基于脑电的上肢动作识别方法的研究

发布时间：2018-06-07 14:37

本文选题：脑电 + 神经解码　；参考：《天津大学》2012年硕士论文

【摘要】：运动能力的恢复对肢体活动能力受损的人群至关重要,近年来虽然针对运动障碍患者的诊断和康复实验研究取得了巨大发展,但其总体疗效却并不令人满意。近些年来将脑机接口技术融入到康复治疗中的思想引起了越来越多的研究者的兴趣。与传统方法相比,基于脑机接口技术的康复器械可以根据患者的实际情况,通过神经解码产生的动作进行相关训练,使患者动作意图和肢体实际动作相互配合协调,达到最佳的训练效果。本文首先根据具体的实验需求设计并搭建了由脑电采集设备、运动采集设备和指示装置组成的实验系统。依据布置方便、实时准确的原则设计了以LM3S9B96为主控芯片、TFT显示屏作指令输出的指示装置,并编写了主控单元对显示屏的控制程序。数据处理阶段首先对采集得到的原始信号进行了预处理,预处理包括数据滤波和脑电数据的分割。对于脑电信号的特征提取采用了功率估计、AR模型参数和小波系数提取。利用提取出的脑电特征参数以运动方向为标签使用支持向量机进行了脑电数据的分类研究,对于三类运动的区分率最高为55.56%。由于支持向量机的参数对结果影响较大并难以规律性确定。本文随后使用了粒子群优化算法对支持向量机进行了参数优化,并讨论了优化算法的过程和结果。由优化结果可看出PSO算法大大减少了计算所需时间,同时在一定程度上降低了分类的成功率。
[Abstract]:The recovery of motor ability is very important to the people with impaired physical activity. Although great progress has been made in the diagnosis and rehabilitation of patients with motor disorders in recent years, the overall effect is not satisfactory. In recent years, the idea of integrating brain computer interface technology into rehabilitation therapy has attracted more and more researchers' interest. Compared with the traditional methods, rehabilitation instruments based on BCI technology can be trained according to the actual situation of the patients and the actions produced by the neural decoding, so that the intention of the patients and the actual movements of the limbs can be coordinated with each other. To achieve the best training results. This paper first designs and builds an experimental system composed of EEG acquisition equipment, motion acquisition equipment and indicator device according to the specific experimental requirements. According to the principle of convenient layout and real time accuracy, an indicator device with LM3S9B96 as the main control chip for instruction output is designed, and the control program of the main control unit to the display screen is compiled. In the data processing stage, the original signal is preprocessed, which includes data filtering and EEG data segmentation. The parameters of AR model and wavelet coefficients are extracted by power estimation. The classification of EEG data was carried out by using the extracted feature parameters of EEG using support vector machine (SVM). The highest discrimination rate of the three kinds of motion was 55.56%. Because the parameters of support vector machine (SVM) have a great influence on the result, it is difficult to determine regularly. In this paper, particle swarm optimization algorithm is used to optimize the parameters of support vector machine, and the process and result of the optimization algorithm are discussed. The optimization results show that the PSO algorithm greatly reduces the computation time and reduces the success rate of classification to a certain extent.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R318.0

【参考文献】