作者: B. Hudgins , P. Parker , R.N. Scott
DOI: 10.1109/10.204774
关键词:
摘要: A novel approach to the control of a multifunction prosthesis based on classification myoelectric patterns is described. It shown that signal exhibits deterministic structure during initial phase muscle contraction. Features are extracted from several time segments preserve pattern structure. These features then classified using an artificial neural network. The signals derived natural contraction which can be produced reliably with little subject training. new scheme increases number functions controlled by single channel but does so in way not increase effort required amputee. Results presented support this approach. >