Post-Processing Techniques to Enhance Reliability of Assessment Algorithm Based Performance Measures

摘要

This study develops an enhanced transportation planning framework by augmenting the sequential four-step planning process with post-processing techniques. The post-processing techniques are incorporated through a feedback mechanism and aim to improve the stability and convergence properties of the solution, thereby improving the reliability of the planning process. There are three building blocks of the proposed post-processing module: slope-based multi-path algorithm or SMPA, perturbation assignment, and O-D prioritization technique. SMPA is the most important part of the module and it can be used as both a post-processing algorithm and an independent static traffic assignment algorithm. In addition to SMPA, the post-processing module consists of perturbation assignment and O-D prioritization schemes. Perturbation assignment provides warm start and the O-D prioritization improves the rate of convergence by deciding the sequence in which the O-D pairs are brought into the flow update process. A detailed implementation procedure for perturbation assignment is developed along with six different criteria for O-D prioritization. Computational experiments using the Sioux Falls, Anaheim and Borman Corridor networks provide insights on these techniques. Results of the computational experiments show that SMPA has a superior rate of convergence compared to state-of-the-practice algorithms even without O-D prioritization and warm start. a