The decoupling in the fiber Bragg grating flow and temperature composite sensing is a difficult problem. To address this issue, a fiber Bragg grating flow and temperature composite sensor based on particle swarm decoupling algorithm is proposed. Firstly, combining the fiber Bragg grating sensing theory and the flow and temperature composite sensing theory, the flow and temperature composite sensing mechanism based on the fiber Bragg grating is studied. Then, a fiber Bragg grating flow and temperature composite sensor that integrate target structure with the cantilever beam of hollow cylinder is designed, a flow and temperature experiment system platform is established. The temperature and flow composite sensing experiments are carried out. Finally, a FBG flow and temperature composite sensor decoupling method based on the particle swarm algorithm is proposed. The proposed particle swarm optimization algorithm is used to decouple the experimental data from the flow and temperature. Research results after decoupling show that the maximum flow error of the sensor in the range of 3~ 8 m 3 / h is 0. 014 m 3 / h, the maximum temperature error is 0. 021℃ , the flow measurement error is 0. 28% , the temperature measurement error is 1. 5% , the flow mean-square error is 1. 16×10 -4 m 3 / h, and the temperature mean-square error is 1. 53 × 10 -4℃ . Compared with the neural network algorithm, results show that the particle swarm optimization algorithm has a good decoupling effectiveness. The measurement accuracy of the sensor could be improved effectively.