摘要
BP神经网络是一种按误差逆传播算法训练的多层前馈网络,其与遗传算法的结合可以得到一种拥有良好的全局优化搜索和局部时频特性的学习训练途径。针对BP网络的不足,提出了一种基于改进遗传算法的BP神经网络控制器,此方法可以克服基本遗传算法收敛速度慢,容易陷入"早熟"收敛,计算稳定性不好等一系列问题,进一步提高了BP神经网络控制器的性能。最后通过对轧制力模型的预报仿真,证明了控制器的有效性。
BP neural network is a kind of error back propagation training algorithm for the multilayer feed forward network. The combination of genetic algorithm and BP neural network can get one type of training way with good global optimization search and local time-frequency characteristics. This paper aims at the shortage of BP network, and proposes a kind of BP neural network controller basett on improved genetic algorithm, which can overcome a series of problems, such as the slow convergence of standard genetic algorithm, falling into the premature convergence easily, and the instability of computation. It further improves the performance of BP neural network controller. Finally, through the simulation of the rolling force prediction model, the paper proves the effectiveness of the proposed controller.
出处
《漳州职业技术学院学报》
2013年第1期1-6,共6页
Journal of Zhangzhou Institute of Technology
关键词
BP神经网络
改进遗传算法
轧制力模型
BP neural network: improved genetic algorithm: rolling force model