基于神经网络的大柔性灵巧手指弯曲角度预测  

The bending angle prediction of large flexible dexterous finger based on neural network

作　　者：杨庆华[1] 罗鑫远[1] 邵铁锋[2] 鲍官军[1] 王志恒[1] 

机构地区：[1]浙江工业大学特种装备制造与先进加工技术教育部/浙江省重点实验室,浙江杭州310032 [2]中国计量学院工程训练中心,浙江杭州310018

出　　处：《浙江工业大学学报》2013年第4期360-363,共4页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金资助项目(51075363);浙江省自然科学基金杰出青年团队资助项目(R1090674);浙江省自然科学基金资助项目(LY12E05022);中国博士后科学基金资助项目(2012M511385)

摘　　要：介绍了一种新型大柔性灵巧手指,该手指弯曲性能好、易于控制但数学建模复杂.基于BP神经网络基本理论,建立预测该手指弯曲角度的BP神经网络模型.通过大柔性灵巧手指弯曲特性实验获得样本数据,借助于MATLAB仿真软件中的神经网络工具箱作为开发平台,将实验样本数据用于BP网络训练.利用训练好的BP网络模型对手指弯曲角度进行预测,预测误差范围控制在3%以内.研究结果表明:这种神经网络模型能够准确预测手指的弯曲角度.A new kind of large flexible dexterous finger was presented, the finger has good bending performance and can be controlled easily however with complicated mathematical modeling. Based on BP neural network theory, a BP network model was established to predict the bending angle of the finger. The sample data obtained through the experiment of the finger bending characteristics were used to train BP network using neural network toolbox in MATLAB software. The trained BP network model was used to predict the bending angle of the finger. The prediction error is controlled less than 3%. The result shows the network model can accurately predict the bending angle of the finger.

关 键 词：大柔性灵巧手指 弯曲角度 神经网络 预测 

分 类 号：TP241[自动化与计算机技术—检测技术与自动化装置]