基于卡尔曼滤波模型的平地行走步频预测方法

摘要

The adjusted damping of intelligent lower limb can only take effect in the next swing cycle because the control is based on the stride frequency of the prior stride. Therefore, this control tactics is ineffective, it is unable to achieve the real⁃time control and is difficult to be applied when the stride frequency changes frequently. A stride frequency prediction method based on the plantar pressure sensing and Kalman prediction model was proposed in this paper to solve the problem. The plantar pressure sensing obtains data of stride frequency, then the Kalman pre⁃diction model predicts the next stride frequency based on the stride frequency that has been known. In the level walking experiment, the deviation between the next step stride frequency predicted by this method and the posterior values reduced about 10% compared with the deviation of the following method. It has good real⁃time performance and provides a feasible plan for the performance improvement of the intelligent lower limb.%现有智能下肢的控制策略都是以刚完成的一步的步频为调节阻尼的依据，调整好的阻尼只能在下一步摆动期生效，因此该控制策略是滞后的，无法做到实时控制，在步频变化频繁的场合难以适用。本文方法利用足底压力传感获取步频数据，然后通过卡尔曼预测方程由已完成的步频预测即将迈出的下一步步频。在模拟日常生活平地行走步频变化的实验中，所预测的下一步步频与后验值之间偏差比跟随方法的偏差大约减小了10％。该方法实时性好，为改善智能下肢的性能提供了新的可行性方案。