Lu Ping, Dong Wei, Jiang Tongqiang, Liu Tianqi, Hu Tianyu, Zhang Qingchuan
National Engineering Research Centre for Agri-Product Quality Traceability, Beijing Technology and Business University, Beijing 100048, China.
School of E-Business and Logistics, Beijing Technology and Business University, Beijing 100048, China.
Foods. 2023 Jan 26;12(3):542. doi: 10.3390/foods12030542.
Focused supervision and early warning of heavy metal (HM)-contaminated rice areas can effectively protect people's livelihood security and maintain social stability. To improve the accuracy of risk prediction, an Informer-based safety risk prediction model for HMs in rice is constructed in this paper. First, based on the national sampling data and residential consumption statistics of rice, we construct a dataset of evaluation indicators that can characterize the level of rice safety risk so as to form a safety risk space. Second, based on the K-medoids clustering algorithm, we classify the rice safety risk space into levels. Finally, we use the Informer neural network model to predict the safety risk indicators of rice in each province so as to predict the safety risk level. This study compares the prediction accuracy of a self-constructed dataset of rice safety risk assessment indicators. The experimental results show that the prediction precision of the method proposed in this paper reaches 99.17%, 91.77%, and 91.33% for low, medium, and high risk levels, respectively. The model provides technical support and a scientific basis for screening the time and area of HM contamination of rice, which needs focus.
对重金属污染稻田进行重点监管和预警，能够有效保障民生安全、维护社会稳定。为提高风险预测的准确性，本文构建了基于Informer的水稻重金属安全风险预测模型。首先，基于全国水稻抽样数据和居民消费统计数据，构建能够表征水稻安全风险水平的评价指标数据集，从而形成安全风险空间。其次，基于K-中心点聚类算法，将水稻安全风险空间进行等级划分。最后，利用Informer神经网络模型预测各省份水稻的安全风险指标，进而预测安全风险等级。本研究对自行构建的水稻安全风险评估指标数据集的预测准确性进行了比较。实验结果表明，本文所提方法对于低、中、高风险等级的预测精度分别达到99.17%、91.77%和91.33%。该模型为筛查需重点关注的水稻重金属污染时间和区域提供了技术支持和科学依据。
