Brain-computer Interface Signal Processing Method And Control Application Research

Brain Computer Interface(BCI)extracts relevant information by translating and decoding brainwave,but not rely on normal brain transmission pathway.Now it has developed into a new way of communication.What's more,it makes a big difference in medical,military,robotics research fields and attracts worldwide attention.Among the brain signal acquisition modes,the event-related desynchronization / synchronization(ERD/ERS)EEG pattern links with movement imagination.Because of obvious physiological characteristics and good discrimination,this pattern has become a hot topic in brain machine interface field.With a large number of BCI data accumulation and based on the natural principle of ERD/ERS pattern,this paper mainly studies EEG data processing methods,including signal processing,feature extraction and pattern classification algorithm,and applies the magnetic bacteria optimization algorithm(MBOA)to optimize support vector machine(SVM)classifier.The processing methods are validated by two categories and four categories offline movement imaginary datasets,and finally establishes a portable brain-computer interface control system.In this paper,there are the definition,main content,research status and future development of BCI.And it also describes EEG Signal Processing preprocessing,feature extraction and pattern classification methods in detail.For pattern recognition classifier SVM,parameters optimization is an important problem.This paper applies MBOA to achieve it,at the same time,compared with other intelligent optimization algorithms such as Particle Swarm Algorithms(PSO),Artificial Bee Colony algorithms(ABC),Genetic Algorithm(GA),Biogeography-Based Optimization Algorithm(BBO).The comparison experiments are verified through the international UCI data sets.For motor imagery EEG datasets of two categories,we use a variety of combinations of feature extraction and classification methods to make a comparison.For the motor imagery EEG datasets of four categories,Common Space Pattern-Support Vector Machine(CSP-SVM)method is outstanding.In order to obtain higher classification accuracy,here we add MBOA-SVM parameter method on EEG data processing optimization problem.Finally,it designs a control application experiment using portable BCI device.The BCI system consists of a robot,a computer and the MindWave.The system complete system communication,use attention degree and eye blink captured by EEG device to control speed and veer of external devices.It makes "Mind Control " come ture.