Optimal Allocation of Water Resources Based on the Multi-Objective Fish-Ant Colony Algorithm
基于多目标鱼群-蚁群算法的水资源优化配置

To resolve complex problems on optimal allocation of water resources with intelligent optimal methods, a multi-objective optimization model was built and the multi-objective fish-ant colony algorithm (MFACA) was designed. This model, based on principles of efficiency, fairness, and harmoniousness, is aimed at producing the largest economic, social, and environmental benefits. The objective of economic benefit is the largest direct economic benefit produced by regional water supply. The objective of social benefit is referred to as the smallest regional water deficit. The objective of environmental benefit is to ensure the smallest discharge of major contaminants. Constraints included water supply, water demand, water settings, economic development, and its harmony. In this model, constraints of water supply include possible water yield and ground water yield. Constraints of water demand include living, industrial, agricultural, and environmental water. Constraints of water settings include overall merit index and water quality. The optimal allocation model had the characteristics of large-scale system, multiple objectives, multiple constraints, multiple levels, and multiple associations. To solve this complicated model, the multi-objective fish-ant colony algorithm was established in accordance with the integration of pheromone positive feedback of the ant colony optimization (ACO) and fast track change and jumping out of local extremum of the artificial fish-swarm algorithm (AFA). A swarm degree in the AFA was used to avoid possible premature problems at the initial stage of ACO. It was not strict for MFACA to set parameters and initial values of a mathematical model. The objective functions and constraints were not necessarily continuous and differentiable. This algorithm has a faster convergence rate and a higher optimization power. In order to validate the feasibility and effectiveness of the MFACA, surveys were done in Zhenping County, Henan Province, China. Data of water resources and other relevant socioeconomic information were obtained and input into a database. Water yield with different planning years and different guaranteed rates was optimized. The largest economic, social, and environmental benefits were effectively calculated with the model and the MFACA. To compare different effects and convergences amongst MFACA, AFA, and ACO, emulations were demonstrated after the same initial parameters of the model were input into MFACA, AFA and ACO, respectively. Optimal solutions were obtained at about the 89th iteration for MFACA, about the 130th iteration for AFA, and about the 110th iteration for ACO in the best condition. The MFACA showed the best result and the fastest convergence rate amongst the three algorithms. It has been shown to be a promising tool for optimal allocation of water resources.