针对机械工程设计中经常遇到的多个局部极值的参数优化设计问题，探索了应用遗传算法求解其全局最优解的途径。文中讨论了遗传算法的基本构造模型，提出了新的适合度技术、新的交叉算子、新的变异算子和算法终止准则。最后，给出了遗传算法在机械工程优化中应用的两个实际算例，算例表明了该方法的有效性 In order to solve the problem of parametric optimization of multiple local extremums often encountered in mechanical engineering design, the ways of solving the global optimal solution by using genetic algorithm are explored. In this paper, the basic structural model of genetic algorithm is discussed. A new fitness technique, a new crossover operator, a new mutation operator and an algorithm termination criterion are proposed. Finally, two practical examples of genetic algorithm applied in mechanical engineering optimization are given. The examples show the effectiveness of the method