工装条件对多维力传感器动态特性的影响机理及其动态校正的研究

Multi-axis force sensors are widely applied in scientific experiments and advanced manufacturing, and their dynamic performance directly affects the developments of advanced technologies. However, the dynamic characteristics of almost all multi-axis force sensors (including new technologies) are affected by their tooling (including support structure and end load), and poor tooling matching will further increase the dynamic measurement error. Whereas, systematic research on this problem has not been seen so far, and actual tooling conditions are also rarely considered in the studies of dynamic error correction. Thus, this project mainly studies the problems of “influence mechanism of tooling condition on the sensor dynamic characteristic” and “dynamic characteristic evaluation and error correction of the multi-axis force sensor under working conditions”. First, through simulation and experimental researches, establish the sensor variation law model of normalized dynamic characteristic parameters under working condition to reveal the common characteristics of the tooling impact; second, study on-site dynamic calibration and modeling methods of multi-axis force sensor, to provide direct model basis; third, for the typical application of wrist force measurement of industrial robots, study the dynamic correction method of multi-axis force sensor under the condition of “moving support + fixed/variable load”; finally, ground test of satellite capture is taken as an example for application verification. Accordingly, the studies of this project are expected to provide basis for the sensor selection, tooling design, performance evaluation and dynamic correction in applications, and thus have extensive significance.

多维力传感器在科学试验和先进制造领域被广泛应用，其动态性能直接影响着先进技术的发展。然而，几乎所有多维力传感器(包括新技术)的动态特性都会受其工装(支撑机构与末端负载)的影响，工装匹配不好时会进一步增大动态测量误差。但目前未见此问题的系统研究，动态误差的校正研究也多未考虑实际工装情况。据此，本项目研究“工装对多维力传感器动态特性的影响机理”以及“工况下多维力传感器的动态特性评估与动态校正”问题。先通过仿真和实验研究，建立工况下传感器的归一化动态特征参数变化规律模型，揭示工装影响的共性特征；再研究工况下多维力传感器的现场动态校准与建模问题，以提供直接模型依据；然后，针对工业机器人腕力测量的典型应用，研究“动支撑+定/变负载”工况下的多维力传感器动态校正方法；最后，以卫星捕获地面模拟试验为例来进行应用验证。本研究预期可为应用中传感器的选型、工装设计、性能评价、动态校正提供依据，具有广泛的意义。

多维力传感器广泛应用于风洞试验、机器人等军民领域的多种应用场合。然而，多维力传感器动态特性通常较差，跟随响应时间长、超调量大、测量带宽低、存在维间动态耦合，使其动态测量误差较大，且传感器的工装对其动态特性也会产生明显影响。这使得多维力传感器在不做动态误差校正的情况下难以直接用于动态力的测量。. 本项目研究旨在为实际工况下多维力传感器的动态误差校正提供软解决方案，降低校正系统复杂度，提高其动态测量精度。项目研究内容包括工装条件对多维力传感器动态特性影响机理及其动态特征参数模型、工况下多维力传感器的现场动态校准与动态建模、不同工况下工业机器人腕力测量的动态校正方法、卫星捕获地面模拟试验中卫星模型动态接触力/力矩测量的动态校正。. 项目研究结果揭示了工装结构参数对多维力传感器动态特性影响的共性规律并建立了BP神经网络模型；提出了混合冲击/阶跃激励的载荷设计方法、输入载荷辨识方法、校准实验方法和频响估计方法，解决了工况下传感器的动态校准与频域建模问题；提出了校正传感器自身特性的基于预矩阵的迭代动态解耦-补偿方法、传感器自身特性校正与支撑端惯性补偿相组合的动支撑动态校正方法、基于共性规律模型与参数自适应二阶补偿器的变负载动态校正方法；针对ATI Mini45传感器的实验结果表明，所提方法能够将不同工况下传感器各通道动态误差降低70%~95%以上。从而，项目研究为应用中多维力传感器动态性能预测、传感器选型及工装设计提供了参考依据，为不同工况下多维力传感器动态误差校正提供了解决方法。项目最终通过卫星捕获地面模拟试验中模型动态接触力/力矩测量的动态误差校正进行了应用验证；结果表明所提方法能够从多维力传感器测量输出中准确恢复模型所受的外部接触力/力矩信号，试验中误差分别为1%和6%，具有实用价值。. 项目研究成果具有重要的理论意义和实用价值，可推广应用于机器人、风洞试验及其它应用中的动态力测量场合。