FOOD SCIENCE ›› 2020, Vol. 41 ›› Issue (9): 15-22.doi: 10.7506/spkx1002-6630-20190427-373

• Basic Research • Previous Articles     Next Articles

Risk Assessment of Hazardous Materials in Grain Supply Chain Based on Frequent Pattern Growth Combined with Self-Organizing Maps (FPG-SOM)

WANG Xiaoyi, WANG Zhenni, KONG Jianlei, JIN Xuebo, SU Tingli, BAI Yuting   

  1. (1. Artificial Intelligence Academy, Beijing Technology and Business University, Beijing 100048, China; 2. Beijing Key Laboratory of Big Data Technology for Food Safety, Beijing Technology and Business University, Beijing 100048, China)
  • Online:2020-05-15 Published:2020-05-15

Abstract: In order to scientifically and reasonably evaluate the comprehensive risks of hazardous materials in each link of the grain supply chain, sample survey data from the grain supply chains in many provinces across the country and data from other dimensions were analyzed in this paper. On this basis, a multidimensional hierarchical risk indicator system was built by using risk factors in the grain supply chain to convert a large number of multidimensional heterogeneous data into semi-quantitative risk indicators. The association rules were applied to excavate the intrinsic correlation between the first-level indicators and the second-level indicators for determining the weight distribution. Further, the self-organizing maps algorithm was used to map each indicator variable to a risk level for analysis of the cross-correlation. Finally, a comprehensive evaluation method for risk levels of hazards in the grain supply chain. By evaluating the risk level of grain products, it was concluded that the key provinces with higher risks were Shandong and Henan provinces, typically in urban areas, and the key link was circulation as well as a series of high-risk hazards, represented by aluminum residues. The evaluation system established in this paper provides a scientific basis for the regulatory agencies to develop target-oriented sample survey strategies, establish priority supervision areas and legitimately allocate supervision resources.

Key words: multidimensional hierarchical index system, association rule mining, self-organizing mapping, comprehensive risk assessment

CLC Number: