Handle reaction vector analysis with fuzzy clustering and support vector machine during FES-assisted walking rehabilitation

Abstract

This paper proposed Fuzzy clustering of C means and K means methods to extract the lateral features of lower limbs movement from handle reaction vector (HRV )data. With C-means clustering, the SVM recognition rate of lateral features was usually above 90% while, with K-means clustering, the recognition rate was close to 85%. The best recognition rate was even reaching up to 97% for some individual subject. Then the samples from all subjects were processed together with the cross-validation. Our experimental results showed that the HRV signal could be used with fuzzy clustering and support vector machine to effectively classify the lateral features of lower limbs movement. It may provide a new choice for FES control signal. The optimizing of the algorism parameters can be introduced to get better control in the future. © 2011 Springer-Verlag.