Research On Dynamic Assessment Of Construction Safety Risks Of Bridge Operation At Height Based On Human Factors Engineering

Since entering the 21st century,the development of transportation infrastructure has grown steadily,China’s bridge production has developed rapidly,and bridge construction has played a vital role in the construction of transportation infrastructure.According to statistics,in the past 20 years of bridge safety accidents,analysis found that 90% of highaltitude work accidents,and human behavior decision-making errors have become the main influencing factors of brewing accidents.Therefore,how to dynamically assess the safety risk of bridge operation at height is very important for the accurate control of bridge operation safety.Taking the construction of bridge works at heights as the research object,based on the analysis of existing research,relevant indicators are summarized from three aspects of physiology,psychology,and environment.According to the length of update time of indicator data,they are divided into two categories: inert indicator system and active indicator system.The reliability and validity of the inert indicator system data were analyzed using EXCEL and SPSS based on on-site questionnaire data,indicating that the reliability and validity of the questionnaire were high.The Pearson correlation coefficient analysis of the inert indicator was conducted,and a multiple linear regression model was established for the basic skills,neuroticism,openness,extroversion,conscientiousness,amiability,social psychology,organizational management,and unsafe behavior of operators.The active indicator data changes in real time.Through field testing and functional correlation analysis using the correlation coefficient matrix,select indicators with strong correlation,and ultimately build a dynamic indicator system for safety risk assessment of bridge work at heights based on human factors engineering.By deploying portable weather stations,high-definition cameras,and workers wearing smart watches at the construction site to obtain active index data,using MATLAB for data processing,taking the frequency of abnormal heart rate and blood pressure per unit time,the abnormal number of workers in the area per unit time,and six active indicators of temperature,humidity,wind speed and noise as input indicators,the structure of "6-4-1" standard BP neural network and the BP neural network optimized by genetic algorithm(GA)were constructed.Through data verification,the accuracy and error of the model are analyzed by comparison,and the GA-BP network model with high accuracy is selected to predict unsafe behavior,and the feasibility of the dynamic prediction model is verified by applying it to the actual project.