|
| 机器学习在农田土壤污染识别、修复和风险评估中的应用研究进展 |
| Application of machine learning in identification，remediation，and risk assessment of farmland soil pollution: a review |
| Received:October 07, 2024 |
| DOI：10.13254/j.jare.2024.0776 |
| 中文关键词: 机器学习，农田土壤污染，污染识别，材料筛选，修复机理，风险评估 |
| 英文关键词: machine learning, farmland soil pollution, pollution identification, material screening, remediation mechanism, risk assessment |
| 基金项目:国家重点研发计划项目（2023YFC3709703）；国家自然科学基金项目（42277012）；广东省自然科学基金项目（2022A1515011031） |
| Author Name | Affiliation | E-mail | | LI Xingzhen | College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China | | | LIN Hansen | College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China | | | QIU Shaojian | College of Mathematics and Information, South China Agricultural University, Guangzhou 510642, China | | | LIN Qingqi | College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China
Guangdong Provincial Key Laboratory of Agricultural and Rural Pollution Control and Environmental Safety, Guangzhou 510642, China
Guangdong Laboratory for Lingnan Modern Agriculture, Guangzhou 510642, China | linqingqi@scau.edu.cn | | YE Long | Guangdong Provincial Academy of Building Research Group Co., Ltd., Guangzhou 510510, China | | | MAI Yuebang | Guangdong Provincial Academy of Building Research Group Co., Ltd., Guangzhou 510510, China | | | WU Peihao | Guangdong Provincial Academy of Building Research Group Co., Ltd., Guangzhou 510510, China | | | NI Zhuobiao | College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China
Guangdong Provincial Key Laboratory of Agricultural and Rural Pollution Control and Environmental Safety, Guangzhou 510642, China
Guangdong Laboratory for Lingnan Modern Agriculture, Guangzhou 510642, China | | | QIU Rongliang | College of Natural Resources and Environment, South China Agricultural University, Guangzhou 510642, China
Guangdong Provincial Key Laboratory of Agricultural and Rural Pollution Control and Environmental Safety, Guangzhou 510642, China
Guangdong Laboratory for Lingnan Modern Agriculture, Guangzhou 510642, China
School of Environmental Science and Engineering, Sun Yat-sen University, Guangzhou 510006, China | |
|
| Hits: 2755 |
| Download times: 3919 |
| 中文摘要: |
| 农田作为国家的保护土地资源，与农业生产、食品安全、人体健康有着密不可分的关系。因此，研究农田土壤污染状况对确保粮食质量、保护农田资源具有重大意义。传统的农田土壤污染研究多针对单一场景、污染物或实验条件，难以应对日益复杂的环境问题。随着大数据时代的来临，机器学习逐渐在农田土壤环境保护领域中得到广泛应用，并在土壤污染识别、修复等方面的研究中展现其客观、准确、擅长处理复杂任务的优势。本文介绍了常用的机器学习流程、方式、算法和模型性能评价指标；通过对Web of Science以及中国知网数据库中2011—2023年间相关领域文献进行统计分析，从农田土壤污染的识别、修复材料的筛选与机理研究、生态风险评估三方面综述机器学习在这些领域研究中的应用，分析了其优势和局限性。最后，本文从提升数据共享、增强模型可解释性以及应用迁移学习等新手段提高模型性能等方面进行了展望。 |
| 英文摘要: |
| Farmland, as a protected land resource of the state, is inextricably related to agricultural production, food safety, and human health. Consequently, investigating the soil pollution status of farmland is of significant importance for ensuring food quality and safeguarding farmland resources. Traditional research on farmland soil pollution have primarily focused on individual scenarios, specific pollutants, or singular experimental conditions, which complicates the resolution of increasingly complex environmental issues. With the advent of the big data era, machine learning has been increasingly applied in the field of farmland soil environmental protection, demonstrating its advantages in objectivity, accuracy, and proficiency in addressing complex tasks related to soil pollution identification and remediation. This paper outlined the commonly used machine learning processes, methods, algorithms, and performance evaluation indicators for models. Through a statistical analysis on relevant literatures from the Web of Science and China National Knowledge Infrastructure（CNKI）database covering the period from 2011 to 2023, it reviewed the application of machine learning in the field across three aspects：the identification of farmland soil pollution, the screening and mechanism research of remediation materials, and ecological risk assessment, while also discussed its advantages and limitations. Finally, the paper anticipated future developments in areas such as enhancing data sharing, increasing model interpretability, and applying novel methods like transfer learning to improve model performance. |
| HTML
View Full Text
View/Add Comment Download reader |
| Close |
|
|
|
