Abstract: In order to realize adaptive and intelligent control of aspirating quantity for coal residual cleaning robot under the different distribution of residual coal in the freight train carriage, a control model based on the aspirating quantity, residual coal volume and the fan current was established, and the aspirating quantity can be controlled intelligently according to the changing trend of fan current. Firstly, fruit fly optimization algorithm was applied to obtain an optimal smooth factor of probabilistic neural network. Then an accurate prediction model for the fan current based on the improved probabilistic neural network was constructed consequently.And a fuzzy step adjustment control rule base was set up by the combination of equipment characteristics and field experience.In order to verify the relevant theories and methods, an intelligent control experiment of aspirating quantity was carried out on residual coal cleaning robot for HCQS-75/110 train carriage. The results show that the prediction accuracy of fan current by the improved probabilistic neural network reaches 97.4%.The cleaning robot equipped with intelligent control system of aspirating quantity can achieve a pass rate of 98.2% and save 26.5% energy consumption, which can meet the requirements of on-site environmental protection and energy consumption.