提出了一种应用RBF神经网络和遗传算法相结合的控制算法 ,用于控制二级倒立摆系统 .这种方法把线性控制的条件作为非线性最优控制性能指标的约束条件 ,用改进的遗传算法求解使性能指标最小的RBFNN控制器参数 .仿真结果表明 ,该方案优于传统的状态反馈方法和模糊控制方法 ,具有更大的稳定域 (0 <θ1,θ2 <2 5 °) ,抗干扰能力更强 . A control algorithm based on RBF neural network and genetic algorithm is proposed to control two-stage inverted pendulum system. This method takes the condition of linear control as the constraint condition of the nonlinear optimal control performance index, The algorithm solves the parameters of the RBFNN controller with the smallest performance index. The simulation results show that this scheme is superior to the traditional state feedback method and fuzzy control method, and has more stable domains (0 <θ1, θ2 <25 °) More capable.