为解决当今世界很多地区出现虫灾的问题，本研究以2021年1月大规模出现在美国华盛顿州亚洲大黄蜂(胡蜂)为例，通过胡蜂出现地理位置，使用灰色系统预测GM(1,1)结合ARIMA预测下一次胡蜂出现的位置来构建虫灾预测与防控系统。该系统具有在灾害区域中预测蜂群的动态、对目击者上传的图像等信息进行核实、实时进行系统更新、预测灾害结束时间的功能。研究对当地用户的评论与图像信息使用CNN结合SVM进行核实，对数据进行准确的筛选来确定虫灾严重的区域。同时，分析两次目击胡蜂的时间进行，对模型更新时间进行预测得到变动的区间。最终通过分析整个蜂群的动态，预测灾害的结束时间，从而提高病虫害防治工作情况的动态、信息化管理。
In order to solve the problem of insect plagues in many parts of the world today, this study uses the large-scale appearance of Asian hornet (Vespa) in Washington State in the United States in January 2021 as an example. The geographical location of the wasp is used to predict GM (1,1) using the gray system, combining with ARIMA to predict the location of the next wasp, build a pest prediction and prevention system. The system has the functions of predicting the dynamics of the bee colony in the disaster area, verifying information such as images uploaded by witnesses, updating the system in real time, and predicting when the disaster will end. The research uses CNN combined with SVM to verify the comments and image information of local users, and accurately screens the data to iden-tify areas with severe pests. At the same time, the time of two sightings of the wasp is analyzed, and the model update time is predicted to obtain the range of change. By analyzing the dynamics of the entire bee colony and predicting the end time of the disaster, we can improve the dynamic and information management of the pest control work.
Alaniz, A.J., Carvajal, M.A. and Vergara, P.M. (2021) Giants Are Coming? Predicting the Potential Spread and Impacts of the Giant Asian Hornet (Vespa mandarinia, Hyme-noptera: Vespidae) in the USA. Pest Management Science, 77, 104-112. https://doi.org/10.1002/ps.6063
[4]
Washington State University (2020) Scientists Predict Potential Spread, Habi-tat of Invasive Asian Giant Hornet.
Ferraz, C.T., Barcellos, W., Pereira Jr., O., et al. (2021) A Comparison among Key Frame Extraction Techniques for CNN Classification Based on Video Periocular Images.
