基于GA神经网络的自适应预测控制的设计与仿真

The Design and Simulation of a Self-Tuning Predictive Control Based on Neural Network Optimized by Genetic Algorithm

针对自适应预测控制抗干扰、鲁棒性与实时性的矛盾,根据遗传算法优化后的神经网络具有很强的自适应性和学习能力、记忆能力、非线性映射能力、鲁棒性和容错能力,文章提出了一种新的自适应预测控制,成功地避免直接矩阵求逆,仿真表明该算法具有良好的综合性能和鲁棒性。

Because of the contradiction of resisting disturbance, robustness and real time, in view of the powerful abilities of self-adapting, learning, remembering , non-linear mapping and robustness of neutral network optimized by genetic algorithm, the article puts forward a new self-tuning predictive control ,which succeeds in avoiding evaluating the inverse matrix on-line. The results of simulation show that the algorithm has better integrative performances and robustness.