Physiological Stimulation Eeg Signal Analysis Based On Deep Learning Algorithm

Using EEG signal data characteristics to analyze,dig deep-level pathological information,and intervene in the operation process of the nervous system is the main research direction of brain science at present.Based on the immersive virtual reality physiological stimulation to induce the intervention mental state,and then analyze the generated EEG signals,the evaluation of the human mental state is one of the best medical intervention diagnosis and treatment programs.Therefore,this project aims at the research on the mental state classification algorithm reflected by the induced physiological stimulation EEG signal,and explores the application of mental state intervention based on immersive VR.This paper analyzes the EEG signals generated under physiological stimuli based on deep learning algorithms.The main research results are as follows:(1)Propose a feature detection model of fusion multi-layer EEG signal eigenmode function based on depth separable convolution.First,use empirical mode decomposition to perform deep feature mining to obtain the eigenmode function characteristics of the multi-layer EEG signal,and then integrate the eigenmode function feature input of the EEG signal under the multi-layer immersive VR physiological stimulation channel by channel and point by point.The convolutional network recognizes the state of psychological fear.This paper collects the EEG signals of patients with phobias under immersive VR stimulation for a week,uses the network model to analyze EEG signals,and counts the changes in desensitization index of immersive VR stimulation exposure therapy.It is found that the desensitization index has a relationship with time.Logarithmic relationship.Therefore,after about 12 days of VR exposure treatment,the desensitization index can reach95%.(2)Proposes a time-frequency signal feature detection model of EEG signal under physiological stimulation based on double network cascade.The model is composed of two parts: EEG time-frequency signal feature extraction algorithm and cascade network.First,two different traditional algorithms are used to perform feature mining and normalization of EEG data,and then the cascade network is used to mine EEG data time-frequency.The in-depth characteristics of information,in which the convolutional neural network mainly performs local feature mining and dimensionality reduction on the input data,and the cascaded recurrent neural network mainly mines the time series information characteristics of the input data.The model built in this article adopts time and space learning strategies.After experiments,the accuracy of this model in predicting the mental state reaches 91.48%.In summary,the EEG signal analysis algorithm proposed in this article can better meet the needs of clinical intervention diagnosis and treatment systems,and provide a relevant research foundation for the subsequent application of virtual reality technology to psychological intervention practice.