A Driver Fatigue Detection Method Based on Improved Particle Swarm Optimization for Long and Short Time Memory Network

In order to enhance the accuracy of detecting driver fatigue, this paper proposed a new method of driver fatigue detection based on improved particle swarm optimization (PSO) to optimize long and short time memory network (LSTM). The experimental data utilized in this study was sourced from the fatigue driving dataset made available by Min Jianliang et al. After pre-processing the original EEG data, wavelet packet decomposition and reconstruction were carried out to extract an energy feature and sample entropy feature. The LSTM neural network optimized by improved particle swarm optimization algorithm was used to train and classify the two kinds of feature data respectively, and the short term memory network model optimized by ordinary particle swarm optimization algorithm (PSO-LSTM) was compared with the short term memory network model without particle swarm optimization (LSTM). The findings indicate that the classification accuracy of energy features is the highest, which is 93.06%. The classification accuracy of sample entropy is the highest, which is 90.83<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>. The feasibility of this method is illustrated.