|
| 污染源自动监测异常数据识别规则及处理方法探索 |
| Exploration on Identification Rules and Processing Methods of Abnormal Data in Automatic Monitoring of Pollution Sources |
| |
| DOI： |
| 中文关键词: 污染源 自动监测 异常数据 识别规则 自动标志 模型训练 |
| 英文关键词: Pollution source Automatic monitoring Abnormal data Identification rule Automatic identification Model training |
| 基金项目:国家自然科学面上基金资助项目（51979136）；广东省普通高校特色创新（自然科学类）基金资助项目（2018KTSCX201）；深圳市科创委技术攻关基金资助项目（JSGG20180508151852303） |
|
| 摘要点击次数: 410 |
| 全文下载次数: 41 |
| 中文摘要: |
| 基于大量污染源自动监测数据的特征分析与异常原因解析，探索建立针对自动监测异常数据的识别规则与标志处理方法，并通过模型训练实现了异常数据的自动标志。经实例验证，该方法可识别异常偏高、异常偏低、异常为0、迟滞不变、逻辑错误等5种类型的异常数据，按照数据有效性及异常原因进行标志处理，可以为后续数据分析及各类模型训练提供数据基础和保障。 |
| 英文摘要: |
| Based on characteristic analysis and abnormal cause analysis of a large number of automatic monitoring data of pollution sources, the identification rules and identification processing methods for automatic monitoring abnormal data were explored and established, and the automatic identification of abnormal data was realized through model training. Examples showed that this method could identify five types of abnormal data, including abnormally high, abnormally low, abnormally zero, hysteresis invariance and logic error. According to data validity and abnormal reasons, this identification and processing method could provide data base and guarantee for subsequent data analysis and various model training. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
