An Enhanced Clustering-Based Method for Determining Time-of-Day Breakpoints Through Process Optimization

Abstract

The fixed-time strategy is crucial in traffic signal control which applies signal plans to different time periods of the day. One critical step is to determine the optimal breakpoints to divide one day into periods with homogeneous traffic flow. Most existing methods are based on k-means clustering algorithm and have to select the optimal number of clusters. Since direct k-means and time-incorporated k-means clustering will result to noncontiguous time periods, several adjustments are needed including further partitioning and re-clustering via empirically adjustment which merges short time periods into adjacent longer ones to finalize the time-of-day (TOD) partition plan. Such adjustments can make the previous optimal number of clusters selection suboptimal. This paper proposes an enhanced method to determine optimal TOD breakpoints through optimizing the process. Instead of choosing the optimal partition plan before adjustments, we propose to determine the optimum after all the adjustments. A case study based on Qingdao City in China is presented to evaluate the added value of the enhanced method. It is shown through simulation experiments that the enhanced method can avoid over-partitioning and substantially improve the traffic operational efficiency especially during high demand periods.

Keywords

• Time-of-day
• TOD breakpoint
• k-means clustering%22%2C%22default_operator%22%3A%22AND%22%7D%7D%7D"> <italic xmlns:ali="http://www.niso.org/schemas/ali/1.0/" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">k</italic>-means clustering
• empirical adjustment