Cognitive Load Evaluation of Human-computer Interface Based on EEG Multi-dimensional Feature

To evaluate accurately the cognitive load (CL) of the operator under the digital interactive interface is helpful to guide the optimization of the digital interface and finally improve the ergonomics. In order to further improve the stability and accuracy of cognitive load evaluation method, combined with EEG experiment, deep learning is applied to CL evaluation problem. Firstly, the preprocessed EEG signals are directly input into CNN-LSTM network to extract the timedomain features of EEG. Secondly, the frequency-domain features of EEG are extracted by FFT and deep belief network (DBN). Thirdly, the time-frequency feature of EEG is obtained by Morlet wavelet transform and the multi-CNN. Finally, the cognitive load of interactive interface is classified by support vector machine (SVM). By recruiting 16 subjects, EEG data under two CL conditions were collected for experiments. The experimental results show that compared with other single deep learning algorithms, it can extract EEG time-domain features, frequency-domain features and time-frequency-domain features more accurately, so has stronger robustness.