改进型神经网络在木件打磨机器人中的应用  

Application of Improved Neural Network in Woodpiece Polishing Robot

作　　者：张连滨[1] 葛浙东[1] 鞠明远 刘存根[1] 周玉成[1] ZHANG Lian - bin GE Zhe - dong JU Ming - yuan LIU Cun - gen ZHOU Yu - cheng(School of Information and Electrical Engineering,Shandong Jianzhu University,Jinan Shandong 250101,China)

机构地区：[1]山东建筑大学信息与电气工程学院,山东济南250101

出　　处：《林业机械与木工设备》2017年第4期19-22,25,共5页Forestry Machinery & Woodworking Equipment

基　　金：山东省泰山学者优势特色学科人才团队支持计划项目(2015162);博士基金项目"基于X射线的木结构建筑用材无损检测系统研究"(XNBS1622)

摘　　要：针对木件机械打磨加工效率低、产品质量差等问题,提出一种打磨轨迹跟踪算法,该算法是对模糊算法和神经网络算法的结合及改进。阐述了打磨机器人的系统结构,并针对两关节机器人建立动力学模型,基于打磨轨迹跟踪算法设计了模糊神经网络控制器。利用Simulink软件对打磨轨迹跟踪算法进行仿真试验,结果表明:改进型神经网络与传统神经网络算法相比,0时刻开始,跟踪误差均在±0.01 mm范围内,系统很好地完成了启动控制,并且打磨机器人在1 s时即可以实现极小误差的曲线打磨轨迹跟踪。该算法可有效控制打磨机器人的运动轨迹,工作时机器人运行稳定可靠,加工效率高。In view of the problems of low polishing efficiency and poor product quality of wood machinery,the paper puts forward a polishing trajectory tracking algorithm,which is a combination and also improvement of a fuzzy algorithm and a neural network algorithm. Firstly,the robot manipulator’s polishing system composition is stated,a kinetic model is established based on two-joint robots and based on the polishing trajectory tracking algorithm,a fuzzy neural network controller is designed. Simulink software is used to conduct simulation testing on the polishing trajectory tracking algorithm,with the result showing that compared with the traditional neural network algorithm,the improved neural network starts at 0： 00,with tracking error within 0. 01 mm,the system able to complete startup control wonderfully. The polishing robot can realize curve polishing trajectory tracking in 1s. This algorithm can be used to effectively control the trajectory of polishing robots,contributing to the stable and reliable robot operation during work and high processing efficiency.

关 键 词：木件打磨机器人 轨迹跟踪 模糊算法 SIMULINK仿真 

分 类 号：TS642[轻工技术与工程]