Abstract: The detection chamber is an important part of the bionic olfactory detection analysis instrument. Unreasonable structure design of detection chamber can cause gas vortex and prolong the sensors' response and recovery time. Non-uniform distribution of airflow field can cause inconsistence of the sensor array in numerical induction. Therefore, it is significant to explore the characteristics of gas flow and distribution, optimize the flow field structure, improve the uniformity and stability of the flow rate and advance the accuracy and repeatability of olfactory detection chamber. However, the traditional method of further manufacture design improvements requires long transformation time, and high costs, and the measurement range is usually disappointing. Computational Fluid Dynamics (CFD) can provide detailed information on airflow simulation and ensure convenient design of agricultural equipments. In order to optimize the structure of bionic olfactory detection chamber of sweeping type, and improve the uniformity and stability of fluid velocity distribution, based on the differential equations of fluid motion, the internal flow field of olfactory detection chamber was numerically simulated by using the CFD. This paper proposed the models of 3 types of chamber detection original structures, and the uniformity and stability of olfactory sensors' detection area from the optimal model were compared with the test results. The 3 types of detection chamber original structures were linear arrangement, parallel arrangement and structure of multi nasal ducts. Each model design was mainly composed of an intake pipe, a detection chamber, a sensor array and a vent pipe. The fluid flow rate was much smaller than the acoustic velocity and the fluid flow was considered as an incompressible process inside the chamber. In the process of olfactory detection, the fluid flowed into the detection chamber from the intake pipe and flowed out of the vent pipe, so the inlet boundary was set to velocity inlet. Outlet pressure boundary conditions were selected, and the natural pressure was taken as the boundary value. The fluid temperature in the chamber was room temperature which was 26 ℃. The wall had no slip boundary condition, and was assumed to be a rigid wall without considering the influence of wall elasticity. The speed deviation ratio and the nonuniformity coefficient were chosen as comprehensive evaluation indicators. The velocity distribution in chamber flow field was obtained and used to analyze the original structures to provide the preference design. The detection chamber of multi nasal ducts corresponded to inclination model. The simulation results indicated that the structures of detection chamber influenced air velocity distribution. The detection chamber of multi nasal ducts along the pipe axis (0.035-0.049 m range) had a velocity smooth region, the velocity of which was stabilized merely at 0.018-0.268 m/s, which could meet the requirement of detection condition. There were no velocity smooth regions in the linear arrangement, and the airflow velocity of smooth regions in parallel arrangement merely ranged from 0.001 to 0.018 m/s. The detection chamber of multi nasal ducts showed the advantage in the uniform and stability of velocity, of which the maximum values of speed deviation ratio and nonuniformity coefficient were calculated to be 0.830 6 and 0.292, respectively. Meanwhile, in the gas detection period by the numerical simulation, the detection time of 3 models was 223.4, 302.0 and 213.8 s, respectively, and the minimum value of the structure of multi nasal ducts showed that it had fast flow response and high working efficiency. Moreover, the optimum structure could effectively improve the consistency of the sensors' numerical detection, and the standard deviation and the coefficient of variation ranged from 0.153 5 to 0.428 3 and from 0.030 5 to 0.082 7, respectively. The results provide a reference for the uniformity design of flow field structures similar to the detection chamber.