基于PLS的多维力传感器动态力反演研究  

DYNAMIC FORCE INVERSION OF MULTI-DIMENSIONAL FORCE SENSOR BASED ON PARTIAL LEAST SQUARES

作　　者：董建平 周兴林[1] 朱攀 Dong Jianping;Zhou Xinglin;Zhu Pan(College of Machinery and Automation,Wuhan University of Science and Technology,Wuhan 430081,Hubei,China)

机构地区：[1]武汉科技大学机械自动化学院,湖北武汉430081

出　　处：《计算机应用与软件》2022年第12期52-57,共6页Computer Applications and Software

基　　金：国家自然科学基金项目(51578430,51778509,51827812)。

摘　　要：多维力传感器动态力反演时测量信号容易受噪声影响,导致被测信号失真,测量精度不高。针对这一问题,将偏最小二乘法应用于传感器被测量求解,建立时域反卷积求解模型,通过提取主要成分来减轻噪声影响,将被测信号作为模型系数辨识求解。采用交叉验证估计均方差的方法确定提取的成分数量,从而得到模型系数,最后重建得到传感器的各维动态力。偏最小二乘法有效避免了矩阵求解的病态问题,具有更好的鲁棒性。仿真结果表明,在较高噪声干扰的情况下,通过该方法依然能有效准确地实现输入信号的反演。The measurement signal is easily affected by noise during the dynamic force inversion of the multi-dimensional force sensor, resulting in distortion of the measured signal and low measurement accuracy. To solve this problem, the partial least square method was applied to the sensor to be measured and solved. A time-domain deconvolution solution model was established, the main components were extracted to reduce the influence of noise, and the measured signal was used as the model coefficient identification solution. The method of cross-validation estimated mean square error was used to determine the number of extracted components, so as to obtain model coefficients, and reconstruct the dynamic force of each dimension of the sensor. Partial least square method effectively avoided the ill-conditioned problem of matrix solution and had better robustness. The simulation results show that, under the condition of high noise interference, the method can still effectively and accurately invert the input signal.

关 键 词：多维力传感器 反演 偏最小二乘法 成分提取 交叉验证 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术] TH823[机械工程—仪器科学与技术]