论文部分内容阅读
体表肌电信号会随着外部或人体内部环境变化而发生改变,这种时变特征使得固定参数肌电模式分类器的分类精度会随着时间的延长而下降。为了获得具有稳定性能的肌电假肢控制系统,在肌电模式分类器中加入自适应机制是很有必要的。本文以传统线性判别分析(Linear Discriminant Analysis,LDA)为基础,尝试在肌电模式分类器中引入三种自适应方案,并探讨了这三种方案在肌电模式分类应用中的优缺点。初步研究表明:自增强线性判别分析(Self-enhancing LDA,SELDA)分类器和循环训练集线性判别(Cycle Substitution LDA,CSLDA)分类器都能够将识别准确率提升5%左右。其中,SELDA是一种有效的自适应方案,而CSLDA可以得到更高的识别率提升和更好的稳定性,但是计算量较大,需要更大的代价。卡尔曼自适应线性判别(Kalman Adaptive LDA,KALDA)分类器单独使用效果不明显,需要进一步改进或结合其他方法使用。
Surface EMG signal will change with the external or human internal environment changes. This time-varying feature makes the classification accuracy of EMG classifier with fixed parameters will decrease with time. In order to obtain a stable electromyographic prosthesis control system, it is necessary to add an adaptive mechanism to the EMG classifier. Based on Linear Discriminant Analysis (LDA), this paper attempts to introduce three kinds of adaptive schemes in EMG classifier, and discusses the advantages and disadvantages of these three schemes in the classification of EMG. Preliminary studies have shown that the Self-enhancing LDA (SELDA) classifier and the Cycle Substitution LDA (CSLDA) classifier can both improve the recognition accuracy by about 5%. Among them, SELDA is an effective adaptive scheme, while CSLDA can achieve higher recognition rate and better stability, but it requires more computational cost. Kalman adaptive linear discriminant (Kalman Adaptive LDA, KALDA) classifier alone is not effective, need to be further improved or combined with other methods.