Monitoring hoisting workers' negative emotions in prefabricated construction using galvanic skin response measurements

Prefabricated building transfers a large number of wet works to factory component production and on-site rapid assembly. However, the increasing number of special hoisting operations and the dangerous construction environment have added significant safety hazards to on-site workers. This paper develops a management-oriented warning framework for prefabricated hoisting by monitoring workers' negative emotions using galvanic skin response (GSR) measurements. The GSR signals of 20 participants are recorded while they performed a simulated hoisting task in the laboratory, and a negative emotion classification K-nearest neighbors algorithm is trained in the laboratory and preliminarily validated on site with one worker. The mapping relationship between emotions and unsafe behaviors is also qualitatively analysed through a questionnaire survey. A total of 16 unsafe behaviors are identified involving three categories of hoisting workers. The negative emotion classification achieves an average accuracy of 91.67% (F1-score: 0.91) in the laboratory and 90% (F1-score: 0.93) on the construction site. The device and algorithm employed in this study can be easily applied to track the negative emotional status of hoisting workers. The framework can support proactive safety management by providing managerial-level warning in precast hoisting scenarios.