论文部分内容阅读
标准BP神经网络算法收敛速度慢是限制其广泛应用的主要原因.为此,以标准BP算法为基础,应用最小二乘法理论,提出了一种收敛速度快的BP算法--NLMSBP算法.仿真结果表明,和标准BP算法及其它改进形式比较,NLMSBP算法收敛速度大大提高,稳定性并未降低,这为BP神经网络应用于实时性要求高的场合提供了算法基础.该算法缺点是计算量大,所需计算机内存大,不适于大型网络的计算.
The slow convergence rate of standard BP neural network algorithm is the main reason that restricts its wide application.Therefore, based on the standard BP algorithm, a least-squares BP algorithm named NLMSBP algorithm is proposed by applying the theory of least square method.The simulation results It shows that compared with the standard BP algorithm and other improved forms, the convergence speed of NLMSBP algorithm is greatly improved and the stability is not reduced, which provides an algorithm basis for applications where BP neural network has high requirements in real time. , The required computer memory is large, not suitable for large-scale network computing.