Emotion Recognition And Regulation System For Children With Autism Based On EEG Signals

Nowadays,brain-computer interface and machine learning technologies are developing rapidly and research on the recognition of emotions from EEG signals is increasing,but most of the research has focused on new algorithms to improve classification accuracy.Most of the research has been conducted on normal subjects,and most of their EEG data has been obtained directly from national and international datasets,such as the DEAP dataset from Queen Mary University of London and the SEED dataset from Shanghai Jiaotong University.There are very few studies on EEG signal emotion recognition for special groups such as autistic children.Therefore,in order to verify the feasibility of using EEG signals to classify and identify emotions of autistic children,this paper investigates the EEG signals of autistic children for emotion recognition.And based on this research,an emotion recognition and regulation system for autistic children is designed to regulate the emotions of autistic children,improve their bad emotional behaviour and protect the physical and mental health of the children.The main work of this paper is as follows:(1)In response to the current problem of insufficient EEG signal data of autistic children,this paper conducted a cooperative study with the Zhoucun District Special Education School in Zibo City to conduct EEG signal data collection experiments on autistic children,and finally succeeded in collecting EEG signal data of five autistic children and producing the materials needed to induce emotions in autistic children in the EEG signal collection experiments.Three professional music therapy teachers first classified the emotion-evoking materials,and then scored and classified the emotions of autistic children in the experiment,and finally classified the results into positive,negative,and neutral types.(2)Emotion recognition study of EEG signals for children with autism.Firstly,the collected EEG data were pre-processed,and noise reduction was performed by using filters and wavelet packet decomposition reconstruction,then the emotional features of EEG signals were extracted,and the power spectral density in the frequency domain analysis method and the differential entropy in the nonlinear analysis method of EEG signals were extracted for comparison and analysis,and finally the extracted features were classified and identified.Support vector machine,random forest,Adaboost,LSTM(Long Short Term Memory)and Res Net(Residual Neural Network)were used to classify and identify the data,and the accuracy of the best classification result reached 90%.(3)In order to further regulate and intervene in the emotions of autistic children,an emotion recognition and regulation system for autistic children was designed at the end of this paper.The system can perform a series of analysis and processing of EEG data,and finally display the EEG signal of the autistic child and the kind of emotion it belongs to in the interface.According to the current emotional state of the child,the system can play the corresponding music therapy material to regulate the emotion of the autistic child according to the principle of music therapy for autistic children.