• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>2011 water quality technology conference amp; exposition: An Oasis of Ideas for a Changing Environment. >Causes of False Alarms in Inferential Event Detection Systems for Distribution System Water Quality Monitoring
【24h】

Causes of False Alarms in Inferential Event Detection Systems for Distribution System Water Quality Monitoring

机译：配水系统水质监测推理事件检测系统中误报的原因

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

There is currently much interest in inferential event detection systems (IEDS) that monitor waterrnquality in distribution systems for deliberate or inadvertent contamination, with a number ofrncommercial systems now available. Unlike techniques such as liquid chromatography thatrnmeasure concentrations of specific compounds, an IEDS measures conventional water qualityrnparameters, i.e., free chlorine and pH, at a site in the distribution system in real time; organizesrnthe measurements into vectors that represent the recent dynamic behaviors of the parameters; andrnpattern-matches new incoming vectors against a database of previously collected and evaluatedrnpatterns. If a good match is found that connotes all is well, the IEDS simply continues tornmonitor. If a match indicates a problem, or there is no match at all, an alarm is generated with arnprotocol to be followed by utility personnel. The sensitivity of the pattern matching can bernadjusted to control how often the alarms are generated. An important issue with IEDS is that theyrncan generate unacceptable numbers of false alarms when their sensitivity is set too high;rnhowever, reducing the sensitivity could cause the rare real event to be ignored.rnThis paper describes research conducted under Water Research Foundation Project 4182 thatrnfocuses on ways to reduce false alarms in IEDS, with a particular focus on accounting for waterrnquality variability that is caused by normal plant and distribution system operations, and seasonalrnchanges and weather, e.g., raw water quality, process control, pumping, and tank cycling.rnBecause a contaminated slug of water could be in contact with a fixed-location sensor array forrnonly a few minutes or considerably less, for this project 20 minutes was established to be arnreasonable time limit for an "event" to manifest itself. The project involves analyses of multiyearrndata sets from more than 40 monitoring sites at five utilities. It has been found at a numberrnof monitoring sites that: water quality parameters largely vary independently of each other;rnindividually, water quality parameters vary non-deterministically, with "apparent" randomness;rnand operational parameters measured nearby generally poorly account for the water qualityrnvariability. The qualifier "apparent" acknowledges that observed behaviors are a consequence ofrnphysical laws, but often their causes are unmeasured or even completely unknown. Thesernfindings suggest that a stand-alone IEDS installation, which performs event detection using only local data, risks being inherently ineffective. An alternative approach is demonstrated thatrnperforms event detection by evaluating how water quality changes between two monitoring sites,rnwith an upstream site providing boundary conditions that substantially account for downstreamrnvariability that otherwise can only be attributed to "unmeasured disturbances".
机译：当前，对于监视故意或无意污染的分配系统中水质的推断事件检测系统（IEDS）引起了很大的兴趣，现在有许多商业系统可用。 IEDS与诸如液相色谱法测量特定化合物浓度的技术不同，IEDS实时测量分配系统中某个站点的常规水质参数，即游离氯和pH。将测量结果组织成代表参数最近动态行为的向量； andrnpattern将新传入的向量与先前收集和评估的rnpattern的数据库进行匹配。如果发现一个很好的匹配表示一切都很好，则IEDS只会继续监视。如果匹配项表明存在问题，或者根本不存在匹配项，则会生成带有arnprotocol的警报，实用程序人员将对其进行跟踪。可以调整模式匹配的灵敏度，以控制生成警报的频率。 IEDS的一个重要问题是，当灵敏度设置得太高时，它们可能会生成不可接受数量的错误警报；但是，降低灵敏度可能会忽略这种罕见的真实事件。本文介绍了根据水研究基金会项目4182进行的研究，减少IEDS中虚假警报的方法，特别着重考虑由正常的工厂和分配系统运行以及季节性变化和天气（例如原水水质，过程控制，抽水和水箱循环）引起的水质变化。被污染的团块可能会与固定位置的传感器阵列接触仅几分钟或更少的时间，为此，该项目被确定为20分钟，这是“事件”表现出来的合理时限。该项目涉及对来自五个公用事业公司的40多个监测站点的多年数据集进行分析。在许多监测点中发现：水质参数彼此之间有很大的不同；个体地，水质参数具有不确定的变化，具有“表观”随机性；在附近测量的运行参数通常不能很好地说明水质的可变性。限定词“明显”承认观察到的行为是物理定律的结果，但是通常它们的原因是无法衡量的，甚至是完全未知的。这些发现表明，仅使用本地数据执行事件检测的独立IEDS安装具有固有的无效性。通过评估两个监测站点之间水质的变化情况，证明了一种替代方法可以执行事件检测，而上游站点提供的边界条件基本上可以说明下游的可变性，否则只能归因于“无法测量的干扰”。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号