一种新的水务信息系统数异常检测算法

摘要

Water information and data abnormal diagnostic is of great significance to the national life. Traditional detection algorithms of water anomaly information system data are mostly modeled with multivariate regression analysis. But this method has some disadvantages. Pro-poses a new method determining whether the data nomal by the nature of the hat matrix diagonal elements. It uses MatLab to perform this algorithm. The algorithm solves the disadvantages of the former and can perform real-time monitoring of water system, accurately i-dentify abnormal data.%水务信息数据异常诊断对国民生计有着举足轻重的意义。传统的水务信息系统数据常检测算法多是采用多元回归分析法，但该方法有缺陷。提出新算法利用帽子矩阵对角线元素的性质，判断数据是否异常，并用MatLab进行算法实现，该算法解决前者的缺点，可对水务系统进行实时监测，准确找出异常数据。