Our streets and highways are getting more congested. Transit signal priority (TSP) control which is widely used at signalized intersections has been recognized as a practical strategy to improve the efficiency and reliability of bus operations. Conventional control strategy suffers from the incompetency to adapt to dynamic traffic situations. Recent studies proposed to use deep reinforcement learning (DRL) method to identify an efficient traffic signal control. However, these existing studies in DRL-based traffic signal control methods focus on private vehicles, paying less attention to the difference between transit vehicles and non-transit vehicles. Recently, the concept of 'pressure' from the traffic field has been utilized as the reward function in RL-based traffic signal control. In this study, we adopt the pressure concept and introduce the priority factor (PF) for TSP control. PF increases pressure and that pressure encourages agents to give the way to the bus movements. This is a simple and effective approach granting the buses crossing the signalized intersection. We tested the proposed method in VISSIM with an arterial and a grid network in a dynamic environment. The experiments demonstrate that agents can reduce bus travel time. Moreover, depending on the priority level, the agents can resolve the conflict of different bus routes by different levels of priority.