Research And Application Of Ant Colony Algorithm In Continuous Space Optimization

Ant colony algorithm is inspired by the true nature of ants foraging behavior by the Italian scholar M.Dorigo, V.MahieZZo, A.Cororni by others in a new simulation optimization algorithm proposed in recent years. Prototype algorithm is ant colony foraging. The algorithm has many advantages, such as stability, simple structure, a positive feedback mechanism, the nature of parallelism, easily combined with other algorithms like. Algorithm was originally used to solve the traveling salesman problem, along with the deepening of the algorithm, ant colony algorithm can also be applied to solve the scheduling problem, assignment problem, NP-complete sequence order and other requirements. And has achieved good results. It shows ant colony algorithm to solve complex discrete optimization problems superiority.But the ant colony algorithm there are some flaws:the contradiction between diversity and algorithm search algorithm resulting solution convergence rate, the random motion of individual ants, although it will gradually evolved toward optimal path, but the larger the sample size, ants individual needs more time to find the optimal solution in many feasible path; if deliberately accelerate the convergence speed, it is possible to make the algorithm into local optima. Also in terms of the scope of application, ant colony optimization algorithm is difficult to deal with continuous space ant solution space because each individual iteration of the resulting solution is always limited, so get is discrete. Basic ant colony algorithm for solving discrete optimization problems has a strong capability, but it is difficult to directly apply for continuous space optimization problem.This paper describes the basic principles of ant colony algorithm and the current research, and to rely on the traveling salesman problem, ant colony algorithm describes the basic model and features of the basic ant colony algorithm thinking, analysis of ant colony algorithm ant search process nature. And in accordance with the basic principles of foraging ants, ants foraging write software for visual display of the entire process of ants foraging behavior.Then the deficiency of basic ant colony algorithm to solve the traveling salesman problem, quadratic assignment problems exist in discrete optimization problems:the search for a long time, it is possible to converge to a local optimal solution, proposed several improvements:join local optimization, pheromone evaporation mode change, change pheromone update rules. Try to balance between the global search ability of the algorithm convergence speed of contradictions ant colony. And different algorithms are analyzed and compared.This paper presents a continuous function can be applied to space optimization ant colony algorithm, the continuous function space domain space is divided into a grid, discrete ideas to solve the continuous space optimization problems. And introduced in the algorithm of ant colony algorithm pheromone global updating rule, the maximum and minimum ant colony algorithm pheromone restriction rules, as well as adaptive ant colony algorithm pheromone evaporation methods thought in order to improve the initial convergence speed algorithms and post global search capability. And the preparation of the ant colony algorithm function space optimization software.Continuous Ant Colony Algorithm in practical problems have been cause for concern. This article will continuously Colony Algorithm optimization capability function solution space in Settlement combination forecasting model parameter calculation and projection pursuit transformation parameters calculated two ways:First, combined with the establishment of a weighted combination of the traditional thinking model, add coefficients for each single model for settlement prediction predictions, take full account of each single model prediction accuracy at different times, so that each single model predictions with actual observations of formed between a function. With continuous ant colony algorithm in the function solution space optimization capabilities to solve the coefficients of a single model predictions, which in combination with sedimentation prediction model based on ant colony algorithm to improve the accuracy.Second, PPC Model on statistical indicators and analysis of multi-sample data, multivariate analysis will be converted into a data element analysis, multivariate data between the data and the unitary form the corresponding function, the function solution with continuous ant colony algorithm space optimization ability to convert PPC model is solved. Taking a forecast of annual runoff and acid rain pH value Observatory forecast to illustrate the continuous ant colony algorithm for solving the conversion parameters of the feasibility and effectiveness.