EEG Signal Characteristics Of Construction Workers’ Hazard Recognition

Hazard recognition has been the first step to recording,rectifying,and reducing the probability of accidents.Although studies attempt to create sues through platforms,devices,and approaches for hazard warning purposes,workers have to leverage safety and production goals during task performance in a fast-changing working environment and thus attenuate their motivation to engage with these technologies.In order to better recognize workers’ risk perception from an objective perspective,this study proposes a procedure to automatically recognize workers’ risk perception in construction sites using EEG signals.Originally,an algorithm is proposed to analyze the EEG signal characteristics of hazard recognition.Firstly,taking the front-line construction workers on the construction site as the experimental object,55 effective participants were collected through portable EEG signal acquisition equipment to observe the EEG signals excited by different types of(hazard/safety)pictures.Secondly,the data is screened,and the features are extracted from the time domain and frequency domain,which are used as the independent variables in the model training of the supervised learning algorithm,and the different hazard perceptions of workers when observing pictures are used as the dependent variables for training.Finally,the accuracy of classification prediction is verified by cross-validation.The results show that: 1)in the application of support vector machine(SVM)classification and prediction,the accuracy of ten-fold cross-validation is 70%,and it is the optimal classification algorithm judging according to the decision boundary.2)When the participants with different lengths of working experience and hazard tendency are classified with the same characteristics,the hazard perception and recognition model based on the SVM algorithm is consistent in different kinds of people;3)For the participants with higher seniority,only EEG signals located in the frontal lobe can effectively capture the perception of hazard recognition;For the participants with lower working age,the EEG signals of parietal lobe should be used as a supplement on the basis of collecting the EEG signals of frontal lobe;4)The activation of the parietal region means the location perception of human and space,which supports the view that lowlevel participants may fail to perceive hazards due to spatial perception obstacles.After mining the time-frequency characteristics of EEG signals and determining different super parameters,this study selects the appropriate classification model and puts forward a feasible classification and recognition method of hazard recognition for construction workers based on EEG.The process of channel selection for capturing different hazard recognition also provides an argument for the subsequent interpretation of brain function from the theoretical level.From the practical level,the proposed feature extraction and classification methods also lay the foundation for the development of hazard perception enhancement technology.