Risk assessment of cadmium pollution in selenium rich areas based on machine learning in the context of carbon emission reduction

Machine learning is of great value for the situation analysis and scientific prevention control soil heavy metal pollution risk. In this paper, taking selenium rich area as research object, improved Genetic Algorithm (GA)–Back Propagation (BP) algorithm was used to construct risk assessment model Cd in area. Firstly, content Se study statistically analyzed based on descriptive statistics correlation analysis. Then, a three-layer BP neural network structure designed optimized by GA algorithm. The individual coding length calculated connecting weights thresholds elements. Based 97 groups field data area, experimental results show that has faster convergence speed, maintains good generalization ability test sample points. Compared with multiple linear regression (MLRM), GA-BP reduces RMSE 64.84, 52.12, 49.53, 63.18% compared M5. accuracy estimating status different areas higher than other three models whole. whole region, samples safe interval, relatively light moderate interval severe accounted 4.12, 8.24, 42.26, 17.52 27.86%, respectively, prediction level showed only 12.36% were state without pollution, while most mild state. Because huge potential carbon sequestration emission reduction agriculture, planting se-rich Cd-low crops these can not promote development local Se-rich industries but also achieve reduction.