Study On Firefly Algorithm And Its Application In Path Planning Of Underwater Vehicles

Firefly algorithm which is derived from the simplification and simulation of groupbehavior of fireflies is a novel intelligent optimization algorithm. In Firefly Algorithm,fireflies with larger brightness attract those whose brightness is smaller to move to them inorder to guide the firefly population continues to move to a better, thus completing theoptimization procedure. It has a simple concept, few parameters to be adjusted and it’s easy toimplement.This thesis studies the standard Firefly Algorithm and proposes a Modified FireflyAlgorithm for its deficiencies. In addition, we propose a multi-objective Firefly Algorithmand use it to solve the path planning problem of underwater vehicle.First, this thesis does a deep study of the standard Firefly Algorithm and elaborates theprinciples of bionics, the basic idea, the mathematical description and the process of thealgorithm. In order to understand Firefly Algorithm deeply and to find out its similarities anddifferences with other algorithms in the intelligent optimization area, we select two classicalgorithms in this field, that is Genetic Algorithm and Particle Swarm Optimization.Second, according to the simulation experiments, we analyze the disadvantage of thestandard Firefly Algorithm that its convergence speed is slow in the larger search rangeconditions. We introduce the distance parameter between the fireflies into the firefly locationupdating formula to propose a Modified Firefly Algorithm. The results of simulationexperiments show that the new algorithm can effectively overcome the weakness that thestandard Firefly Algorithm is sensitive to search ranges and improve the convergence speedand the search accuracy of the original algorithm.Third, through taking into account the characteristics of multi-objective optimizationproblem and combining with the unique optimization mechanism of Firefly Algorithm, wepropose the Multi-objective Firefly Algorithm. This algorithm introduces the concept ofPareto dominance to distinguish the brightness of the fireflies, uses an external file to save thePareto optimal solutions generated in the iterative process of the algorithm, and maintains theexternal file with an adaptive grid method. Simulation results for test functions inmulti-objective optimization problem show that the algorithm is able to get Pareto optimalsolutions with nice approximation and uniformity.Finally, Multi-objective Firefly Algorithm is applied to the underwater vehicle pathplanning problem. We determine methods to describe the environment of underwater vehicle path planning problem and its cost functions and conducts simulation experiments in threekinds of specific path planning environment. Experimental results show that runningMulti-objective Firefly Algorithm once, we can get a group of Pareto optimal paths, whichsatisfies the requirements of path planning.