Decoding System Of EEG Information Based On Embedded

Brain computer interface(BCI)is an interdisciplinary research subject involving cognitive neuroscience,control theory,artificial intelligence and other disciplines.We transmitted the subjects' thinking information directly through EEG signals without relying on the peripheral nerves and muscles of the human body.It has great scientific research value in the fields of aiding the aged,aiding the disabled and military application.Among them,the accurate decoding of EEG information is the key to realize a reliable BCI system.In order to realize the direct brain control of robot,the paper used the embedded development platform to build a portable and practical EEG information decoding system.Aiming at the problems of limited resources and low classification accuracy in EEG information decoding system,we carried out the following research from two aspects of improving the accuracy of attention detection and control intention recognition.First of all,we improved the optimized complex network method(OCNM)to monitor the subjects' attention levels during the use of the system and we proposed a paradigm simplification and parameter optimization method suitable for embedded computing platform,which was implemented on the embedded platform based on stm32f407 single-chip microcomputer.The online experiment results of 6 subjects showed that the improved algorithm had a slight decrease in classification accuracy(80.67% to 77.78%)compared with the original algorithm,but the rate of single data classification was significantly improved(10.76 seconds to <1 second).Secondly,we choosed small computational complexity in the existing algorithm of Power Spectral Density Analysis algorithm(PSDA)and Canonical Correlation Analysis algorithm(CCA),we put forward a new method for decision fusion to improve the steady-state visual evoked potential(SSVEP)decoding accuracy.The offline test results of 6 subjects showed that the PSDA algorithm and CCA algorithm achieved average accuracy of 73.11% and 81.14% respectively,while the new method proposed in this paper achieved average accuracy of 83.68% through the fusion of classification decision of the two algorithms,thus improving the overall performance of the BCI system.Finally,we built an omni-directional human brain control system based on SSVEP signal.We installed LED lights with fixed frequency flashing on the robot platform,the subjects induced the SSVEP signal by watching the flashing lights at different frequencies.Then we outputted robot motion control instructions according to different signals.Considering the influence of robot distance on SSVEP signal strength,this paper studied SSVEP signal characteristics and classification accuracy at four kinds of distances.The results of the online experiment showed that the classification performance decreased with the increase of distance,and the classification accuracy decreased to 51.67% when the distance increased to 3 meters.These results provided a basis for the establishment of the application range of the omni-directional robot human brain control system in the subsequent research.