Abstract:
Urban traffic congestion is becoming a major issue, resulting in prolonged travel times,
air pollution, fuel consumption, and driver resentment, which is frequently created by
vehicles searching for vacant parking spaces. In the development of a parking manage
ment system, the optimization of parking pricing, demand, and supply throughout the
course of a day is a significant issue. Dynamic vehicle parking pricing has emerged as
a promising strategy to manage parking demand, minimize cruising time, and maxi
mize revenue and enhancing traffic flow. This study employs competitive game theory
among parking agents to formulate a new dynamic parking pricing model. Bi-level op
timization problem is developed to regulate traffic flow and maximize profit for parking
agents. Evolutionary algorithm is used to solve the problem. Extensive numerical sim
ulations are executed using randomly generated data to evaluate the optimal pricing
strategies and maximization of profit for parking agents. Predicting parking demand
is also the focus of this study. Time varying discrete non-homogeneous Markov chain
model is used to predict demand. An adaptive learning algorithm is proposed to enable
the non-homogeneous Markov chain to respond effectively to changes in the dynamic
demand environment. Case study has been carried out based on the proposed al
gorithm. The result derived from data predictions, along with the integration of an
adaptive strategy, is presented to enable the system to learn from new changes. Sensi
tivity analysis is performed to assess the impact of learning parameters on prediction
accuracy. Optimal parking lot choice strategy is also further investigated in this study.
The total travel time and the expected cost of parking lots are considered as conflict
ing objectives for the decision-maker. Bi-criteria optimization is used to formulate the
parking lot choice problem. The Pareto optimal Parking lot is identified by applying
ideal point method. We conduct numerical simulations and apparently generate data
to show the feasibility of the proposed algorithm.
