In order to improve the measurement accuracy of droplet particle size of domestic droplet spectrometer, standard particles were used to test the scattering response of the droplet spectrum, and the least mean square algorithm was used to invert the scattering response signal of the actually measured droplet spectrum. Through theoretical analysis and experimental verification, the particle size inversion data of the least mean square algorithm under different standard particle response signals were obtained. The results show that the inversion of standard particle distribution by the least mean square algorithm can reduce the distribution error of particle size so that the particles of different sizes can be accurately distinguished. The size distinction accuracy can reach 1 μm. This study provides a reference for improving the accuracy of the droplet spectrometer.