The Multi-Object Ant-Genetic Algorithm And Its Application In Regional Water Resource Allocation

Multi-objective optimization is widely used in many areas, such as economic, management, military affairs, human affairs, and so on. The traditional multi-objective optimization algorithms often can only obtain some local optimal Pareto Front instead of global one. Now some intelligent-colony algorithms based on biology or physical phenomenon become the frontiers of Multi-objective optimization research, for they can not only have a high speed of searching the global optimal solutions, but also have generality and can be easily applied to other fields.Genetic Algorithm and Ant Colony Algorithm are two kinds of mature and widely used intelligent optimization algorithms. Genetic Algorithm can have better global search performance independent with problems in many wide scopes. Based on the global search and population evolution, the algorithms have many good characteristics, such as potential parallel property, multi-value comparability, strong robust and so on. On the other hand, Ant Colony Algorithm has not only features such as intelligent search and global optimization, but also the characteristics such as robust, positive feedback, parallel distributed computation and combining with other algorithm easily.Note that the Genetic Algorithm has the advantage of speediness, randomness and global convergence, and its disadvantage in redundant iteration, precocity, sensory to parameters. Meanwhile, the Ant Colony Algorithm possess the merit of parallelism, positive feedback mechanism and high proficiency of resolution, but it has a poor performance on global searching and often leads to local solutions. Considering all of these, we propose a new algorithm---Multi-Objective Ant Genetic Algorithm, which is based on the fusion of Genetic Algorithm and Ant Colony Algorithm, to solve the problems of multi-objective optimization with constraint conditions. The algorithm introduces the information feedback mechanism of Genetic Algorithm to improve the computing rates in Pareto compositor, in another words, pheromone is used to guide the search and choose of Multi-Objective Genetic Algorithm. The most important is that this new algorithm can not only guarantee to get the high-accuracy solution, but also reduce the complexity of computation.In this algorithm, firstly, the solution space is divided into some subspaces, and all the subspaces are labeled by pheromone, then the pheromone guides inheritance searching and updates itself. At the same time, strategy of updating the Pareto optimal decisions, scheme of converging and exiting the searching are used to promote the efficiency and reduce the complex degree of algorithm. The experiments shows that compared with other algorithms, this new algorithm can approximate Pareto faster and more accurately.In the application of Multi-Objective Ant-Genetic Algorithm, we analysised the regional water resources optimal allocation model and designed a proper scheme of resolution for a real decision-making problem in the last part of the paper. Our proposed algorithm helped us to make proper decisions, which satisfied the practical demand.