基于PSO和BP复合算法的模糊神经网络控制器

Fuzzy neural network controller based on particle swarm optimization and back-propagation complex algorithm

为了克服单独应用粒子群算法(PSO)或BP算法训练模糊神经网络控制器参数时存在的缺陷,提出了一种训练模糊神经网络参数的PSO+BP算法。该算法将二者相结合,即在PSO算法中加入一个BP算子,以充分利用PSO算法的全局寻优能力和BP算法的局部搜索能力,从而更有效地提高其收敛速度、训练效率和提高该模糊神经网络控制器的控制效果。最后的仿真实验结果验证了该基于PSO+BP复合算法的模糊神经网络控制器的有效性和可行性。

In order to overcome the shortcomings of the algorithm when applying the particle swarm optimization algorithm （PSO） or the back-propagation algorithm （BP） to train the fuzzy neural network controller parameters, a combined algorithm of both PSO and BP was proposed. The combined algorithm takes full advantages of the global optimization ability of PSO and local search ability of BP by inserting a BP operator into the PSO algorithm. The convergence speed, the training efficiency and the performance of the fuzzy neural network controller are improved greatly. The final simulation results verify the effectivity and feasibility of the fuzzy neural network controller based on the combined algorithm of both PSO and BP.