遗传算法、进化规划和进化策略这三类进化算法都是基于对自然进化的模拟，其区别在于产生下一代群体的规则不同，但下一代群体的产生又都是仅依赖于其父代，因而进化算法的运行过程可以视为一个Ｍａｒｋｏｖ过程，其状态转移矩阵可以表示成一个统一的形式．利用矩阵范数的基本性质，得到了进化算法收敛速度的一个下界，同时也得到了进化算法收敛性的一个证明，并由此解释了遗传算法能很快地得到一个较好的解而要花费较长时间才能得到最优解的原因，为今后加快进化算法收敛速度指出了一个可行的研究方向 Genetic algorithms, evolutionary programming and evolutionary strategies are all based on the simulation of natural evolution. The difference between them is that the rules for generating the next generation are different. However, the generation of the next generation depends on their parents only. The operation of evolutionary algorithm can be regarded as a Markov process whose state transition matrix can be expressed in a unified form. By using the basic properties of matrix norm, a lower bound of convergence speed of evolutionary algorithm is obtained, and a convergence of evolutionary algorithm is also proved. It also explains that genetic algorithm can quickly obtain a good solution and cost A long time to get the optimal solution, pointing out a feasible research direction for accelerating the convergence rate of evolutionary algorithm in the future