Gait Phase Recognition Based on Improved Multi-Objective Sine-Cosine Algorithm

Gait phase recognition has been shown to facilitate the analysis of gait feature, evaluation of ankle rehabilitation, and assessment of physical ability. However, the efficiency of gait phase recognition is generally limited by the shortcoming of single sensor and data acquisition method. The application of multi-sensor fusion improves the recognition accuracy, whereas it increases data redundancy and reduces system real-time performance. This paper proposes an improved multi-objective sine-cosine algorithm for gait phase recognition which aims to identify the most effective feature combinations to achieve optimal recognition accuracy. We explore the effect of different sensor positions on gait phase recognition, and introduce integrated classifiers and multi-sensor to improve the performance. Experimental results demonstrate that the proposed method effectively increases the convergence speed, avoids local optimal solutions, reduces data dimensionality, and recognizes gait phases. The proposed method has the potential to provide effective support for personalized control in gait rehabilitation.