Discriminant Analysis of Soil Type by Laser Induced Breakdown Spectroscopy Combined with GA-BP Neural Network

Laser induced breakdown spectroscopy (LIBS) and genetic algorithm (GA) were used to optimize the neural network for classification and identification of six kinds of crop soils. Through laser-induced breakdown of soil surface to produce plasma spectrum, 100 sets of spectral data of each soil were collected by spectrometer, and the main element characteristic spectral lines were accurately calibrated with NIST atomic spectrum database as reference. In the experiment, six characteristic spectral lines are selected for analysis. After the dimension of spectral data is reduced by principal component analysis (PCA), the recognition efficiency of BP neural network is greatly improved, and the average recognition accuracy is as high as 99.5%. The study of LIBS technology and BP neural network can provide reference for finding suitable soil for crop planting.