Aiming at the unclear relationship between spray parameters of nozzle and atomization parameters in the aviation spray mode, and considering the urgent need of precision aerosol spray, spray performance test and agent mathematical model modeling method were combined to discuss the influence of main working parameters on atomized particle size and spray width for the CN1215 special aviation centrifugal nozzle. The operating parameters of the liquid supply system corresponding to the spray parameters of the centrifugal nozzle were calibrated, and then the variation laws of droplet size and spray width under the influence of the working parameters (flow rate range was 100~350mL/min, nozzle rotating speed was 8000~10000r/min) were also analyzed in an indoor windless environment. Secondly, taking the spray parameters of the nozzle (nozzle flow, nozzle speed) as the test factor, and taking the droplet diameter (Dv50) and the spray width corresponding to droplet diameter (Dv50) as response factor, three kinds of mathematical methods, including fourthorder response surface method (RSM), Kriging method and ellipsoidal basis function neural network (EBFNN), were used to approximate the relationship between experimental factors and response factors respectively. The agent mathematical models between the nozzle atomization parameters (Dv50, and corresponding spray width) and nozzle operating parameters (nozzle flow, nozzle speed) were established. The decisive coefficients R2 of the three agent models for the particle size Dv50 were 0.705, 0.718 and 0.925, and the decisive coefficients R2 of the three agent models for the Dv50 corresponding spray width were 0.819, 0.890 and 0.930, respectively. Based on the EBFNN implicit proxy mathematical model, the response surface of two atomization parameters was established, the rapid prediction of droplet Dv50 and spray width under the influence of working parameters was achieved. Based on the EBFNN implicit proxy mathematical model, the response surfaces of two atomization parameters were established, which realized the rapid prediction of droplet Dv50 and spray amplitude under the influence of spray parameters. It was of great significance for accelerating the development of aviation precision pesticide application.