dual spatial-temporal feature pyramid is proposed to enable the direct enhancement of multi-scale features for target frames and unlash the design of temporal information mining modules. It contains a spatial feature pyramid from a target frame and a spatial-temporal feature pyramid from multiple reference frames. Building on the dual feature pyramid, we further propose to decouple motional contexts and static contexts to fully leverage temporal information. Specifically, multi-scale motional contexts are mined with the introduced dedicated module and static contexts are enhanced by making temporally fused category-level representations interact with the target frame feature. The final segmentation maps are obtained by regarding the enhanced category-level representations as powerful feature classifiers to classify the target frame feature of rich motional contexts. Experimental results on two popular VSS benchmarks demonstrate that the proposed method with decent parameter and inference efficiency clearly outperforms previous advanced methods. IEEE

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Hyperbolic Sine Function-Based Full-State Feedback Attitude Tracking control for Rigid Spacecraft

引用

IEEE Transactions on Systems, Man, and Cybernetics: Systems 2025年第5期55卷 3412-3424页

作者： Dong, Rui-Qi Wu, Ai-Guo Li, Bin Duan, Guang-Ren Department of Automation Shenzhen518055 China Sichuan University School of Aeronautics and Astronautics Chengdu610207 China Southern University of Science and Technology Center for Control Science and Technology Shenzhen518055 China Harbin Institute of Technology Center for Control Theory and Guidance Technology Harbin150001 China

The attitude tracking control with unwinding-free performance for rigid spacecraft is studied in this article. A full-state feedback control law based on a hyperbolic sine function is developed such that the resulted closed-loop system can achieve two stable equilibria. By Lyapunov stability theory and Barbalat's Lemma, it is proven that the obtained closed-loop system is almost globally asymptotically stable, and achieves unwinding-free performance. Further, by constructing a strict Lyapunov function, it is demonstrated that the two stable equilibria are exponentially stable. Moreover, subsets of attraction regions corresponding to each stable equilibrium are characterized. The simulation results illustrate that the proposed attitude control scheme can effectively avoid the unwinding problem during attitude tracking. © 2025 IEEE.

关键词： Lyapunov functions

来源：评论

学校读者我要写书评

暂无评论

Friction Modeling and control Compensation for Electromechanical Braking Systems 8

Friction Modeling and Control Compensation for Electromechan...

引用

8th CAA International Conference on Vehicular control and Intelligence, CVCI 2024

作者： Houhua, Jing Qinggan, Lin Mingqi, Liu Harbin Institute of Technology Department of Control Science and Engineering Harbin China

ISBN: (纸本)9798331504892

Influenced by the electrification and intelligence of automobile chassis, electromechanical braking has become the future brake-by-wire trend choice by virtue of its simpler mechanism and faster servo performance, while system friction, as a key factor of its servo control, must be handled properly. This work studies the system friction modeling of electromechanical braking system and the corresponding control compensation approach. First, by evaluating the structure and working principle of the electromechanical braking system, the system model is produced, the braking control test rig is built and the system characteristic test is carried out. Then, the friction model parameters were identified from the experimental data based on the examination of system features. Finally, the brake control performance after friction compensation was experimentally validated by the physical rig, which verified the effectiveness of the friction model and its control compensation. © 2024 IEEE.

关键词： Braking performance

来源：评论

学校读者我要写书评

暂无评论

Estimation of RoA for Nonlinear Systems Based on Gaussian Processes 6

Estimation of RoA for Nonlinear Systems Based on Gaussian Pr...

引用

6th International Conference on Electronic engineering and Informatics, EEI 2024

作者： Chen, Yucheng Zhu, Yupeng Zhu, Hongming Zhang, Quanqi Wu, Chengwei Harbin Institute of Technology Department of Control Science and Engineering Harbin China

ISBN: (纸本)9798350353594

This paper employs Gaussian Processes (GPs) to estimate uncertainty dynamics inherent in nonlinear systems. A control policy ensuring the system satisfies initial stability constraints is derived using linear quadratic regulator (LQR). By modeling the Lyapunov function based on uncertain dynamics as GPs, the stability of the system dynamics in the Lyapunov sense is ensured. To discretize the Lyapunov function with Lipschitz continuity, a finite number of discrete points are chosen to represent the properties of the function. The region of attraction (RoA) based on Lyapunov function is widely utilized as a safe constraint in the field of control. Additionally, the maximum uncertainty dynamics are chosen as the boundary for exploration to expand the evaluation of the RoA. Finally, a simulation example is provided to demonstrate the effectiveness of the proposed approach. © 2024 IEEE.

关键词： Lyapunov functions

来源：评论

学校读者我要写书评

暂无评论

Generative Adversarial Network with In-Betweening for Fine-Grained Skeleton-Based Action Generation 50

Generative Adversarial Network with In-Betweening for Fine-G...

引用

50th Annual Conference of the IEEE Industrial Electronics Society, IECON 2024

作者： Qi, Xiangyuan He, Zhen Wang, Qiang Department of Control Science and Engineering Harbin Institute of Technology Harbin China

ISBN: (纸本)9781665464543

Human action generation has shown critical value in both industry application and academic research, especially fine-grained skeleton-based action generation. At present, there are several main ways to achieve fine-grained skeleton-based action generation. In this work, we utilize a generative adversarial network with in-betweening, in order to generate natural and continuous skeleton sequences and better capture complex motion details. In-betweening can be viewed as a prediction task based on the past and future context. Besides, it is convenient to control the result in a personalized way and combine the framework with other network structures to further improve the performance. In particular, two simple discriminators are applied on different timescales and LSTM serves as the foundation of the action generator because of its capability of effectively capturing long-term dependencies and processing sequence data of different lengths. The target noise added to the input of the generator can contribute to the robustness of the system and further improve the performance. Moreover, a novel Tai Chi dataset composed of fine-grained skeleton-based action data is created using high quality motion capture technology. The model is evaluated on two fine-grained skeleton-based datasets, LaFAN1 and Tai Chi, which contain common daily actions and professional sports actions respectively, and it achieves superior performance both qualitatively and quantitatively. © 2024 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Hyperparameter Optimization of Object Detection Networks Based on Large Language Models 50

Hyperparameter Optimization of Object Detection Networks Bas...

引用

50th Annual Conference of the IEEE Industrial Electronics Society, IECON 2024

作者： Hu, Yushuo Wang, Qiang Department of Control Science and Engineering Harbin Institute of Technology Harbin China

ISBN: (纸本)9781665464543

Since the emergence of deep learning, the configuration of hyperparameters has been one of the most significant problems that people have paid attention to. However, traditional hyperparameter optimization techniques typically have drawbacks such as expensive calculation consumption and lengthy training time. With the development of large language models (LLMs), particularly models such as generative pretrained transformer (GPT), it is achievable to utilize the models’ analysis ability to guide parameter space search. This paper investigates the use of LLMs for hyperparameter optimization in machine learning problems. To evaluate the effectiveness of this strategy, four object detection models, including YOLOv5, YOLOv8, Faster R-CNN, and R-FCN, were used to conduct experiments. At the same time, we compared the distinctions between this approach and traditional ways. The experimental results indicate that combining LLMs with the hyperparameter optimization process can perform comparably or better than conventional methods. © 2024 IEEE.

关键词： Deep 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：