Risk assessment on health damage of workers due to dust pollution at construction sites based on predicted value of backpropagation neural network

Dust pollution has serious health implications on construction workers. Past research has shown that dust exposure data can help improve the construction workers' awareness of dust pollution and urge them to adjust their work behavior. Therefore, in order to control dust pollution on sites, accurate measurement of dust concentrations is very important. However, the current risk assessment of health damage to workers is mostly based on measured values of health hazards at the construction site, and the error between measured values and actual concentrations of pollutants are often ignored. To overcome the above limitation, the aim of this research is to obtain risk of health damage to construction workers by simulating the exposure dose of dust in construction sites. Using an experimental design, this research measured dust concentrations at a construction site in Nanchang city of China and simulated it using backpropagation neural network. This is a widely used technique in the field of pollutant concentration prediction because of its powerful computational ability to solve nonlinear relationships. The study used nine monitoring points representing different operations while capturing four types of construction dust. It found work faces that have high dust concentrations and the ones that are more hazardous to workers with stable results and relatively high accuracy levels. In addition, this paper showed that backpropagation neural network could be used to assess health risks of dust pollution which is often neglected as a mere nuisance or discomfort due to its slow onset. The results provide a number of practical implications for construction project managers who are very keen to mitigate its impact with the help of the worker.