残差补偿粒子滤波在六维力传感器下E膜中的应用

APPLYING RCPF TO LOWER E-TYPE MEMBRANE OF SIX-AXIS FORCE SENSOR

为减小动载环境下,噪声信号对六维力传感器测量精度的影响,提出一种具有残差补偿步骤的改进粒子滤波算法(RCPF)。该算法以双E型弹性体六维力传感器下E型膜为研究对象,根据挠度和应变的关系,建立非线性系统模型。根据权值蜕化程度,将样本集分层;借鉴残差补偿思想,在粒子重采样前修复先验分布的累积误差,将最新的观测信息融入样本集;基于Thompson-Taylor算法,通过聚合重采样将高、低权值粒子随机组合,产生新粒子集。将优化算法应用于六维力传感器动态测试系统。结果表明,RCPF算法具有更好的估计精度,在保持实时性的同时,有效地提高了六维力传感器的测量精度。

In order to reduce the influence of noise signal on the measurement accuracy of six-axis force sensor in dynamic load environment, we propose an improved particle filtering algorithm which has the residuals compensation steps （RCPF）. This algorithm takes the lower E-type membrane of dual-E elastic body six-axis force sensor as the research object, and builds the nonlinear system model based on the relation of deflection and strain. According to the degeneration level of weights, it stratifies the sample sets. Learning from residuals compensation idea, the cumulative error of prior distribution is repaired before resampling the particles and the latest observation information is merged into sample sets as well. Finally, based on Thompson-Taylor algorithm, the RCPF produces new particles set by the random combination of the particles with high and low weights through aggregation resampling. We apply the optimised algorithm to the dynamic test system of six-axis force sensor. Results indicate that it effectively enhances the measurement accuracy of six-axis force sensor while maintaining the real-time performance.