Traffic safety assessment method of the immersed tunnel based on small target visual recognition image

Abstract

The quality of lighting installation performance has a direct impact on the traffic safety of immersed tunnels. To effectively investigate and judge the traffic safety of immersed tunnels having different lighting installations, a traffic safety assessment method for immersed tunnels based on lighting performance degradation was put forward in this study by using big data technology. Numerical simulation was used to simulate the lighting environment in an immersed tunnel under different conditions of lighting performance degradation, conduct the small target recognition test in a physical tunnel, and calculate the traffic safety factor; then, a real-time kinematic assessment model of traffic safety in immersed tunnels was built in combination with the key index factors influencing lighting installations in immersed tunnels. The test results showed that the performance degradation of lighting installations positively correlated with the visual cognition of drivers and passengers. long short-term memory neural network model can effectively assess the traffic safety of immersed tunnels, and the root mean square error (RMSE) and coefficient of determination of the model were separately 1.029 and 0.95, which were superior to the RMSE and coefficient of determination of random forest and recurrent neural network model, and the running time was often less than 1min, complying with the rea; -time assessment requirements; the boundary value of the traffic safety factor of immersed tunnels was 0.6304, and if a value was less than the boundary value, it indicated that the performance of lighting installations was not good and might pose a threat to traffic safety. The research results provided a new perspective for the status assessment of lighting installations in immersed tunnels and also offered a theoretical basis for fine maintenance and repairs of lighting installations.