Traffic flow modelling for uphill and downhill highways: Analysed by soft computing-based approach

Researchers have made significant strides in understanding car-following behaviour and traffic flow, especially with the advent of intelligent and networked technologies. Diverse mathematical models analyse traffic flow, each with pros and cons. This study focuses on a sensitivity-based mathematical model for uphill and downhill highways, examining position, velocity, and acceleration profiles to predict traffic jam occurrence. The model employs an ordinary differential equation and a machine learning-based approach (machine learning procedure neural network) for numerical solutions, exhibiting high accuracy (10-8 - 10-10) compared to the reference Runge-Kutta method For accuracy, reliability and stability of the results are evaluated by various performance indicators and statistical terms. For multiple independent executions, mean absolute deviation, root mean square error and error in Nash- Sutcliffe efficiency are calculated. Their values are lies in range 10-8-10-14. Moreover, graphical analysis is established for better visualization of traffic flow and congestion.

Automated summary

Researchers have developed a sensitivity-based mathematical model for predicting traffic congestion on uphill and downhill highways. The model utilizes an ordinary differential equation and machine learning approach, providing accurate results. The reliability and stability of the model are evaluated using various indicators and statistical terms, and a graphical analysis is also presented for better visualization of the traffic flow and congestion.