基于禁忌BP神经网络的动态测量误差预测研究

本文针对BP算法在神经网络参数学习中局部性能好但易陷入局部极小值而禁忌搜索算法拥有良好的全局性能的特点，提出了神经网络的综合训练方法（禁忌BP算法，TSBP Algorithm）即首先利用禁忌搜索算法对神经网络参数进行金局搜索训练，然后再利用BP算法对参数进行局部学习。设计了一种多样化和集中化并行搜索的禁忌搜索算法，一定程度上解决了传统禁忌算法局部寻优和拓展搜索空间之间的矛盾。最后利用仿真实例验证了TSBP神经网络较之BP神经网络和TS神经网络在动态误差预测方面的优越性，得出了相关结论。

Research on forecast of dynamic measurement error based on TSBP neural network

While training neural network,the BP Algorithm has good local performance but it is easy to fall into local minimum;and the tabu search algorithm has good global performance,so in this paper the following combinatorial method(TSBP Algorithm)is put forward:first of all,train neural network based on tabu search algorithm in global space;then,train the parameters of neural network using BP algorithm in local space.A kind of tabu search algorithm with diversification and intensification search is designed.The n...