ABSTRACT
Although, GA has many advantages in searching for the optimal values and many experts and scholars research on the GA continually, too many problems still exist.[3] The global search ability of the standard GA comes from the random creation of the initial population and the crossover operation. But due to contrast the limitation of the population with more dimension of the solution space, this makes the global search is limited, cannot find the global optimal value and computational precision is low. The mutation operator is small in the SGA, a few new mutated individuals are easily assimilated by the many old individuals. Close relative propagate and premature convergence
1 INTRODUCTION
Recently, the presentation of genetic algorithms and development is a large progress, which are evolutionary algorithms based on the nature biology evolution. The advantage of this method is not necessary to compute the function gradient and not dependent on the problems self. Besides, the algorithms search the global optimum and find the global optimal solution at large probability.[1]
Continuous rolling is generally used in bar production, the process is divided into three stage, which is rough rolling, intermediate rolling and finishing rolling. According to commonly used pass design method of rolling bar, it is difficult to make balance loads between the mills and make the least total energy consume. With the objective of least energy consumption and mill loads balance, the passes of bars are optimized by the improved Genetic Algorithm compiled by Matlab language, thus, the reasonable rolling process is obtained.
