Research On Intelligent Wheelchair Control Based On EEG

Brain-computer interface(BCI)is a direct way of communicating between the brain and a computer or other electronic device without the involvement of peripheral nerve tissue.How to improve the identification accuracy of BCI equipment and the stability of EEG signal is an important research topic.To solve the above problem,this paper through the imagination of motion of brain electrical signal preprocessing,feature extraction and classification recognition algorithm research,improve the identification precision of the BCI on EEG signals,signal-to-noise ratio and stability,for more information,after the EEG signals processing into a control instruction of external equipment,in order to solve the mutilation disabled people to control the wheelchair for autonomous mobile through the brain.Finally,an intelligent wheelchair motion control experiment is designed to verify the proposed algorithm.The specific work of this paper is as follows:Preprocessing of EEG signals.Wavelet threshold de-noising processing method for EEG,by finding a suitable threshold as a door way,eliminate noise keep useful information,and further used forced de-noising,the default threshold de-noising and the given threshold de-noising three wavelet threshold de-noising methods,on brain electrical signal de-noising comparison analysis,Finally,the wavelet threshold is selected as the signal pretreatment method.Feature extraction of motion imaging signals.In this paper,DWT,CSP,AR model and PSD are used to extract the features of EEG signals.PSD takes the total energy of each group of signals as the feature vector,which contains little feature information and overlaps.DWT extracted the frequency band containing ERS/ERD phenomenon of motion imagination and used the energy of the frequency band as the feature vector.The eeg information represented by DWT was more obvious than that proposed by PSD.CSP is suitable for feature extraction of EEG signals with few classification tasks.The AR model is used to extract the features of EEG,and the extracted feature information is closer to the original signal.Finally,the classifier constructed in chapter 4 is used to classify and identify the feature vectors extracted by the four feature extraction algorithms,and the comparison results show the superiority of DWT.Classification and recognition of motion imaging signals.Support vector machine(SVM)is the traditional motion image signal classifier.Bagging and Boosting are two kinds of ensemble learning algorithms,and Boosting algorithms include Ada Boost and Gradient Boosting.Three classifiers Bagging,Ada Boost and Gradient Boosting are constructed based on Bagging,AdaBoost and Gradient Boosting algorithms,and the feature vectors are classified and recognized.Among them,the AdaBoost algorithm uses exponential loss,and the Gradient Boosting algorithm can use any loss function.Finally,the simulation results show that the performance of Gradient Boosting classifier is better than that of Bagging,Ada Boost and support vector machine.The collection experiment of motion imaginary signals is designed in the actual laboratory environment,and the collected EEG signals are preprocessed,feature extracted and classified,and the DWT-Gradient Boosting combination is verified to be superior to other combinations by simulation.Finally,the recognized signals are converted into external instructions to achieve the motion control of the wheelchair.The experimental results verify the effectiveness of the proposed EEG signal feature extraction algorithm and classification recognition algorithm.