An Improved Genetic Algorithm for Nonlinear Programming Problems

In this paper, an improved genetic algorithm is proposed for nonlinear programming problems. In this algorithm, each individual is taken as a particle with mass. The mass of each individual is defined and the center of gravity is computed according to physical formula. A simplex is formed randomly from the population. One of the points of the simplex is reflected in the center of gravity of the remaining points to obtain a new trial point. The crossover operator, based on above method, is modified to improve the efficiency of genetic algorithm. To evaluate the efficiency of the improved algorithm, the algorithm is applied to five test problems, and our results are compared with other methods. The numerical results illustrate the efficiency of our method.