传统遗传规划常因树复杂度的无节制增大 ,导致运行时间过长而不能直接在工程应用中使用。文章在遗传规划中引入了多目标优化方法 ,寻求在解的性能和复杂度之间的均衡。该方法不仅提供了一种在随机搜索过程中有效控制树结构长度的方法 ,而且产生精度更高的最优结果。通过对符号回归问题的应用实例 ,验证得到了较好的结果。 Traditional genetic programming often grows out of control of tree complexity, resulting in long run times that can not be used directly in engineering applications. The article introduces a multi-objective optimization method in genetic programming to seek a balance between the performance and complexity of solutions. This method not only provides a way to effectively control the length of tree structure in random search, but also produces the best result with higher accuracy. Through the application example of the symbol regression problem, we get better results.