With the rapid development of the catering industry, catering fume pollution has become one of the important sources of urban air pollution. How to realize the purification of catering fumes is a huge problem related to public life health and safety. This paper first constructs the structure of self-cleaning low-temperature plasma soot purification and provides the equipment, materials, and process required for soot purification experiments. Secondly, CFD simulation of the flue of the oil smoke purification device is introduced into Computational Fluid Dynamics, and the relevant physical parameters of the oil smoke flow are obtained by solving the equations of mass conservation, momentum conservation and energy conservation. The self-cleaning low-temperature plasma oil smoke purification device was also analyzed by numerical simulation using Fluent software, and the method for detecting VOCs in restaurant oil smoke was given. Lastly, the oil smoke purification device constructed in this paper was tested for its impact on oil smoke. The results show that for every 100W·m−2 increase in the intensity of oil smoke, the temperature error will increase by about 0.01℃ accordingly, and when the intensity of oil smoke reaches 500W·m−2, the temperature error of the oil smoke purification device is 0.074℃. The highest imported concentration of all kinds of VOCs in the barbecue smoke produced by the experiment reached 3762.53 μg / m
3, and the average treatment efficiency of 98.69% was achieved under the actual air volume of about 800m3/h. This shows that the use of computational fluid dynamics can realize the simulation analysis of a self-cleaning low-temperature plasma fume purification device, and the fume purification device has a strong fume purification ability.