Dynamic total least squares estimation of intersection traffic flow patterns

Abstract

Total least squares (TLS) technique is introduced to dynamically identify intersection traffic flow patterns when both of the observations for entering and exiting vehicles have random measurement errors. An algorithm of dynamic TLS estimations of intersection traffic flow patterns is proposed. Simulation experiments show that it can improve estimation accuracy substantially in comparison with recursive ordinary least squares estimations. Hence, the dynamic TLS estimations provide a competitive alternative for identifications of intersection traffic flow patterns.