生物灾害科学2026，Vol.49Issue(2)：215-221,7.DOI:10.3969/j.issn.2095-3704.2026.02.25
水稻害虫物联网系统性能评估与数据校准研究
Performance evaluation and data calibration of ATCSP
摘要
Abstract
[Objective]This study aimed to evaluate the performance of the Jiaduo Major Agriculture and Forestry Pest Automatic Monitoring and Control System(ATCSP)in monitoring major rice pests and to develop an effective calibration method for its systematic errors.[Method]Field experiments were carried out in Hanshou county,Hunan Province,from June to November 2024,covering a total of 24 weeks.The monitoring targets included four major rice pests:Chilo suppressalis,Cnaphalocrocis medinalis,Sesamia inferens,and Spodoptera litura.By comparing the system's automatic identification results with manual counting data,the Pearson correlation coefficient was used to assess their consistency.A piecewise linear calibration model was established based on the characteristics of error distribution,and the leave-one-out cross-validation method was adopted to verify the model's correction effect.[Result](1)The system counts were highly consistent with manual counts in identifying the occurrence peaks of the four pests,with all Pearson correlation coefficients above 0.97(P<0.000 1),indicating high reliability of the system in trend monitoring.(2)System counts were generally lower than manual counts,with relative errors ranging from 17.2%to 49.1%;among them,Spodoptera litura showed the largest error(+49.1%)and Cnaphalocrocis medinalis the smallest(+17.2%).(3)The errors exhibited obvious density-dependent characteristics,increasing significantly at high pest densities.Accordingly,piecewise linear calibration models were constructed for each of the four pests respectively.Validation via the leave-one-out cross-validation method showed that the mean relative error between the corrected estimates and manual counts decreased significantly from+33.6%to+8.7%.[Conclusion]The ATCSP automatic monitoring and control system presents favorable reliability in monitoring the occurrence trends of major rice pests and can effectively identify pest peaks.The established piecewise linear calibration model can remarkably improve the accuracy of system counting,providing data support and technical basis for the practical application of Internet of Things(IoT)automatic monitoring systems in the precise prevention and control of rice pests.关键词
病虫害监测/物联网/误差分析/分段校正模型Key words
pest and disease monitoring/Internet of Things(IoT)/error analysis/piecewise correction model分类
农业科技引用本文复制引用
李晨宇,杨国萍,熊美云,孙明凤,徐雅倩,邹钦旋,乐熹子,李忠彩..水稻害虫物联网系统性能评估与数据校准研究[J].生物灾害科学,2026,49(2):215-221,7.基金项目
国家重点研发计划项目(2024YFD1400902) （2024YFD1400902）
