School of Computer Science, China University of Geosciences, Wuhan, Hubei, 430074, China.
Environ Sci Pollut Res Int. 2019 Jun;26(18):17901-17910. doi: 10.1007/s11356-017-0516-y. Epub 2017 Oct 27.
Urban water supply networks are susceptible to intentional, accidental chemical, and biological pollution, which pose a threat to the health of consumers. In recent years, drinking-water pollution incidents have occurred frequently, seriously endangering social stability and security. The real-time monitoring for water quality can be effectively implemented by placing sensors in the water supply network. However, locating the source of pollution through the data detection obtained by water quality sensors is a challenging problem. The difficulty lies in the limited number of sensors, large number of water supply network nodes, and dynamic user demand for water, which leads the pollution source localization problem to an uncertainty, large-scale, and dynamic optimization problem. In this paper, we mainly study the dynamics of the pollution source localization problem. Previous studies of pollution source localization assume that hydraulic inputs (e.g., water demand of consumers) are known. However, because of the inherent variability of urban water demand, the problem is essentially a fluctuating dynamic problem of consumer's water demand. In this paper, the water demand is considered to be stochastic in nature and can be described using Gaussian model or autoregressive model. On this basis, an optimization algorithm is proposed based on these two dynamic water demand change models to locate the pollution source. The objective of the proposed algorithm is to find the locations and concentrations of pollution sources that meet the minimum between the analogue and detection values of the sensor. Simulation experiments were conducted using two different sizes of urban water supply network data, and the experimental results were compared with those of the standard genetic algorithm.
城市供水管网容易受到人为、意外的化学和生物污染，这对消费者的健康构成威胁。近年来，饮用水污染事件频繁发生，严重威胁社会稳定和安全。通过在供水管网中放置传感器，可以有效地实现水质的实时监测。然而，通过水质传感器获得的数据检测来定位污染源是一个具有挑战性的问题。其难点在于传感器数量有限、供水管网节点数量众多以及用户对水的动态需求，这使得污染源定位问题成为一个不确定性、大规模和动态优化问题。本文主要研究污染源定位问题的动态性。先前的污染源定位研究假设水力输入（例如，消费者的水需求）是已知的。然而，由于城市水需求的固有可变性，该问题本质上是消费者水需求的波动动态问题。在本文中，将水需求视为具有随机性，并可以使用高斯模型或自回归模型进行描述。在此基础上，提出了一种基于这两种动态水需求变化模型的优化算法，用于定位污染源。该算法的目标是找到污染源的位置和浓度，以满足传感器模拟值和检测值之间的最小值。使用两种不同规模的城市供水管网数据进行了仿真实验，并将实验结果与标准遗传算法进行了比较。
