Two Adaptive Algorithms Based On The Law Of Universal Gravitation

In real life, we often encounter some global optimization problems, in order to solve the problem of global optimization,many algorithms have been presented. Most of the algorithms are the heuristic algorithms, most of which are built on simulating the biological group behavior or physics principle. The heuristic algorithms are widely applied to various fields. In recent years, it has appeared the algorithms based on the universal gravitation and Newton’s second law of kinematics algorithm. Such as the Central Force Optimization Algorithm, Gravitational Search Algorithm and Search Charge System Algorithm. In the thesis we study the central force optimization algorithm and gravitational search algorithm. We do some research given as follows:1、We research on the improvement of center force optimization algorithm. On the basis of the basic central force optimization algorithm, we introduce the velocity updating formula and weight, and then analyze the stability of the algorithm. We determine the parameters setting range of the weight and gravitational constant through the stability analysis. On this basis, we present an adaptive central force optimization algorithm. In order to verify the validity of the algorithm, we select 23 test functions to do numerical experiments and compare with other algorithms. The numerical results show that the ACFO algorithm achieves good convergence rate.2、We research on the gravitational search algorithm. Firstly, we convert the velocity updating formula and location updating formula of gravitational search algorithm into a second order differential equation. Secondly we analyze the second order difference equations to determine the convergence region of the parameters. According to the parameters setting area, we put forward an adaptive gravitational search algorithm. At last, through the numerical experiment of 23 test functions, we verify the validity of the algorithm. The experiment results show that the algorithm has faster convergence rate.