A Peak and Trough Detection Algorithm Based on Sliding Window Variance

Abstract

AbstractThe application object of the algorithm proposed in this paper is a compound hip exoskeleton, which uses the control strategy based on linkage state variables to execute walking assistance. In the control process, it is necessary to detect and identify the peaks and troughs of linkage state variables under regular gait, and eliminate the peaks and troughs under irregular gait in the process of movement. The existing peak and trough detection algorithms are usually used in the field of medical and image processing for electrocardiogram QRS peak recognition and edge detecting. These algorithms focus on the accuracy and robustness of peak and trough recognition in the signal containing high-frequency noise. The signal processing process is mostly off-line mode, which can not meet the special peak and trough recognition and real-time requirements in the compound hip exoskeleton control strategy. In this paper, a real-time peak and trough recognition algorithm based on sliding window variance (SWV-PTD) is proposed. By setting an appropriate variance threshold, the algorithm can effectively detect the peaks and troughs of linkage state variables only in regular gait, eliminate the peaks and troughs of linkage state variables in irregular gait, and the delay interval of peak and trough detection is within an acceptable range. Finally, the reliability and effectiveness of the algorithm are verified by experiments.KeywordsCompound hip exoskeletonWalking assistanceSliding windowVariancePeak trough detectionLinkage state variable