SIA OpenIR  > 信息服务与智能控制技术研究室
可变形体模特机器人控制策略研究
Alternative TitleResearch of the control strategy for the Deformable Humanoid robot
武云峰1,2
Department信息服务与智能控制技术研究室
Thesis Advisor南琳
ClassificationTP242
KeywordCan总线 Pid 尺寸拟合 Bp神经网络 模糊推理
Call NumberTP242/W94/2013
Pages73页
Degree Discipline模式识别与智能系统
Degree Name硕士
2013-05-28
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract本文重点对SIAHMR-I的CAN总线智能节点设计、上身躯干关键尺寸拟合和尺寸变形控制进行了研究、仿真和实现。在CAN总线智能节点设计方面,阐述了智能节点的设计方案和软件设计方式;利用MFC双缓冲的原理在上位机实现了控制响应曲线的动态显示软件,方便了执行机构控制器的调试;详细介绍了自制微型电动推杆的双闭环PID控制器设计、调试和实现,并对双闭环PID控制器进行了一系列的改进,提高了电动推杆的控制效果;分别介绍了SIAHMR-I所用直流无刷电机和步进电机的控制器设计方案。 在SIAHMR-I上身躯干关键尺寸拟合方面,针对SIAHMR-I关键尺寸仿人特征和建模难的特点,提出采用神经网络拟合尺寸的算法。首先,运行SIAHMR-I关键尺寸获取样本数据,在Matlab中对样本数据进行反向传播算法(BP)网络进行训练拟合;其次,对比BP网络和其改进型网络的迭代次数、拟合误差和训练时间等参数,选择了基于Levenberg-Marquardt的反向传播算法;最后,在Simulink中利用基本Matlab数学工具箱和神经网络工具箱实现选择的BP网络并进行了样本输入输出的验证对比。 在尺寸变形控制方面,提出了基于模糊推理的尺寸变形方案。首先,针对关键尺寸变形特点,对比了不同的控制方案并借鉴一阶直线倒立摆控制原理提出了基于模糊推理的控制方案;其次,以腰关节为例分别定义了隶属度函数、 设计模糊控制规则集和选择解模糊策略;最后,在Simulink中实现模糊控制器,通过仿真实验对比选定解模糊算法、比例因子等。仿真结果验证了本文所提出的控制方案,为今后的实践提供了理论依据和指导。
Other AbstractThe research, simulation and implementation for the CAN bus intelligent node design, the size fitting for the upper torso and shape-size changing stressed in this paper. In CAN bus bus intelligent node design, elaborate the design scheme and design mode of software; By taking advantage of the principle of double buffering in MFC, implemented the dynamic display software for the response curve, facilitating the adjusting of the controller. The designing, debugging and implementing of the double closed PID loop is exhaustive introduced in this paper, and for the purpose of improvement of the control effect, some improvement is added to the PID controller; The controller designing for the brushless DC motor and stepper motor used into the SIAHMR-I is presented respectively in this paper. In size fitting of the upper torso for SIAHMR-I, As characteristic such as humanoid shape-size changing and the difficulty for modeling, shape-size fitting make use of Neural-Network is presented. Firstly, obtaining the sample data by means of running of the crucial joint of SIAHMR-I and training the Back-Propagation(BP) Network in Matlab environment by using of the sample data. Secondly, the BP Network based on Levenberg-Marquardt method is adapted by contrast of the basic BP Network and some improvement methods for the BP Network.Finally, implementing that BP Network, testing and verifying the input data and output data in the Simulink environment by using of the mathematical toolbox and Neural Network toolbox. In size fitting for the upper torso, the shap-size changing based on the Fuzzy reasoning is presented. Firstly, Take the humanhoid shape-size changing trait into consideration, the shape-size changing based on the Fuzzy reasoning is adapted by contrasting the different control scheme and learning the control strategy of an Inverted Pendulum system; Secondly, taking the waist joint of the SIAHMR-I for example, define the membership function, fuzzy control rules set and select the defuzzification method; Finally, realize the fuzzy logic controller in Simulink, selecting the defuzzification method and proportion factor etc. by comparing the simulation result. The result proved the correctness of the control strategy presented in this paper and provide the theoretical basis and direction for the practical application in the future.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/10748
Collection信息服务与智能控制技术研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
武云峰. 可变形体模特机器人控制策略研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2013.
Files in This Item:
File Name/Size DocType Version Access License
可变形体模特机器人控制策略研究.pdf(1886KB) 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[武云峰]'s Articles
Baidu academic
Similar articles in Baidu academic
[武云峰]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[武云峰]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.