Research On BCI Based On Multi-class Motor Imagery

Brain science and intelligent tech similar to brain currently are at the edge of science and technology,and they are of great significance to human health,artificial intelligence technology and new industry.The BCI is a control system that doesn't rely on peripheral nerve of brain and normal output channel of muscle,which can connect directly to brain and external equipment,bringing the hope for patients with neuromuscular channel failure like amyotrophic,myasthenia gravis and stroke.Although motor imagery EEG signal has gotten a lot of attention,there are many problems to be solved.For example,the research on motor imagery of left and right hand has gained very good recognition accuracy,but the two generated control commands still can't meet the large demand for application.Therefore,this paper conducts research on BCI tech of 8 class of motor imagery.On the basis of deep research of relative theories about BCI tech of multi-motor imagery,firstly conducted the secondary development of MindWave headset devices,collected and stored the EEG signals;Secondly FIR digital wave filter based on hamming window was used to filter wave for removal of artifacts and power-line interference;In three methods of nonlinear dynamics analysis,according to the feature extraction effect choosed approximate entropy as the feature extraction algorithm,and used wavelet packet energy entropy extract the multi-class motor imagery EEG.Respectively recognized the pattern of the feature vector by using generalized regression neural network,probabilistic neural network and constrained extreme learning machine.Then chose the extreme learning machine algorithm,its recognition accuracy is up to 87.5%;Finally developed brain-machine interface control software based on Android by combining MindWave equipment.Using the software,unmanned aerial vehicle has achieved the eight kinds of action like take-off,landing unmanned and rolling etc.At last design a scheme of EEG signal controlling unmanned aerial vehicle and verifies this whole system by controlling the toy unmanned aerial vehicle.At the end of this paper are a summary of the main content,innovation point,insufficient point of this paper and a forecast for the future of BCI tech.