论文部分内容阅读
复杂庞大的供水管网系统拥有众多监测点,在人工判断的情况下,各个监测点采集的海量数据无法被及时有效地处理,数据准确性无从保障,这对供水管网异常情况的判断造成极大阻碍。针对此情况,将北京市某供水管网监测站56个月的在线监测数据进行时段上和季节上的切分,构建自回归滑动平均(autoregressive moving average,ARMA)模型,并通过该模型建立的置信区间识别人工模拟序列中的异常值,从而实现独立节点自身数据的自识别。结果表明:经过数据反馈矫正,该自识别过程能够准确提取人工模拟监测数据中的异常值。ARMA模型的建立极大限度压缩了需人工处理的数据量,以便在异常数据中人工甄选无效数据,实现数据质量控制。
The complex and large water supply pipe network system has a large number of monitoring points. Under the circumstance of human judgment, the massive data collected at each monitoring point can not be processed timely and effectively, and the data accuracy can not be guaranteed. Therefore, the judgment of abnormalities in the water supply network Great obstruction. In view of this situation, the online monitoring data of a 56-month monitoring station of a water supply network in Beijing were divided into periods and seasons, and an autoregressive moving average (ARMA) model was constructed and established by the model Confidence intervals identify outliers in artificial analogue sequences to achieve self-identification of the individual nodes’ own data. The results show that: after data feedback correction, the self-identification process can accurately extract the abnormal values in the artificial simulated monitoring data. The establishment of ARMA model greatly reduces the amount of data to be manually processed so as to manually select invalid data in abnormal data and achieve data quality control.