Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
6864126 | Neurocomputing | 2018 | 18 Pages |
Abstract
Leak source localization is a very important topic that has receivedmuch research attention in recent years. The “Request and Reset” strategy is introduced here into the GC-PSO and D-PSO algorithms to improve the localization strategy. In the “Request and Reset” process, some of the low fitness particles are requested to be removed from their current group with their positions reset to assist the globally best particle, with those particles then combined into an “optimal group” to enhance the search around the globally best particle. Other two existing algorithms are improved by modifying the learning factor and inertia weight as comparison. Serval experiments are conducted by simulation to investigate the feature of the proposed algorithms, in which the impact of environmental size and population size as well as error adaptability are considered. Experimental results demonstrated the feasibility and advantage of the proposed approaches. MGC-PSO has the superior performance to the other methods in aspect of success rate and iteration time.
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Yan Yuting, Zhang Ru, Wang Ji, Li Juming,