Research on Multimodal Optimization Algorithm for the Contamination Source Identification of City Water Distribution Networks

Abstract

In recent years, drinking water contamination happens from time to time and causes severe damage to social stability and safety. Setting the sensor in the town water distribution networks can dramatically decrease the occurrence of contamination events by real-time monitoring on water quality. However, how to make a reverse localization on contamination source by the detection information of water quality sensor is a challenging issue. The difficulty is that the limited sensor amounts, large-scale nodes in town distribution networks and changing water demands from users lead to the uncertainty of the optimal problem. In this paper, we mainly study the uncertainty issue of the Contamination Source Identification(CSI) problem. In the previous studies, simulation-optimization model has been utilized for the conversion from CSI problem to the unimodal function optimization problem in many documents. But it is a multimodal function optimization problem in essence and the number of its solution has non-uniqueness. This paper uses dynamic niching genetic algorithm and can calculate multiple contamination sources through one operation, which provides the possibility for screening the true contamination source. Furthermore, this paper has a try and verifies the validity after the threshold formulation as well as the effectiveness of algorithm.