The Improvement Of Ant Colony Optimization Algorithm And Its Application Research

With the development of science and technology,optimization theories and methods play an important role in the fields of transportation,social production and industrial design.In recent decades,a series of modern heuristic optimization methods that simulate natural processes have been proposed one after another,such as simulated annealing,genetic algorithms,and evolutionary planning.However,these classic optimization methods are difficult to deal with complex problems.In view of this,through research on the behavior of ant colonies,biologists have found that ants are not smart,but ant colonies are very smart.In the natural environment,ants must solve several problems to maintain the operation of a large society,such as maintaining a stable food supply and assigning their partners to different tasks.An ant cannot complete tasks alone,but collaborative work enables them to complete a larger range of tasks,and can complete overly complex tasks quickly and efficiently.This interesting phenomenon has aroused great research interest and enthusiasm,and proposed an ACO algorithm that simulates the process of ants searching for the shortest path for foraging.The ACO algorithm is first used to solve the traveling salesman problem.It uses distributed feedback parallel computing,is easy to integrate with other algorithms,and has strong robustness.It has a wide range of applications in the fields of path planning,data mining and combination optimization.Path planning is one of the important steps in robot navigation tasks.It means that the robot plans an optimal movement path from the starting point to the ending point according to some or some evaluation criteria in the running environment with obstacles.However,the classical ACO algorithm has some shortcomings in robot path planning.Such as the search time is long,the convergence speed is slow and local optimal phenomena are prone to occur.Outlier detection is one of the important research directions of data mining.Its purpose is to find out objects whose behavior is different from other objects in the data set.It has been widely used in intrusion detection,medical diagnosis and credit card fraud.However,the research of mixed attribute outlier detection based on ACO algorithm is still rare.In response to these problems,this paper mainly studies the improvement based on ACO algorithm and its application in robot path planning and mixed attribute outlier detection.Its main work is as follows:(1)Based on the classic ACO algorithm,this paper makes some improvements to its heuristic function and pheromone allocation mechanism,and applies the improved ACO algorithm to robot path planning using the grid method.Experimental results show that compared with the classic ACO algorithm,the improved algorithm has a shorter search time,faster convergence speed,and more precise solution results.(2)Aiming at the problem that the rough set method cannot effectively deal with numerical and mixed data,a mixed attribute outlier detection method based on neighborhood granules is constructed.The outlier degree and outlier factor are defined to characterize the outliers of neighborhood granules and objects,respectively,and related neighborhood granules based outlier detection(NGOD)algorithm is designed.(3)The ACO algorithm is further applied to the NGOD algorithm,and a mixed attribute outlier detection(ACO-NGOD)algorithm based on ACO algorithm is proposed.Experimental results show that these two methods have better adaptability and effectiveness than the existing outlier detection methods.It can be applied to three types of data,including categorical,numeric,and mixed attribute data.