机器学习算法在森林火灾风险预测中的有效性比较研究
南京大学, 信息管理学院, 江苏南京 210023
| 摘 要： | 森林火灾作为一项全球性的环境问题，对生态系统、经济发展和社会安全构成严重威胁。准确预测森林火灾对于制定有效的预防措施和减少火灾带来的损失至关重要。随着机器学习技术的发展，其在森林火灾风险预测领域中的应用逐渐成为研究热点。鉴于不同的机器学习算法具有不同的数据处理能力和预测精度，比较和选择最适宜的算法对提高预测模型的性能显得尤为关键。本研究通过分析1990至2019年广西省的气象数据和相应的火灾等级信息，选取了10种机器学习算法进行实验。在数据预处理阶段，严格处理了缺失值和异常值，以确保数据质量可靠。随后，利用准确率、精确率、召回率、F1分数以及ROC曲线等多项指标，全面评估各算法的预测性能。为进一步检验模型的稳定性和鲁棒性，并确保所选模型具有更强的泛化能力，研究实施了5折交叉验证。结果显示，SVM、贝叶斯、BP神经网络、逻辑回归、AdaBoost、Gradient Boost和XGBoost在AUC上表现较佳，而KNN和随机森林在精确度和准确率上具有优势。研究指出，结合多种算法能够进一步提高预测的准确性和可靠性，未来研究可考虑纳入更多影响因素，并探索采用深度学习技术来进一步提升预测性能，为森林火灾的预防和应对提供更加科学、有效的支持。 |
| 关 键 词： | 森林火灾; 机器学习; 数据预处理; 预测模型 |
| DOI： | 10.57237/j.jaf.2024.02.002 |
View:
Downloads:
收稿日期: 2024-04-01
接受日期: 2024-05-06
在线出版日期: 2024-05-23
The Effectiveness of Methods Designed to Predict Forest Fire Risk: A Comparison of Machine Learning Algorithms
School of Information Management, Nanjing University, Nanjing 210023, China
| Abstract: | Forest fires represent a global environmental issue, posing severe threats to ecosystems, economic development, and social security. Accurate prediction of forest fires is crucial for formulating effective preventive measures and minimizing the associated losses. With the advancement of machine learning technologies, their application in forest fire risk prediction has become an emerging research focus. Diverse machine learning algorithms exhibit varying data processing capabilities and predictive accuracies; hence, comparing and selecting the most suitable algorithms is significant for enhancing the performance of predictive models. This study analyzed meteorological data and corresponding fire severity information from Guangxi Province between 1990 and 2019, employing 10 machine learning algorithms in experiments. Initial data preprocessing, including handling of missing and outlier values, ensured data quality. Subsequently, predictive performance across algorithms was assessed using accuracy, precision, recall, F1 score, and the Receiver Operating Characteristic (ROC) curve. To further examine the stability and robustness of the models, a 5-fold cross-validation was implemented. Results indicated that SVM, Bayesian classifiers, BP neural networks, logistic regression, AdaBoost, Gradient Boost, and XGBoost demonstrated superior performance in terms of AUC, while KNN and Random Forest algorithms showed advantages in precision and accuracy. The 5-fold cross-validation confirmed the stability and robustness of the models, revealing that most models maintained stable predictive performance across different datasets. The study suggests that integrating multiple algorithms can improve the accuracy and reliability of predictions and recommends that future research consider additional influencing factors and employ deep learning techniques to further enhance predictive performance. |
| Keywords: | Forest Fires; Machine Learning; Data Preprocessing; Predictive Models |
参考文献
| [1] | Edwards R B, Naylor R L, Higgins M M, et al. Causes of Indonesia’s forest fires [J]. World Development, 2020, 127: 104717. |
| [2] | Pourghasemi H R, Gayen A, Lasaponara R, et al. Application of learning vector quantization and different machine learning techniques to assessing forest fire influence factors and spatial modelling [J]. Environmental research, 2020, 184: 109321. |
| [3] | Ying L, Cheng H, Shen Z, et al. Relative humidity and agricultural activities dominate wildfire ignitions in Yunnan, Southwest China: Patterns, thresholds, and implications [J]. Agricultural and Forest Meteorology, 2021, 307: 108540. |
| [4] | Zheng Z, Gao Y, Yang Q, et al. Predicting Forest fire risk based on mining rules with ant-miner algorithm in cloud-rich areas [J]. Ecological Indicators, 2020, 118: 106772. |
| [5] | Alley R B, Emanuel K A, Zhang F. Advances in weather prediction [J]. Science, 2019, 363(6425): 342-344. |
| [6] | Ma W, Feng Z, Cheng Z, et al. Identifying Forest fire driving factors and related impacts in china using random forest algorithm [J]. Forests, 2020, 11(5): 507. |
| [7] | 朱馨, 李建微, 郭伟, 毕胜, 伍跃飞. 基于机器学习的森林火险预测模型 [J]. 中国安全科学学报, 2022, 32(09): 152-157. |
| [8] | 高超, 林红蕾, 胡海清, 宋红. 基于气象因子的黑龙江黑河林火发生概率预测 [J]. 森林与环境学报, 2022, 42(05): 529-535. |
| [9] | Huesca M, Litago J, Palacios-Orueta A, et al. Assessment of forest fire seasonality using MODIS fire potential: A time series approach [J]. Agricultural and Forest Meteorology, 2009, 149(11): 1946-1955. |
| [10] | 高博, 陈响, 单仔赫, 韩喜越, 单延龙, 尹赛男, 于渤. 基于Logistic回归模型的大兴安岭地区林火发生概率预测研究 [J]. 中国安全生产科学技术, 2022, 18(11): 163-168. |
| [11] | 王丹丹, 黄家荣, 刘伟, 孟庆玲, 田鹏飞. 基于人工神经网络的森林火险预测 [J]. 西北林学院学报, 2010, 25(03): 143-146. |
| [12] | Xie L, Zhang R, Zhan J, et al. Wildfire risk assessment in Liangshan Prefecture, China based on an integration machine learning algorithm [J]. Remote Sensing, 2022, 14(18): 4592. |
| [13] | Pastor E, Zárate L, Planas E, et al. Mathematical models and calculation systems for the study of wildland fire behaviour [J]. Progress in Energy and Combustion Science, 2003, 29(2): 139-153. |
| [14] | Pai M L, Varsha K S, Arya R. Application of Artificial Neural Networks and Genetic Algorithm for the Prediction of Forest Fire Danger in Kerala [C] // Intelligent Systems Design and Applications: 18th International Conference on Intelligent Systems Design and Applications (ISDA 2018) held in Vellore, India, December 6-8, 2018, Volume 2. Springer International Publishing, 2020: 935-942. |
| [15] | Lin H, Liu X, Wang X, et al. A fuzzy inference and big data analysis algorithm for the prediction of forest fire based on rechargeable wireless sensor networks [J]. Sustainable Computing: Informatics and Systems, 2018, 18: 101-111. |
| [16] | 姚艳霞, 苗双喜, 黄旭, 张波. 耦合Rothermel模型与粒子系统的林火蔓延模拟方法 [J]. 测绘通报, 2022(08): 75-80. |
| [17] | 张文文, 王劲, 王秋华, 曹恒茂, 王金波, 左军宏, 王加庆. 森林火灾遥感探测技术研究进展 [J]. 西北林学院学报, 2023(01): 123-130. |
| [18] | Chuvieco E, Aguado I, Salas J, et al. Satellite remote sensing contributions to wildland fire science and management [J]. Current Forestry Reports, 2020, 6: 81-96. |
| [19] | 高超, 林红蕾, 胡海清, 宋红. 我国林火发生预测模型研究进展 [J]. 应用生态学报, 2020, 31(09): 3227-3240. |
| [20] | Karo I M K, Amalia S N, Septiana D. Wildfires Classification Using Feature Selection with K-NN, Naïve Bayes, and ID3 Algorithms [J]. Journal of Software Engineering, Information and Communication Technology (SEICT), 3(1): 15-24. |
| [21] | Jaafari A, Zenner E K, Pham B T. Wildfire spatial pattern analysis in the Zagros Mountains, Iran: A comparative study of decision tree based classifiers [J]. Ecological informatics, 2018, 43: 200-211. |
| [22] | Jaafari A, Pourghasemi H R. Factors influencing regional-scale wildfire probability in Iran: an application of random forest and support vector machine M] // Spatial modeling in GIS and R for Earth and environmental sciences. Elsevier, 2019: 607-619. |
| [23] | 黄琼, 司颖, 王浩宇. 基于加权贝叶斯网络的森林火灾风险评估 [J]. 消防科学与技术, 2021, 40(11): 1671-1676. |
| [24] | Sirat E F, Setiawan B D, Ramdani F. Comparative analysis of K-means and isodata algorithms for clustering of fire point data in Sumatra region[C]//2018 4th International Symposium on Geoinformatics (ISyG). IEEE, 2018: 1-6. |
| [25] | WU C, ZHONG Y, LIU H, et al. A Novel Application of BP Neural Networks to Evaluate the Safety of Power Grid in Wildfire Disasters [J]. International Journal of Simulation: Systems, Science and Technology, 2016, 17(27): 32.1-32.5. |
| [26] | Papagiannaki K, Giannaros T M, Lykoudis S, et al. Weather-related thresholds for wildfire danger in a Mediterranean region: The case of Greece [J]. Agricultural and Forest Meteorology, 2020, 291: 108076. |
| [27] | Demin G, Haifeng L, Anna J, et al. A forest fire prediction system based on rechargeable wireless sensor networks [C] // 2014 4th IEEE International Conference on Network Infrastructure and Digital Content. IEEE, 2014: 405-408. |
| [28] | 高德民, 林海峰, 刘云飞, 吴国新. 基于无线传感网的森林火灾FWI系统分析 [J]. 林业科技开发 2015, 29(01): 105-109. |
| [29] | Vikram R, Sinha D, De D, et al. PAFF: predictive analytics on forest fire using compressed sensing based localized Ad Hoc wireless sensor networks [J]. Journal of Ambient Intelligence and Humanized Computing, 2021, 12: 1647-1665. |
