论文部分内容阅读
针对前馈网络ＢＰ算法所存在的收敛速度慢且常遇局部极小值等缺陷，提出一种基于Ｕ－Ｄ分解的渐消记忆推广卡尔曼滤波学习新方法，与ＥＫＦ相比，不仅大大加快了学习收敛速度、数值稳定性好，而且比ＢＰ算法需较少的学习次数和隐节点数仍可达到更好的学习效果。仿真计算表明，该方法是提高网络学习速度、改善学习效果的一种有效方法，可有效解决非线性系统建模、辨识与控制问题。
Aiming at the shortcomings of the BP algorithm such as slow convergence rate and frequent local minima, a novel learning method of extended memory based on U-D for extended memory is proposed. Compared with EKF, it not only accelerates greatly, The convergence rate of learning, numerical stability is good, and the number of learning and hidden nodes less than the BP algorithm can still achieve better learning results. The simulation results show that this method is an effective way to improve the speed of network learning and improve the learning effect, which can effectively solve the problems of modeling, identification and control of nonlinear systems.