Predicting transient particle transport is crucial to address the risks posed to human health by indoor contaminants and to improve the design and control of ventilation systems. The Markov chain model with a coarse matrix has proven to be efficient in this regard, demonstrating faster speeds than traditional methods and higher computational accuracy when based on non-uniform states. However, its application under unsteady airflow conditions is limited, and discussions of how to determine non-uniform Markov states and their corresponding transient transition probabilities when the flow field changes are limited. Therefore, this study developed an application method of Markov chain model based on fixed non-uniform states in unsteady airflow fields. First, this model rapidly provides various Markov state schemes based on airflow field velocity and clustering methods. It then selects the optimal scheme through frequency and finally obtains the transition matrix through Lagrangian tracking for particle transport prediction. This model reduces the computational cost of transient particle transport when the wind speed changes while ensuring high computational accuracy. The proposed method was validated using published flow field experiments and simulation data. The computational speed of the proposed method was compared with computational fluid dynamics (CFD), and in the case studies considered in this research, the normalized root-mean-square deviation of the model was less than 10% and the total computation time was more than 75% faster than that of CFD.