A risk assessment model for the entire rice processing chain based on Kmeans++ and extreme learning machine

The rice processing chain plays a crucial role in the rice supply chain, making the risk assessment of its main pollutants essential. Existing methods often rely on subjective weight determination and fail to consider the characteristics of indicators, leading to unreasonable results. Additionally, these models suffer from poor robustness and low accuracy. A risk assessment method based on Kmeans++ and extreme learning machine (ELM) is proposed in this paper, utilizing a multi-level risk indicator system. By integrating toxicological characteristics into the entropy weight, the comprehensive risk indicator for the processing chain is obtained. The Kmeans++ algorithm clusters the indicators to classify risk levels, while ELM is used for assessment. Validated on 75 rice processing data sets with six pollutants (Pb, Cd, Hg, aflatoxin B1, zearalenone, and deoxynivalenol), the method achieved a 93 % risk classification accuracy, outperforming back propagation (BP) and radial basis function (RBF) methods. This approach highlights the role of AI in enhancing food safety and supervision, supporting decision-making and rapid response in the digital transformation of agriculture. • A rice processing chain risk assessment method combining K-means++ clustering and ELM is proposed. • A multi-level risk assessment system integrated with entropy-weight method models rice chain contaminant risks • The method shows better risk classification performance than traditional BP and RBF networks. • This research improves AI application in food safety, providing reliable tools for authorities. • The approach lays a solid foundation for the development of rapid risk early-warning and mitigation systems in food safety.