摘要
本研究提出了一种表面肌电信号(SurfaceElectromyogram ,sEMG)至单纤维动作电位(SingleFiberActionPotential,SFAP)新的分解算法。由于sEMG分解的复杂性,本研究将sEMG分解问题转化为SFAP三基函数参数的优化问题和同一SFAP参数的聚类问题。在算法中,运用改进的遗传算法(GeneticAlgorithm ,GA)进行参数的优化,运用无监督学习的Kohonen神经网络进行参数的聚类。遗传算法的运用加强了算法的搜索能力,提高了分解的正确率,加快了算法的收敛速度。本分解算法的运用使得医疗诊断和假肢控制等领域可以通过非侵入式测量得到SFAP随时间的变化规律。
A new method was proposed in this paper of decomposing surface electromyogram (EMG) signals into their constituent single fiber action potentials (SFAPs). Because of the complexity of decomposition, the problem of sEMG decomposition was translated into three-base-function parameter optimizing and parameter clustering of the same SFAP. In the algorithm, improved genetic algorithm (GA) was used to optimize the parameter, and unsupervised learning Kohonen neural network was used to cluster the parameter. The using of GA enhanced the searching ability of algorithm, improved the decomposition correctness, and increased the decomposition convergent speed. The significance of such solution is that the variation of SFAP can be obtained by a non-invasive manner for physical diagnose and artificial limb control.
出处
《中国生物医学工程学报》
EI
CAS
CSCD
北大核心
2005年第3期343-349,共7页
Chinese Journal of Biomedical Engineering
基金
国家自然科学基金资助项目 (5 0 1770 2 3 )
关键词
SEMG
遗传算法
分解
SFAP
聚类
Artificial limbs
Biomedical engineering
Genetic algorithms
Neural networks
Optimization
