文献类型：期刊文献

中文题名：基于人工智能的森林火灾发生预测研究进展

英文题名：Research Progress on the Prediction of Forest Fire Occurrence Based on Artificial Intelligence

作者：王京鲁[1,2] 李杰[3] 冯晓川[3] 朱林[4] 李鹏[4] 郝立坤[4] 宋增兵[4] 王明玉[1,2] 舒立福[1,2] 司莉青[1,2] 李伟克[1,2]

第一作者：王京鲁

机构：[1]中国林业科学研究院森林生态环境与自然保护研究所,北京100091;[2]国家林业草原火灾监测预警与防控工程技术研究中心,北京100091;[3]国家林业和草原局调查规划设计院,北京100714;[4]北京市园林绿化局森林防火事务中心,北京102200

年份：2025

卷号：5

期号：3

起止页码：81-89

中文期刊名：陆地生态系统与保护学报

外文期刊名：Terrestrial Ecosystem and Conservation

基金：“十四五”国家重点研发计划项目(2023YFD2202005,2023YFD2202001);国家林业和草原局森林雷击火防控-揭榜挂帅项目(2023132032)。

语种：中文

中文关键词：人工智能;森林火险;森林火灾发生预测;气象预测;预测模型

外文关键词：artificial intelligence;forest fire risk;forest fire occurrence prediction;weather forecasting;prediction model

分类号：S762.2

摘要：在气候变暖和极端天气的影响下,森林火灾发生日趋频繁,准确预测森林火灾发生对保护森林生态系统、经济和人员安全具有重大意义。本文首先回溯了典型森林火灾发生预测系统模型,其次综述了森林火灾发生预测的人工智能算法模型及其应用,并讨论了气象预测大模型驱动森林火灾发生预测模型的必要性、可实施性及模型构建思路,以实现中长期森林火灾发生预测,最后总结了人工智能在森林火灾发生预测应用中的优势和问题。同时,本文展望了基于人工智能的森林火灾发生预测的研究方向,一是引入其他模型或结合知识驱动提高预测模型的可解释性;二是提高气象预测模型输出数据的时空分辨率,更好地监控天气行为,增强基于人工智能的森林火灾发生预测模型的应用能力;三是建立预测大模型,保障应急安全;四是具身智能设备可植入森林火灾发生预测模型,收集分析林火因子,输出林火发生概率,回传数据并自主调整行动策略。
Under the influence of climate warming and extreme weather,forest fires are occurring more and more frequently.Accurately predicting the occurrence of forest fires is of great significance for protecting forest ecosystems,the economy and the safety of people.This paper first reviews the typical prediction system models for the occurrence of forest fires.Secondly,it summarizes the artificial intelligence algorithm models for predicting the occurrence of forest fires and their applications,and discusses the necessity,feasibility and model construction ideas of meteorological prediction large models driving the prediction models for the occurrence of forest fires,so as to achieve medium and long-term prediction of the occurrence of forest fires.Finally,it sums up the advantages and problems of the application of artificial intelligence in predicting the occurrence of forest fires.Meanwhile,this paper looks forward to the research directions of predicting the occurrence of forest fires based on artificial intelligence.First,introduce other models or combine knowledge-driven approaches to improve the interpretability of prediction models.Second,improve the spatio-temporal resolution of the output data of meteorological prediction models,better monitor weather behaviors,and enhance the application ability of artificial intelligence-based prediction models for forest fires.Third,establish prediction large models to ensure emergency safety.Fourth,embodied intelligent devices can be implanted into the prediction models for forest fires,collect and analyze forest fire factors,output the probability of forest fire occurrence,send back data and independently adjust action strategies.