关键词: 幅值立方法, 神经网络, 特征向量, 偏差率

Abstract:

Feature vector obtained from traditional characteristic vector extraction algorithm cannot identify the different angle motion of the finger, this paper introduced a method of amplitude cubic to the characteristic vector extraction algorithm, and then decreased the orders of magnitude.The comparison with traditional feature vector identification algorithms suggests that the zero error rate in identification to the tiny different SEMG feature of the feature vector extraction algorithms proposed reaches more than 75%,meanwhile the huge error rate remained below 5%.And the finger movement pattern recognizer based on BP neural network was designed to enhance the correct rate of the mode of finger movement.

Key words: method of amplitude cubic, neural network, feature vector, error rate