This paper investigates the age of information (AoI)-based joint user scheduling and transmit power optimization on high-speed railway mobile networks, where a base station (BS) positioned close to the railway schedules track monitoring sensors and train passengers with a time division multiple access (TDMA) manner to fulfill the uplink transmissions. In order to minimize the average achievable AoI of the sensors, we formulate an optimization problem subject to the constraints of the minimal amount of data required to be uploaded by passengers and the available transmit power of the users including both the sensors and the passengers, by optimizing the user scheduling and transmit power. To efficiently solve the problem in the fast varying high-mobility environment, an algorithm based on deep Q-networks (DQN) is proposed, where the weighted sum of the sensors’ AoI and the passengers’ remaining amount of data to transmit are integrated into the reward function. Simulation results show that the proposed DQN-based algorithm reduces the average AoI of the sensors greatly, about 70% and 60% compared with traditional random algorithm and round-robin algorithm, respectively. Moreover, the increment of the numbers of passengers deteriorates the AoI performance, but among all simulated algorithms, the AoI achieved by our presented DQNbased algorithm is least affected by the numbers of passengers.