机构地区: 华南理工大学土木与交通学院
出 处: 《船海工程》 2007年第6期107-110,共4页
摘 要: 针对BP算法存在收敛速度慢和局部极值的问题,采用此方法对BP学习算法进行改进,结合实验采集的数据进行仿真计算,结果表明采用此方法可以有效地提高Elman网络的训练速度和稳定性,使El-man网络能够更有效地对波浪补偿平台垂向位移进行短时预报。 Since BP algorithm had the defects that the speed of constriction was slow and had local extreme data, the modified BP algorithm was put forward. Using the collected real time data, the results indicated that the modified BP algorithm could improve the training speed and stability of Elman network effectively so that Elman neural network could realize shor^-time prediction for vertical displacement of wave compensating platform more effectively.