Prediction of Cognitive Errors Based on EEG Data Mining

Cognitive process is the premise for people to make relevant operations. Once cognitive errors occur, human behavior will have operational errors. Therefore, it is particularly important to find prediction methods for cognitive errors. Because the cognitive process is mainly completed by the human brain, and among many parameter indicators, EEG signal is the most direct measurement of brain activity related indicators, this paper uses EEG signal to study cognitive errors. This paper designed and carried out the pre-experiment and formal experiment of cognitive error prediction experiment, and collected the original EEG data when cognitive error occurred. After the original EEG data is obtained, the collected original EEG data are preprocessed. From the quantitative point of view, based on the one-way ANOVA method, the power spectral density of the relevant frequency bands and EEG acquisition channels is extracted as the characteristics of EEG data. Finally, using the power spectral density (PSD) values, the cognitive error prediction based on the cubic kernel support vector machine is carried out. By comparing the cognitive error coverage of models with different EEG data, we get appropriate data. The cognitive error coverage of the cognitive error prediction model with the best classification effect reaches 74%.