摘要
针对医用三维力传感器容易受电磁场等外部环境的影响,产生大量相似特征数据,导致其输出紊乱信号,降低传感器控制精度和测量速度的问题,提出一种多级冗余强干扰下三维力传感器数据挖掘方法。根据角度标定理论采集三维力传感器冗余数据;引入相似度指数函数计算冗余因子,获取三维力传感器冗余数据活跃度,完成数据冗余分类;通过差值去噪算法高性能过滤三维力传感器冗余数据;利用谱聚类算法构建拉普拉斯矩阵,剔除冗余数据,实现三维力传感器数据自动挖掘。仿真结果表明,所提方法在多级冗余强干扰下的三维力传感器控制精度为96.54%,测量速度为0.61 ms,能量消耗为0.26 kcal。由此证明,所提方法的控制精度高、测量速度快、传输效果优,能够满足机器人辅助手术过程中的力反馈控制需求。
Targeting at the problem that medical 3D force sensor is easily affected by external environment such as electromagnetic field,which produces a large number of similar characteristic data,resulting in the output of disordered signals and reducing the control accuracy and measurement speed of the sensor,a data mining method of 3D force sensor under multi-level redundancy and strong inter-ference is proposed.Redundant data of the three-dimensional force sensor are acquired according to the angle calibration theory.The similarity index function is introduced to calculate the redundancy factor,and the activity of redundant data of three-dimensional force sensor is obtained to complete the data redundancy classification.High performance of filtering redundant data of three-dimensional force sensor is realized by using difference denoising algorithm.Spectral clustering algorithm is used to construct Laplacian matrix,elim-inating redundant data,and realizing automatic data mining of three-dimensional force sensor.The simulation results show that the con-trol accuracy of the proposed method is 96.54%,the measurement speed is 0.61 ms,and the energy consumption is 0.26 kcal.It is proved that the proposed method has high control precision,fast measurement speed and excellent transmission effect,and can meet the demand of force feedback control in robot-assisted surgery.
作者
岳根霞
王剑
刘金花
YUE Genxia;WANG Jian;LIU Jinhua(Fenyang College,Shanxi Medical University,Fenyang Shanxi 032200,China)
出处
《传感技术学报》
CAS
CSCD
北大核心
2024年第8期1383-1388,共6页
Chinese Journal of Sensors and Actuators
基金
山西省教育厅项目(J2021972)
山西省高等学校教学改革创新项目(J2020437)。
关键词
三维力传感器
冗余数据
数据挖掘
角度标定
指数函数
差值去噪
谱聚类算法
three-dimensional force sensor
redundant data
data mining
angle calibration
exponential function
difference denoising
spectral clustering algorithm
