ABSTRACT

Space–time prism (STP) modeling for activity programs with various realizations of activity chains has been a challenging research topic. This study claims that a bidirectional search scheme in a multi-state supernetwork is capable of pinpointing the exact STP of an activity program. The correctness and search space are first analyzed for the existing two-stage bidirectional search methods originally suggested for constructing trip-based STPs. Two simultaneous bidirectional search methods are further suggested. Travel time lower bounds based on A*, landmarks, and triangular inequalities are applied in goal-directed searches to reduce the search space. The small twists in formalism over the existing methods ensure accuracy and computational efficiency. The performances of the different search methods are compared for conducting activity programs in large networks.

摘要

具有各种活动链实现的活动程序的时空棱镜 (STP) 建模一直是一个具有挑战性的研究课题。这项研究声称多状态超网络中的双向搜索方案能够精确定位活动程序的确切 STP。首先分析了最初建议用于构建基于旅行的 STP 的现有两阶段双向搜索方法的正确性和搜索空间。进一步提出了两种同时双向搜索方法。基于 A*、地标和三角不等式的旅行时间下限应用于目标导向搜索以减少搜索空间。对现有方法的形式主义的小改动确保了准确性和计算效率。