L-Town Water Distribution Network

该数据集是一个真实的水分销网络异常检测基准，包含了661个节点、766条连接以及29个最佳放置的压力传感器所在的区域A。数据集模拟了泄漏和传感器故障的场景，并采用了滑动窗口方法处理了包括单一泄漏和多种传感器故障在内的各种情况。该数据集的规模包括661个节点、766条连接和29个压力传感器，其任务是水分销网络中的异常检测。

This dataset is a real-world benchmark for water distribution network anomaly detection. It encompasses 661 nodes, 766 connections, and Zone A where 29 optimally deployed pressure sensors are situated. The dataset simulates scenarios involving pipe leaks and sensor faults, and adopts a sliding window approach to handle various cases including single leaks and multiple sensor failures. Featuring 661 nodes, 766 connections and 29 pressure sensors, its primary task is anomaly detection within water distribution networks.

搜集汇总

背景与挑战

背景概述

该数据集关注水分配网络中的在线学习问题，包含一个实现方法用于构建基于残差的异常检测系统，能够适应各种变化。数据集本身未公开发布，需联系作者获取。