فهرست نشریات علمی دسترسی آزاد ایران

This paper examines specific types of fractional integro-differential equations, using an advanced method rooted in fractional calculus. The proposed technique extends the classical Frobenius approach by using the K-transform and the binomial series to solve these equations. We then apply this framework to emotion recognition in neuroscience, specifically by modeling Electroencephalography (EEG) signals generated during emotion experience. Those signals exhibit complex timing features, such as memory effects and long-term dependencies, which fractional models are well-suited to capture. The K-transform also handles nonlocal behavior more effectively, making it very suitable for systems involving fractional orders.  We validated our framework on a real-time EEG dataset collected from 20 participants (ages 18–40) using a Central Drugs Standard Control Organisation (CDSCO)-approved BCI device during baseline and neurofeedback sessions. The proposed method achieved an average classification accuracy of 89.3% (±2.1%), approximately 6% higher than that of integer-order models on the same dataset. While these findings highlight the advantages of fractional modeling for EEG-based emotion recognition, we note that the dataset size is modest, and further validation on larger multi-channel datasets will be pursued.