基于PSO—SA的二级倒立摆前馈补偿模糊神经网络控制

以模糊神经网络为基础，结合误差前馈补偿完成了二级倒立摆系统的稳定控制，并采用模拟退火粒子群算法对控制参数进行全局寻优。与基于状态变量合成的模糊神经网络控制器相比，该控制方法豢仅解决了多变量系统模糊控制器的“规则爆炸”问题，并且。由于所有状态变量直接参与控制输出，控制精度亦有所提高。仿真结果表明，该控制方案所需规则数目少，响应速度快，有良好的鲁棒性和非线性适应能力。

Fuzzy Neural Network Control Combined with Feed-forward Compensation for a Double Invert Pendulum Based on PSO-SA

The stability control of the double invert pendulum system was completed based on a fuzzy neural network and combined with the error feed-forward compensation in this paper and,an algorithm of particle swarm optimization combined with simulated annealing(PSO-SA) was used in global optimizing of control parameters.Compared with the fuzzy controller based on the synthesis of state variables this control scheme has not only solved the rule number explosion problem of multi-variable system fuzzy controller,but also has improved the control accuracy because of the direct participation of control output by all state variables.Simulation results show that this control scheme has the advantages of less rules,fast response speed,good robustness and strong nonlinear adaptive ability.