Abstract: The pseudorandom numbers generator based on the primary-cellular automata（CA） and hybrid-CA is studied.The comparative experiment and the analysis show that the chaos primary-CA can produce the stable and superior pseudorandom numbers.Though the performance of hybrid-CA is better than the average performance of primary-CA，it is far worse than the performance of chaos primary-CA.Furthermore，the pseudorandom numbers generation algorithm based on hybrid-CA and particle swarm optimization（PSO） is proposed in view of pseudorandom numbers generator based on the hybrid-CA.In the algorithm，the cellular of CA can be considered as the particle of PSO.The iterative process of each cellular based on its own rule is corresponding to the flight process of each particle in search rule-space.Each particle iteratively evaluates the fitness of the candidate solutions according to the entropy value of pseudorandom numbers produced by corresponded cellular.The algorithm can improve the correlation of pseudorandom numbers by searching the best rules for each cellular.Finally，the further research of pseudorandom numbers generator coupled with CA-PSO based on niche technology is presented.

摘要： 针对基本元胞自动机（Cellular Automata，简称CA）、混合CA的伪随机数发生器进行了深入的研究，通过对比实验观察到混沌型基本CA输出的伪随机序列质量稳定并较优，而混合CA输出伪随机序列的相关性，尽管优于基本CA的平均表现，但远差于混沌型基本CA的表现。针对混合CA的伪随机数发生器，提出了一种基于混合CA与粒子群优化（Particle Swarm Optimization，简称PSO）算法融合的伪随机数产生算法。在该算法中，元胞对应于PSO的粒子，每个元胞按照各自不同的规则进行迭代演化，其对应粒子在迭代规则空间中飞行。该算法通过计算每个元胞产生伪随机序列的熵值作为粒子的适应度函数值，有效地实现每个元胞最佳规则的搜索，一定程度上提高了混合CA产生伪随机序列的质量。给出了基于小生境技术、构造出最优CA-PSO耦合伪随机数发生器的研究方向。