Smart Home Control System Based On Motion Imagination Of EEG

Brain Computer Interface(BCI)is a system that enables communication between the human brain and a computer or other device,and it provides a completely new way for information exchange between the human brain and the outside world.The brain-computer interface was originally designed to help people with severe movement disorders to control external devices.Based on this,a smart home control system based on motion imagination is designed to help disabled people control external home equipment through their imagination of EEG(Electroencephalogram).The work of the thesis is mainly reflected in the following aspects:(1)In the pre-processing phase,based on the Independent Component Analysis(ICA)method,an Independent Vector Analysis(IVA)solution based on the fastest step size reduction is proposed,and the artificially mixed data and actual EEG data are collected.Two simulation experiments demonstrate the separation effect and convergence speed of the proposed algorithm.The experiments also show that the signal-to-interference ratio of the proposed algorithm is better than the ICA and naturally decreasing IVA methods;the subdiagonal value of the correlation matrix is also maintained near 0.85,indicating that the separation effect is obvious;In addition,the fastest convergence rate obtained from real EEG data.(2)In the feature extraction stage,in order to overcome the shortcomings of unstable feature extraction of small samples,a regularization method can be introduced in Common Spatial Pattern(CSP),and training data is weighted and then merged,which effectively avoids training samples.Unity.Simplifying the sample can reduce the individual differences of the data and improve the stability of the system.(3)At the stage of classification and identification,the main features of a linear classifier include a small amount of calculation,a simple model,and superior performance for binary classification problems.However,for the current trend of more brain-computer interfaces applied to online experiments,linear classifiers still have two important issues to be resolved.First: In the experimental stage,only a small sample size can be collected within a short time,which does not achieve the best results of machine learning.In order to expand the sample size,only the sample data previously collected by the subject can be added to the sample data collected now for common analysis.Secondly,feature extraction and pattern classification models of EEG signals in real-time operating brain-computer interface systems are obtained through offline training.In order to solve the two problems of linear classifiers,this paper optimizes the Fisher linear classifier,that is,an algorithm based on adaptive perceptron linear classifier(adaption perceptron LDA,ap LDA)is used to apply it to online classification and provide a system Accuracy of online classification.(4)Finally,the thesis designed a smart home control system based on motor imagination.Subjects can control the home equipment by using the Emotiv Insight EEG device for motor imagination.And through experimental analysis,an accuracy of 90.25% was obtained.The experimental results show the feasibility,accuracy and practicability of the smart home system based on motion imagination proposed in this thesis.