基于大数据多模式互验的水资源计量数据异常检测方法

A Method for Abnormal Detection in Water Resource Measurement Data Based on Big Data Multimode Mutual Verification

摘要: 水资源计量过程环节多，异常数据难于避免，开展数据异常检测是水资源计量管理重要内容。模型算法是快速自动发现水资源计量异常数据的核心技术，大数据挖掘技术已有较多算法在数据异常检测中应用，对模型算法有效利用是准确发现异常数据的工作基础。本异常检测方法是基于统计、分类、预测等3种大数据模型算法进行数据异常检测，研究完整设计了计算处理流程，并应用软件技术完成系统开发，实现计算参数自定义，采用消息控制多模式计算，有效实现交互验证处理；经实测数据应用检验，实验分析表明，该检测方法能有效提升异常发现准确率、提高修正的便利性。

Abstract: The process of water resource measurement involves multiple stages, and abnormal data is difficult to avoid. Conducting data anomaly detection is an important part of water resource measurement management. Model algorithms are the core technology for quickly and automatically discovering abnormal data. Big data mining technology has been widely applied in data anomaly detection, and effective utilization of model algorithms is the foundation for accurately discovering abnormal data. Based on three big data model algorithms, including statistics, classification, and prediction, research was conducted, and a calculation processing flow was designed. Software technology was applied to complete system development, achieving custom calculation parameters and using message control for multi-mode calculation and effectively achieving interactive verification processing; Applying measured data for application calculations, experimental analysis shows that this detection method can effectively improve the accuracy of anomaly detection and the level of intelligence for correction.