Abstract: It is the main factors for fitness functions to guide the search of the genetic algorithm optimization process. The exponential fitness functions are improved by exponentiation scale transformation. They are used to evaluate several common fitness functions to keep their diversity of population and convergence of the algorithms. The optimal computation is compared for the usual and the improved fitness functions under the same conditions of genetic manipulation and their parameters in using three typical test functions. Numerical results show that it is significant for the new fitness functions of a power optimal algorithm to improve the overall performance including the accuracy, convergence speed, and convergence stability of the ameliorated genetic algorithms.

Key words: genetic algorithms, fitness functions, testing functions, optimal computation

摘要： 在遗传算法优化过程中，引导搜索的主要依据是适应度函数。通过评估常见的几种适应度函数，兼顾保持种群的多样性和算法的收敛性，由乘幂尺度变换，提出了一种改进的乘幂适应度函数。以三个典型的测试函数为例，在相同遗传操作和参数情况下，分别采用常见的与改进的适应度函数进行优化比较。结果表明，所改进的乘幂适应度函数能明显提高算法的收敛精度、收敛速度和收敛稳定性，对提高遗传算法的整体性能有重要的意义。

关键词: 遗传算法, 适应度函数, 测试函数, 优化计算