Font Size: a A A

Improved Bacterial Foraging Optimization Algorithm And Its Application

Posted on:2015-02-02Degree:MasterType:Thesis
Country:ChinaCandidate:T W CaoFull Text:PDF
GTID:2208330434451429Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
The swarm intelligence optimization algorithms were proposed to solve variety of real-life optimization problems through analyzing its highly gregarious self-organization and constructing mathematics model. For large-scale optimization problems, swarm intelligence algorithms are better than the traditional mathematical programming methods in computing time and complexity significantly. Swarm intelligence algorithms gradually become a research focus in the field of optimization because they can get the approximate optimal solution efficiently. At present, the applications of swarm intelligence algorithm are mainly about genetic algorithm, ant colony algorithm and particle swarm algorithm. These algorithms have made significant achievements in various fields of engineering practice. Bacterial foraging optimization (BFO) algorithm is a novel smart optimization algorithm developed in recent years. BFO algorithm has advantages of constructed intuitive, easy to understand, not sensitive to initial conditions, strong stability, and the outstanding ability of local search, but it also has disadvantages of slow convergence, lack of global searching capability, prematurity and other shortcomings. With respect to the several mature intelligent optimization algorithms are concerned, BFO algorithm are slightly less from the theoretical depth and breadth of practical applications. BFO algorithm is significant in potentiality for further excavation, analysis and improvement. This paper has analyzed and improved weakness of BFO algorithm, and then applied to simulation experiments. The main works are as follows:(1) To overcome slow convergence and low accuracy in the standard BFO algorithm, this paper proposed a dynamic maximum swimming steps BFO (DMSS-BFO) algorithm based on "classic effect rate" which comes from psychologist Edward· Thorndike to improve the operation of chemotaxis. The "classic effect rate" contains three principles of "strengthening, punishment and subsided". The new operation includes three types of individual bacteria:elite, ordinary, and mediocre. Elite individuals has an increase of the maximum number of steps swimming which gives it more opportunity to explore; mediocrity individuals has an reduce of the maximum number of steps swimming which gives it less opportunity to explore; ordinary individuals continue to use the initialized maximum number of swimming steps. Testing on function optimization problems shows that DMSS-BFO algorithm has fast convergence and high precision. The performance of DMSS-BFO algorithm was better than the standard BFO algorithm both in unimodal and multimodal function optimization problems.(2) To improve the global search capability of the standard BFO algorithm, a CS-BFO algorithm was proposed based on Levy flight model. That is, the elite individuals of the population is reserved, and then all non-elite individuals on the basis of Levy flight is updated. Elimination and dispersal operate was improved and incorporated into the tendency operate, which changing the original cycle of three layers to two and simplifying the algorithm structure. Several classic benchmark function optimization tests showed that:CS-BFO algorithm’s convergence speed and global search capability have been significantly improved, and the algorithm run faster due to its simple structure; CS-BFO algorithm can find a global optimal solution efficiently, and its overall performance is better than standard BFO and DSN-BFO algorithm.(3) In order to investigate the performance of BFO algorithm for combinatorial optimization, the improved DMSS-BFO and CS-BFO algorithm were applied to solve0-1knapsack problem. Using0-1binary coding in BFO algorithm according to the characteristics of0-1knapsack problem solution, each individual bacterium generates a random coding region as a tendency operating region, in which bacteria individuals’ elements changes randomly. If the fitness value is better than the target bacteria after tendency operation, the coding region is recorded and retained for the next tendency operation until the area no longer generates better solution by randomly changing. The experimental results of the improved algorithms show high efficiency and accuracy for solving0-1knapsack problem compared with other intelligent algorithms.
Keywords/Search Tags:intelligent optimization, bacterial foraging optimization algorithm, dynamic maximum swimming steps, Levy flight, 0-1Backpack
PDF Full Text Request
Related items