Bond Graph Modeling of Mobile Robot and Its Parameter Identification
-
摘要: 为了研究移动机器人CyCab的转向系统,本文对该系统进行了键合图建模,得到了电机和皮带构成的子系统的模型方程。为了对模型中的未知参数进行精确辨识,提出了一种将最小二乘法与BP神经网络相结合的方法,将最小二乘法辨识结果作为BP神经网络训练的初始权值,并且引入权重因子以改善训练过程的收敛性。利用最小二乘法和改造后的BP神经网络对模型中的未知参数进行了辨识,实验结果表明经过改造的BP神经网络辨识算法对系统模型未知参数具有更高的辨识精度。Abstract: We build the bond graph model of the steering system of the mobile robot CyCab and obtain the modeling equations for its motor and belt. In order to accurately identify the unknown parameters in the model, we propose a method that combines the least squares method with the BP (back propagation) neural network. We utilize the identification results obtained with the least square method as the initial weighing values of the BP neural network training. We also introduce weighing factors to improve the convergence of the training process. Finally we use the least squares method and the improved BP neural network to identify the unknown parameters in the model. The ex-perimental results show that our improved BP neural network identification algorithm has higher precision for identif-ying the unknown parameters.
-
Key words:
- bond graph modeling /
- parameter identification /
- neural network
-
[1] 王中双,陆念力.键合图理论及应用研究若干问题的发展及现状[J].机械科学与技术,2008,27(1):78～83 [2] Zhao J,Wang F Q.Parameter identification by neural network forintelligent deep drawing of axisymmetric workpieces[J].Journalof Materials Processing Technology,2005,166:387～391 [3] Liang Y C,Feng D P,Liu G R,Yang X W,Han X.Neural iden-tification of rock parameters using fuzzy adaptive learning parame-ters[J].Computers and Structures,2003,81:2373～2382 [4] 陈恩伟,刘正士,干方建.机器人末端臂惯性参数辨识的人工神经网络方法[J].中国机械工程,2006,17(3):173～178 [5] 庄未,刘晓平.机器人关节面参数的行波与神经网络混合辨识方法[J].振动工程学报,2010,23(2):268～271 [6] 王中双.键合图理论及其在系统动力学中的应用[M].哈尔滨工程大学出版社,2007 -
点击查看大图
计量
- 文章访问数: 235
- HTML全文浏览量: 13
- PDF下载量: 2
- 被引次数: 0
下载:
