检索结果-内蒙古大学图书馆

6th International Conference on Energy, Power and Grid, ICEPG 2024

作者： Zhou, Ninghui Su, Baisong Yang, Peng Liu, Gaowei Gu, Yue State Grid Corporation of China East China Branch Shanghai200120 China East China Electric Power Design Institute Shanghai200063 China

ISBN: (纸本)9798350377798

In this paper, a method for optimizing the retrofit scheme of relay protection devices in East China Power Grid based on Deep Deterministic Policy Gradient (ddpg) algorithm is proposed. By constructing a state evaluation model and life cycle management method, the timing and strategy of relay protection device retrofit are optimized by combining the ddpg algorithm. The experimental results show that the average trouble-free operation time of the modified equipment increased from 555 hours to 710 hours;the correct action rate increased from 89.83% to 94.83%;the number of power outages was reduced from 5.3 to 3.3, and the duration of outages was reduced from 10.6 hours to 6.6 hours. In the conclusion of the above data, the ddpg optimization scheme can significantly improve the operational reliability of relay protection devices and the stability and security of the power grid. © 2024 IEEE.

关键词： ddpg algorithm Deep Reinforcement Learning Device Modification Optimization Power System Relay Protection

来源：评论

学校读者我要写书评

暂无评论

Fuzzy-based collective pitch control for wind turbine via deep reinforcement learning

引用

ISA TRANSACTIONS 2024年 148卷 307-325页

作者： Nabeel, Abdelhamid Lasheen, Ahmed Elshafei, Abdel Latif Zahab, Essam Aboul Cairo Univ Fac Engn Elect Power Dept Giza 12613 Egypt

Wind turbines (WTs) have highly nonlinear and uncertain dynamics due to aerodynamic complexity, mechanical factors, and fluctuations in wind conditions. Turbulence and wind shear add complexity to modelling, especially in constant power region (region 3). Thus, an effective control design demands a deep understanding of the nonlinearities and uncertainties. This paper suggests a novel model-free reinforcement learning (RL) collective pitch angle controller to operate efficiently in region 3. The proposed controller stabilizes generator speed, maximizes power output, and minimizes fluctuations while accommodating system uncertainties, nonlinearity, and pitch limits. The disparity between WT dynamics due to wind speed perturbations and uncertainties is measured using a gap-metric criterion. The controller design adopts a deep deterministic policy gradient (ddpg) algorithm to train six agents in a medium-fidelity WT environment at different mean wind speeds to ensure the controller's robustness. Initially, imitation learning is used for efficient sample collection to fasten training convergence. Afterwards, the agent learns by interacting with the environment. After the training, the pitch control outputs from multi-trained agents are processed by a fuzzy system to have smooth transitions under different operating conditions. The resulting fuzzy ddpg (F-ddpg) controller is deployed to obtain the optimal pitch control action. The performance of the proposed F-ddpg controller is compared to the gain-scheduled PI (GSPI), Linear-Quadratic-Regulator (LQR), and single-ddpg-agent controllers. The controllers are simulated in high-fidelity onshore and offshore 5-MW WT environments using the OpenFAST/MATLAB simulation tools. The results reveal the superiority of the proposed controller in generalizing its optimal performance in different operating conditions.

关键词： Reinforcement learning Imitation learning ddpg algorithm Fuzzy logic Collective pitch control Wind turbines

来源：评论

学校读者我要写书评

暂无评论

Neural Network algorithm for Intercepting Targets Moving along Known Trajectories by a Dubins' Car

引用

AUTOMATION AND REMOTE CONTROL 2023年第3期84卷 196-210页

作者： Galyaev, A. A. Medvedev, A. I. Nasonov, I. A. Russian Acad Sci Trapeznikov Inst Control Sci Moscow Russia

The task of intercepting a target moving along a rectilinear or circular trajectory by a Dubins' car is formulated as a problem of time-optimal control with an arbitrary direction of the car's velocity at the time of interception. To solve this problem and to synthesize interception trajectories, neural network methods of unsupervised learning based on the Deep Deterministic Policy Gradient algorithm are used. The analysis of the obtained control laws and interception trajectories is carried out in comparison with the analytical solutions of the interception problem. Mathematical modeling of the target motion parameters, which the neural network had not previously seen during training, is carried out. Model experiments are conducted to test the stability of the neural solution. The effectiveness of using neural network methods for the synthesis of interception trajectories for given classes of target movements is shown.

关键词： interception task Dubins' car ddpg algorithm neural network synthesis of trajectories

来源：评论

学校读者我要写书评

暂无评论

DRL-based frequency response of wind turbine generators adapting their variable regulation capabilities

引用

IET RENEWABLE POWER GENERATION 2023年第3期17卷 495-506页

作者： Ding, Zhetong Wu, Jiping Wang, Ying Shan, Shuo Yuan, Kun Zhang, Kaifeng Southeast Univ Key Lab Measurement & Control CSE Minist Educ Nanjing 210000 Jiangsu Peoples R China NARI Technol Co Ltd Nanjing Jiangsu Peoples R China

Frequency response of wind turbine generators (WTGs) is a crucial technology to ensure safe integration of increasing future wind power into power grids. WTGs' regulation capabilities are variable compared to conventional generations due to uncertain wind speed and dynamic deloading factors, which affects the design of their frequency response strategy. This paper proposes a novel deep reinforcement learning (DRL)-based frequency response framework of WTGs, combining a deep deterministic policy gradient (ddpg)-based control and a proportional power sharing (PPS) strategy. The ddpg-based control dynamically determines the optimal total regulation power of WTGs and the PPS strategy allocates the total regulation power to each WTG. Benefiting from the ability of ddpg algorithm to handle multi-dimensional continuous system states and actions, the proposed ddpg-control enables a wind farm to dynamically determine the optimal total frequency regulation power under WTGs' variable regulation capabilities. Besides, the application of deep neural networks (DNN) in ddpg algorithm makes the ddpg-based control possible to be implemented in complex nonlinear systems. Simulations on an IEEE 39-bus system demonstrate that effectiveness of the proposed frequency response strategy of WTGs.

关键词： dynamic deloading factors ddpg algorithm frequency response strategy WTGs' variable regulation capabilities Optimisation techniques deep deterministic policy gradient-based control Wind power plants uncertain wind speed frequency control wind farm safe integration optimal total frequency regulation power deep reinforcement learning-based frequency response framework future wind power WTGs' regulation capabilities proportional power sharing strategy gradient methods reinforcement learning Frequency control Neural nets ddpg-based control deep neural networks wind turbines frequency response Control engineering computing DRL-based frequency response wind power plants deep learning (artificial intelligence) ddpg-control crucial technology power generation control Control of electric power systems Reinforcement learning power grids control engineering computing Interpolation and function approximation (numerical analysis) wind power PPS strategy Power engineering computing power engineering computing optimal total regulation power optimisation turbogenerators wind turbine generators

来源：评论

学校读者我要写书评

暂无评论

A LARGE-SCALE PATH PLANNING algorithm FOR UNDERWATER ROBOTS BASED ON DEEP REINFORCEMENT LEARNING

引用

INTERNATIONAL JOURNAL OF ROBOTICS & AUTOMATION 2024年第3期39卷 204-210页

作者： Wang, Wenhui Li, Leqing Ye, Fumeng Peng, Yumin Ma, Yiming CSG PGC Power Storage Res Inst Guangzhou 510000 Peoples R China

To ensure the effect and improve the accuracy of large-scale path planning for underwater robots, a large-scale algorithm for planning the path for underwater robots based on deep reinforcement learning is proposed. Deep reinforcement learning is analysed, and the idea, structure, network update method, and training process of deep deterministic policy gradients (ddpg) algorithm are described. A fitness learning model of the robot which under water is confirmed to describe the mathematical relationship between the geographical location and operating speed of the underwater robots. On this basis, ddpg algorithm is applied in large-scale path planning of underwater robots. TensorFlow is used to build Actor and Critic neural network structures, and design environment state models, action state spaces, and reward functions. In deep reinforcement learning, the large-scale navigation planning for the underwater robot, through exploration-online trial and error, finds the optimal search strategy, and considers obtaining the maximum expected reward during the path planning procedure, achieving the large-scale path planning for the underwater robot. According to the experimental results, the proposed algorithm demonstrates good performance in large-scale path planning for underwater robots and effectively improves both the accuracy and efficiency of the planning process.

关键词： ddpg algorithm reward function deep reinforcement learning underwater robot large-scale path planning

来源：评论

学校读者我要写书评

暂无评论

A Reinforcement Learning Approach for Continuum Robot Control

引用

JOURNAL OF INTELLIGENT & ROBOTIC SYSTEMS 2023年第4期109卷 1-14页

作者： Kargin, Turhan Can Kolota, Jakub Poznan Univ Tech Inst Automat Control & Robot Piotrowo 3A PL-60965 Poznan Poland

Rigid joint manipulators are limited in their movement and degrees of freedom (DOF), while continuum robots possess a continuous backbone that allows for free movement and multiple DOF. Continuum robots move by bending over a section, taking inspiration from biological manipulators such as tentacles and trunks. This paper presents an implementation of a forward kinematics and velocity kinematics model to describe the planar continuum robot, along with the application of reinforcement learning (RL) as a control algorithm. In this paper, we have adopted the planar constant curvature representation for the forward kinematic modeling. This choice was made due to its straightforward implementation and its potential to fill the literature gap in the field RL-based control for planar continuum robots. The intended control mechanism is achieved through the use of Deep Deterministic Policy Gradient (ddpg), a RL algorithm that is suited for learning controls in continuous action spaces. After simulating the algorithm, it was observed that the planar continuum robot can autonomously move from any initial point to any desired goal point within the task space of the robot. By analyzing the results, we wanted to recommend a future direction for research in the field of continuum robot control, specifically in the application of RL algorithms. One potential area of focus could be the integration of sensory feedback, such as vision or force sensing, to improve the robot's ability to navigate complex environments. Additionally, exploring the use of different RL algorithms, such as Proximal Policy Optimization (PPO) or Trust Region Policy Optimization (TRPO), could lead to further advancements in the field. Overall, this paper demonstrates the potential for RL-based control of continuum robots and highlights the importance of continued research in this area.

关键词： Reinforcement Learning ddpg algorithm Continuum robot

来源：评论

学校读者我要写书评

暂无评论

A Novel Approach to Light Detection and Ranging Sensor Placement for Autonomous Driving Vehicles Using Deep Deterministic Policy Gradient algorithm

SAE INTERNATIONAL JOURNAL OF CONNECTED AND AUTOMATED VEHICLE...

引用

SAE INTERNATIONAL JOURNAL OF CONNECTED AND AUTOMATED VEHICLES 2024年第3期7卷 303-308页

作者： Berens, Felix Ambs, Jordan Elser, Stefan Reischl, Markus Univ Appl Sci Ravensburg Weingarten Inst Artificial Intelligence Weingarten Germany Univ Appl Sci Ravensburg Weingarten Weingarten Germany Karlsruhe Inst Technol Inst Automat & Appl Informat Karlsruhe Germany

This article presents a novel approach to optimize the placement of light detection and ranging (LiDAR) sensors in autonomous driving vehicles using machine learning. As autonomous driving technology advances, LiDAR sensors play a crucial role in providing accurate collision data for environmental perception. The proposed method employs the deep deterministic policy gradient (ddpg) algorithm, which takes the vehicle's surface geometry as input and generates optimized 3D sensor positions with predicted high visibility. Through extensive experiments on various vehicle shapes and a rectangular cuboid, the effectiveness and adaptability of the proposed method are demonstrated. Importantly, the trained network can efficiently evaluate new vehicle shapes without the need for re-optimization, representing a significant improvement over classical methods such as genetic algorithms. By leveraging machine learning techniques, this research streamlines the sensor placement optimization process, enhancing the perception capabilities of autonomous driving vehicles. The optimized sensor configurations obtained from the ddpg algorithm lead to safer and more reliable autonomous driving systems, contributing to the advancement and widespread adoption of autonomous driving technology.

关键词： Autonomous vehicle LiDAR configuration Optimization problem ddpg algorithm

来源：评论

学校读者我要写书评

暂无评论

A gait control method for bipedal robots based on reward function optimization 4

A gait control method for bipedal robots based on reward fun...

引用

4th International Conference on Intelligent Communications and Computing

作者： Pei, Ruoyan Ma, Pengge Zhang, Tong Li, Lan Zhengzhou Univ Aeronaut Zhengzhou Peoples R China Beijing Inst Technol Beijing Peoples R China

ISBN: (纸本)9798331530372;9798331530365

The intelligent control of walking of humanoid robots is one of the key research directions in the field of robotics research, but the traditional motion model's constraint on the center of mass makes it difficult for robots to maintain walking stability and simulate human gait at the same time for its excessive attention on stability. Compared with the traditional deep reinforcement learning algorithm, the ddpg algorithm has the ability to handle continuous action space and high-dimensional state space, and has a wide application prospect in many practical problems. We proposed a bipedal robot control method based on reward function optimization to enable the robot to achieve both stable high-speed movement and human gait imitation. This paper combines the physical characteristics of the bipedal robot to establish a control system based on the ddpg algorithm. At the same time, the reward function is designed to guide the robot to learn the correct walking strategy. Through the comparative test, the weight limit ratio of each reward function for the bipedal robot to stably increase the speed is given. The simulation results show that the method proposed in this paper has good practicality and effectiveness.

关键词： Bipedal robot gait control deep reinforcement learning ddpg algorithm reward function

来源：评论

学校读者我要写书评

暂无评论

The Strategy for Lane-keeping Vehicle Tasks based on Deep Reinforcement Learning Continuous Control 24

The Strategy for Lane-keeping Vehicle Tasks based on Deep Re...

引用

International Conference on Machine Intelligence and Digital Applications (MIDA)

作者： Li, Qianxi Fei, Rong Xian Univ Technol Dept Comp Sci & Technol Xian Shaanxi Peoples R China

ISBN: (纸本)9798400718144

Autonomous driving technology has become a hot research field, with lane-keeping being one of the fundamental tasks in autonomous driving systems. Deep reinforcement learning can enhance the lane-keeping performance of autonomous vehicles by learning from experience, exploring vehicle-following control behaviors, and providing a solid technical reference for intelligent driving control. ddpg, as a classic algorithm in deep reinforcement learning, has advantages in addressing continuity problems. However, for lane-keeping task decisions, suitable reward functions are essential to ensure the safety, efficiency, and comfort of vehicle-following behavior. This paper analyzes collision time, headway distance, and acceleration change rate to design corresponding reward functions that guarantee safe, efficient, and comfortable driving. Additionally, when designing the action-state space, human psychological factors and expected deviations are considered. Parameters such as deviation, velocity difference, and following vehicle speed are taken as inputs to the model. Through experimenting with different actions and adjusting acceleration decisions based on feedback from these actions and the environment, the cumulative changes in reward functions during driving are analyzed. It is found that vehicles can drive smoothly and efficiently in both congested and uncongested conditions.

关键词： Autonomous driving technology Lane keeping task Deep reinforcement learning ddpg algorithm Reward function

来源：评论

学校读者我要写书评

暂无评论

Speed and heading control of an unmanned surface vehicle using deep reinforcement learning 12

Speed and heading control of an unmanned surface vehicle usi...

引用

IEEE 12th Data Driven Control and Learning Systems Conference (DDCLS)

作者： Wu, Ting Ye, Hui Xiang, Zhengrong Yang, Xiaofei Jiangsu Univ Sci & Technol Sch Automat Zhenjiang 212100 Jiangsu Peoples R China Nanjing Univ Sci & Technol Sch Automat Nanjing 210094 Peoples R China

ISBN: (纸本)9798350321050

In this paper, a deep reinforcement learning-based speed and heading control method is proposed for an unmanned surface vehicle (USV). A deep deterministic policy gradient (ddpg) algorithm which combines with an actor-critic reinforcement learning mechanism, is adopted to provide continuous control variables by interacting with the environment. Moreover, two types of reward functions are created for speed and heading control of the USV. The control policy is trained by trial and error so that the USV can be guided to achieve the desired speed and heading angle steadily and rapidly. Simulation results verify the feasibility and effectiveness of the proposed approach by comparisons with classical PID control and S plane control.

关键词： Deep reinforcement learning ddpg algorithm unmanned surface vehicle

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：