A New Crossover Operator for Improving Ability of Global Searching

Considering the deficiency of the traditional crossover operator of Genetic Algorithms (GAs) in global searching, the partheno-crossover operator is proposed in this paper. The partheno-crossover, a new crossover operator, still uses the traditional method of crossover. The individual of the population doesn´t cross over the individual of the same population but the individual of the completely different with it. When these two individuals have many offspring, only the fittest offspring can replace his father´s position in the population. So this offspring either inherits the father´s good gene fragments or have a better gene fragments. It both ensures that the population develops to excellent direction, and avoids that the population converges to a local optimal solution. Compared to the Standard Genetic Algorithm (SGA), the traditional improved Genetic Algorithm (TIGA) and the Partheno-Genetic Algorithm (PGA), the improved Genetic Algorithm used the partheno-crossover operator (PCGA) allows better stability of global convergence and some comparative results relative to the optimization of test functions has increase of 10 times, even Several decuples.