Mental Workload Assessment System During Controlling Uav

Mental Workload(MWL)represents the occupancy of brain resources under working conditions and is an important factor affecting job performance.Appropriate brain load will significantly improve the efficiency of workers in the work.EEG can sensitively respond to the neural electrical response caused by changes in MWL,which is a hot spot in current research.The breakthrough of this technology will promote the development of adaptive automation.Based on the UAV simulation platform,this paper studies the relationship between the MWL of the operator and the EEG,in the UAV control task.Finally,from multi-modal physiological signal synchronous acquisition platform(MPSSA)construction,experimental design,EEG signal processing,The construction of the MWL recognition model is studied to establish MWL recognition system.For the problem of how to improve the signal-to-noise ratio,combining with the characteristics of ECG,EMG,and EEG,A new analog amplification filter structure is established,which increases the signal-to-noise ratio of the physiological signal,expands the dynamic input range of the signal,and solves the problems of baseline drift and polarization voltage;designing power circuits for safety issues;Designing digital circuits,computer programs for signal synchronization.The entire system can simultaneously collect valid signals.The experimental design is divided into two parts.The MWL experiment is supplemented by the subjective scale method and task performance method.Based on the Phoenix RC drone simulation system,the experiments are designed to induce different levels of MWL.After statistical calculations,there were significant differences in task performance and subjective scale scores under different MWL experiments,and the experimental design was reasonable and effective.The noise experiment design is based on the existing research to determine the main interference source signals of EEG,guide the subjects to complete specific actions to induce noise,collect the noise source signal and the signal that the noise spreads to the scalp.Through analysis and research developing a set of fully automated EEG noise processing algorithms for few channels(less than 16).Process the experimental data to explore the relationship between EEG activity and MWL,and establish a brain load recognition model across time periods and individuals.The SVM model based on power spectrum and mutual information features can reach an accuracy of 92.6 %.The algorithm has short calculation time and strong real-time performance,and is used for real-time evaluation of brain load.The transfer learning model based on EEMD and recursive graph can reach an accuracy of 92.8%.The algorithm has high accuracy and stability,and is used for offline evaluation of mental load.In short,this research has completely set up a set of automatic MWL assessment system based on the drone control task.