基于机器学习的碳排放数据异常识别方法
Machine Learning Based Anomaly Recognition Method for Carbon Emission Data
杨若钧 1陈爱忠 2曾金灿 3姚腾1
作者信息
- 1. 国家环境保护环境影响评价数值模拟重点实验室,北京 100041
- 2. 国家环境保护环境影响评价数值模拟重点实验室,北京 100041;生态环境部环境工程评估中心,北京 100012
- 3. 南方电网能源发展研究院有限责任公司,广东广州 510700
- 折叠
摘要
提出了一种基于机器学习的碳排放数据异常的识别方法,采用了密度聚类、统计方法和机器学习相结合的策略.通过引入统计方法与机器学习方法检测异常的双重判据,该方法能有效检测异常数据.针对碳排放分析多维数据中参数间相关性难以明确的问题,采用机器学习技术研判高维异常数据中存在的复杂多变量关系,并深入分析了异常产生的原因.通过该方法提升数据质量,更精准、有效地开展碳核查,以满足碳排放量准确计量的要求.
Abstract
A machine learning based method for identifying anomalies in carbon emission data was proposed,using a combination of density clustering,statistical methods,and machine learning strategies.By introducing a dual criterion of statistical methods and machine learning methods for detecting anomalies,this method can effectively detect abnormal data.In response to the problem of unclear correlation between parameters in multidimensional data in carbon emission analysis,machine learning technology is used to analyze the complex multivariate relationships in high-dimensional abnormal data,and the reasons for the occurrence of anomalies are analyzed in depth.By using this method to improve data quality,carbon verification can be carried out more accurately and effectively to meet the requirements of accurate measurement of carbon emissions.
关键词
碳排放/机器学习/密度聚类/碳核查Key words
carbon emissions/machine learning/density clustering/carbon verification引用本文复制引用
基金项目
中国南方电网有限责任公司管理创新项目(ZBKJXM20210437)
出版年
2024
NETL
NSTL
万方数据