Abstract: ‍ ‍To address the issues of high energy consumption and delayed data collection in wireless sensor networks， an unmanned aerial vehicle （UAV）-based data collection optimization scheme utilizing intelligent reflecting surfaces （IRS） is proposed. Among them， multiple ground sensors with buffer zones collect environmental information， and a UAV with limited energy is deployed to collect data with the assistance of the IRS. By considering the freshness of information and the propulsion energy of the UAV， through the joint optimization of the UAV’s 3D flight trajectory， ground sensor scheduling， and IRS configuration， the weighted and optimal problem of the expected average age of information （AoI） and UAV propulsion energy consumption was constructed. Subsequently， the non-convex optimization problem was modeled as a Markov decision process， and a deep reinforcement learning algorithm was proposed to optimize the UAV data acquisition process based on the Manhattan city simulation environment. Finally， the optimized 3D flight trajectory of the UAV and the optimal configuration of the IRS were obtained. Simulation results show that the proposed optimization algorithm can effectively reduce system energy consumption while improving information freshness. When the number of IRS reflection units is the same， the system performance is improved by approximately 50.64% compared to the baseline schemes. The effectiveness of the proposed scheme and the contribution of the IRS to enhancing system performance are demonstrated.