摘要
耦合现象普遍存在于多维力传感器中,多维力传感器输出如不经过解耦,数据直接应用到机器人操作中,会导致机器人的误操作。针对存在的耦合现象,本文首先分析了多维力存在的耦合原因,根据产生原因将耦合分为两种形式:结构耦合和误差耦合,然后提出了一种新的解耦方法-基于线性神经网络解耦方法,与传统解耦方法相比,该方法大大提高了解耦合精度。最后通过实验验证了该方法的有效性和优越性。
Coupling is a common phenomenon in the multi-axis force sensor.If we measure the data without decoupling,it will cause incorrect operation of the robots.Aiming at coupling phenomenon,this paper analyzes the reasons why coupling exists in the multi-axis force sensor.Dividing from the causes there are two types of multi-axis force sensors,structure coupling and error coupling.We put forward a linear decoupling method based on neural network.Comparing to the traditional decoupling method,this method improves the precision of decoupling greatly.In the end of this paper,compared with the traditional decoupling method from expertments,this method was proved its effectiveness.
出处
《传感技术学报》
CAS
CSCD
北大核心
2011年第8期1136-1140,共5页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金重大国际合作研究项目"仿土拨鼠矿难探测与救援机器人基础理论与关键技术研究"(60910005)
关键词
多维力传感器
解耦
神经网络
结构耦合
误差耦合
multi-axis force sensor
decoupling
neural network
structure coupling
error foupling