基于自组织模糊神经网络在脑机接口中的应用及其研究

发布时间：2018-03-29 17:38

本文选题：脑机接口　切入点：自组织　出处：《江南大学》2012年硕士论文

【摘要】：人们一直幻想利用自身大脑意念来直接控制外部世界，然而这个愿望一直没能实现。直到随着计算机和信息处理技术的飞速发展，科学家们终于将这个梦想变成现实，提出了脑机接口(Brain computer interface，BCI)技术。脑机接口技术是一种新的人机接口方式，它为大脑建立一条全新的体外信息交流与控制通道，通过对脑电信号的分析，理解人的意图，直接将大脑信息传达到外部世界。作为一种全新的信息交流技术，它可以为那些思维正常但肢体残疾的人提供一种新的交流方式，并在军事医疗等领域有很大的应用前景。脑电信号的分类识别是脑机接口技术的关键，提高其分类的准确率在科研中有很重要的意义。自组织模糊神经网络将模糊逻辑理论和人工神经网络这两种技术有机的结合起来，既拥有了模糊逻辑理论模仿人类推理的能力，又拥有了人工神经网络的自适应学习能力。而将其应用到脑机接口中，可以很好的预测基于运动想象的脑电信号，更好的让大脑与外部世界进行交流。本论文的工作重点是：首先，介绍了脑机接口技术的相关知识，如应用前景、目前的发展趋势。阐述了脑机接口技术的概念及其系统组成部分。提出了几种脑电信号的特征提取方法和模式识别算法，并用MATLAB编程实现了对BCI数据的特征提取。其次，介绍了几种常用的模糊神经网络的经典算法，包括模糊神经网络的基本组成，理论的本质和需要解决的问题(如识别率和抗噪性以及参数初始化等)，给出了算法的具体实现步骤，为将其引入脑机接口的应用中做理论基础。再次，在上面算法的基础上，提出了基于有效神经元的自组织模糊神经网络算法在脑机接口中的应用，，此算法能减少BCI样本训练时间，与其他BCI竞赛算法比较，具有较好的正确率。最后，又介绍了改进的基于三阶B样条函数的自组织模糊神经网络算法。讨论了其在脑机接口中的应用，并用MATLAB编程实现对BCI脑机接口数据集的分类识别任务。从实验中可看出，与其他的模糊神经网络算法比较，此算法的识别率还有待提高。
[Abstract]:People have been fantasizing about using their brains to control the outside world directly, but that didn't happen until, with the rapid development of computers and information processing technology, scientists finally turned that dream into reality. Brain-computer interface (Brain-computer interface) is a new method of man-machine interface. It establishes a new way for brain to communicate and control information in vitro. Through the analysis of EEG, we can understand people's intention. To communicate brain information directly to the outside world. As a completely new communication technology, it can provide a new way of communication for people with normal thinking but physical disabilities. The classification and recognition of EEG is the key of brain-computer interface technology. It is very important to improve the accuracy of classification in scientific research. The self-organizing fuzzy neural network combines the fuzzy logic theory with artificial neural network, and has the ability to imitate human reasoning by fuzzy logic theory. It also has the adaptive learning ability of artificial neural network. Applying it to the brain-computer interface, it can well predict the EEG signals based on motion imagination, and better let the brain communicate with the outside world. The main work of this thesis is:. First of all, the related knowledge of brain-computer interface technology, such as application prospect, is introduced. In this paper, the concept of BCI technology and its system components are introduced. Several feature extraction methods and pattern recognition algorithms for EEG signals are presented, and the feature extraction of BCI data is realized by MATLAB programming. Secondly, several classical algorithms of fuzzy neural network are introduced, including the basic composition of fuzzy neural network. The essence of the theory and the problems to be solved (such as recognition rate, noise resistance and parameter initialization etc.) are given. The implementation steps of the algorithm are given, which is the theoretical basis for introducing the algorithm into the application of brain-computer interface. Thirdly, on the basis of the above algorithm, a self-organizing fuzzy neural network algorithm based on effective neuron is proposed in the brain-computer interface. This algorithm can reduce the training time of BCI samples and compare with other BCI competition algorithms. It has good accuracy. Finally, an improved algorithm of self-organizing fuzzy neural network based on third-order B-spline function is introduced, and its application in brain-computer interface is discussed. The classification and recognition task of BCI brain-computer interface data set is realized by MATLAB programming. It can be seen from the experiment that the recognition rate of this algorithm needs to be improved compared with other fuzzy neural network algorithms.
【学位授予单位】：江南大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP334.7

【相似文献】