Research On Brain-computer Interface Based Self-organizing Fuzzy Neural Network

People have been fantasizing to use their own ideas to directly control outside world,however, this desire has not been achieved. Until computer and information processingtechnology have rapid development, scientists finally turn this dream into a reality andpropose brain computer interface (Brain computer interface, BCI) technology.BCI technologyis a new approach of man-machine interface.It establishs a new channel of communicationand control of information outside the body for the brain.And through the analysis of the EEGsignal, then understand people’s intention and directly transmit brain information to theoutside world.As a new information and communication technology, it can provide a new wayof communication for those people whose thought is normal but body is disabled.It has a wideapplication prospects in military、medical and other fields.EEG signal classification is a keyof BCI technology,so it has a very vital significance that the classification accuracy hasimproved in scientific research. The self-organized fuzzy neural network(SOFNN) combinestwo technologies that are fuzzy logic theory and artificial neural network. It not only has thecapability to imitate human reasoning of the fuzzy logic theory, but also has adaptive learningcapability of artificial neural network. Applying SOFNN to BCI,EEG signals based onmovement of imagination can be predicted very well and the brain have better communicationwith outside world. The work of this paper focuses on as follows：Firstly, introduce related knowledge about BCI technology, such as application prospects,current development trends.Elaborate concepts and system components of BCI technology.Propose several algorithms of EEG feature extraction and pattern recognition classificationand achievet algorithm of feature extraction using MATLAB.Secondly, introduce contents of some classic algorithms of fuzzy neural network,including the basic components,the nature of theories and the problem that need to beresolved(such as the recognition rate and resistance to noise and parameter initialization andso on).And give concrete steps of the algorithms.These algorithms provide theoretical basisthat we introduce them into BCI application.Thirdly,propose application of self-organizing fuzzy neural network algorithm based oneffective neurons in BCI. And this algorithm can reduce training time of BCIsamples.Compared with other BCI competition algorithms, it has good accuracyFinally,introduce improved self-organizing fuzzy neural network algorithm based onthree order B-spline function.Discusses its application in BCI and achieve classification andrecognition task about BCI data set with using MATLAB.Through experiments we can seethat the recognition rate of the algorithm needs to improve compared with other fuzzy neuralnetwork algorithms.