Modelling individual and household activity : travel scheduling behaviours in stochastic transportation networks

xvi, 117 leaves : illustrations ; 30 cmPolyU Library Call No.: [THS] LG51 .H577M CEE 2014 XiongThe fundamental objective of transportation planning is to provide appropriate transportation facilities to meet future travel demand. Knowledge of travel demand can help improve the efficiency and sustainability of the existing transportation systems. Various approaches, such as network equilibrium models and simulation systems, have been proposed to predict travel demand. The activity-based network equilibrium approach for analysing travel behaviour and predicting travel demand has emerged over the last two decades. The activity-based network equilibrium approach covers a new class of models that predict where and when activities are pursued. The main contribution of this thesis is the development of a modelling framework for the individual and household daily activity-travel scheduling behaviour. The existing activity-based network equilibrium modelling methodologies are extended in two directions: (1) developing a dynamic model for individual’s daily activity-travel scheduling behaviour in congested networks, (2) taking into account the impact of intra-household interactions on daily activity-travel scheduling. In previous activity-based network equilibrium models, travellers are assumed to care only about the utility that can be obtained immediately. Travellers would thus choose the activity that provides the highest immediate utility but would ignore the utility that could be obtained during the remainder of a day. The daily activity-travel schedule is thus composed of a sequence of repeated static choices over an entire day. In this thesis, a Markov Decision Process (MDP) model is proposed to capture two types of within-day dynamics in individual’s activity-travel scheduling behaviour. Firstly, travellers take into account the expected future utility and perceive the impact of the current decision on subsequent activities and trips. Secondly, activity-travel decisions have a strong dependency on contextual situations, such as time of the day and location.The impact of travel time uncertainty on activity-travel scheduling is also considered in the proposed MDP model. For example, a sudden increase in travel time to the destination of a non-compulsory activity may reduce the probability of choosing this activity. The MDP model provides a framework for modelling inter-temporal activity-travel decisions in such an uncertain environment. The development of a bridge between the activity choice behaviour and the long-standing network equilibrium models has attracted considerable research efforts. The supernetwork representation is adopted as a unified framework for modelling activity location, time of participation, duration, and route choice in congested networks. However, imposing simple constraints on the activity-travel schedules makes the supernetwork models computationally intractable. In this thesis, the supernetwork models are shown to be special cases of the MDP model. The computational burden can be alleviated by exploiting the special structure of the MDP model. Dynamic programming algorithms are developed to solve the MDP model without enumeration of all the possible activity-travel schedules. In traditional travel demand models, travellers within a household are treated separately. Each household member makes activity-travel decisions independently. As a result, the estimation of joint activity participation can be biased. In this thesis, a household MDP model is proposed to explore the influence of intra-household interactions on individual’s activity-travel scheduling behaviour. A utility function is adopted to represent joint household preference. The function consists of a weighted sum of each household member’s utility, together with a term measuring the level of intra-household interactions. The variation in intra-household interactions across activity types is thoroughly examined using the household MDP model. The intra-household interactions are considered as weak or negligible for compulsory activities, such as work and school. These activities are conducted with significant regularity and under strict spatial and temporal constraints. Joint non-compulsory activities are motivated by collaboration and companionship. The intra-household interactions are positive for these activities. For example, one household member’s choice of activities, such as social visits and outdoor sports, is highly related to the other’s choice. The intra-household interactions are negative for other non-compulsory activities, such as cooking and cleaning. If one household member has completed one such activity, the other does not have to undertake this activity but still benefits from its results. To minimize the effort spent on these activities, a household simply allocates each of these activities to one household member.Department of Civil and Environmental EngineeringM.Phil., Dept. of Civil and Environmental Engineering, The Hong Kong Polytechnic University, [2014