We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于改进 MPSO-SVM 算法的下肢连续运动预测模型.
- Authors
隋修武; 石峰
- Abstract
Aiming at the low prediction accuracy of lower limb continuous motion, an improved MPSO -SVM algorithm is proposed to predict the continuous motion of lower limbs. Firstly, the surface EMG signal, acceleration and knee joint angle information of the lower limbs when the human body is walking normally,and the signal amplitude range of the root mean square value, the integral myoelectric value and the acceleration of the EMG signal after denoising are extracted as feature samples. The principal component analysis method is used to perform eigenvalue fusion, and then the adaptive particle swarm optimization (MPSO) algorithm is improved by- introducing sine function and exponential function. The modified MPSO algorithm is used to optimize the penalty parameters and kernel function parameters of the support vector machine (SVM). Finally,the knee joint angle prediction based on the improved MPSO -SVM algorithm is constructed by using the EMG signal and the acceleration signal feature sample of the lower limb during walking. The model is trained and tested. The results show that the root mean square error of the knee joint based on the improved MPSO-SVM algorithm is 2.67°,the average error is 1.40°,and the maximum absolute error is 7.72°,which are far superior to the 21,27°,8.02°, and 58,38° predicted by the SVM algorithm, and 23.60°,13.59°, and 63.69° of BP neural networks.
- Subjects
STANDARD deviations; KNEE; LEG; SUPPORT vector machines; PARTICLE swarm optimization; SIGNAL denoising; DATA fusion (Statistics)
- Publication
Journal of Tiangong University, 2019, Vol 38, Issue 6, p69
- ISSN
1671-024X
- Publication type
Article
- DOI
10.3969/j.issn.1671-024x.2019.06.012