针对微机电测量系统（MEMS）波峰检测计步算法和自相关分析计步算法仅利用单轴加速度和固定阈值对传感器姿态和运动状态变化适应性较差的问题，提出了一种自适应波峰检测算法。该算法将行人运动状态分为正常状态与非正常状态，根据行人每一步的最大整体加速度与运动状态的内在相关性，获取不同运动状态的波峰检测经验阈值，实现不同运动状态下的自适应计步。通过实验对比分析，自适应波峰检测算法在传感器不同姿态和行人不同运动状态下的计步正确率均可达到99%以上，而常规波峰检测算法和自相关分析算法对正常态的计步精度虽然达到97%和99%以上，但对非正常状态下的计步精度仅有70%和50%，无法适应行人运动状态的变化。结果表明：自适应波峰检测算法对MEMS传感器姿态和运动状态的变化适应性较强，能够实现传感器不同姿态和不同运动状态下的可靠性计步。另外，自适应波峰检测、常规波峰检测、自相关分析算法的时间运算效率分别为0.036 s、0.046 s、0.131 s，自适应波峰检测算法时间效率明显优于其他两种算法。%In view that conventional peak detection algorithm and self-correlation analysis algorithm have poor adaptability to sensor attitudes and motion states of MEMS measurement system pedometer using single axis data and fixed threshold, an adaptive peak detection algorithm is proposed. According to inherent correlation of maximum acceleration and motion states, the algorithm gets peak detection experi- ence thresholds of different motion states, and realizes adaptive step counting. Experiments show that the adaptive peak detection step counting accuracy reaches above 99% for both different sensor attitudes and pedestrians motion states. In contrast, the accuracies of conventional peak detection and self-correlation analysis algorithms reach 97% and 99% under normal state, but only 70% and 50% under abnormal state. The results show that the adaptive peak detection algorithm has strong adaptability to sensor attitudes and motion states, and achieves reliable step counting under various conditions of sensor attitudes and motion states. In addition, the time calculation efficiencies of the adaptive, conventional peak detection and self-correlation analysis algorithms are 0.036 s, 0.046 s and 0.131 s, respectively, which prove that the adaptive peak detection algorithm is significantly superior to the other two algorithms.
展开▼
