论文部分内容阅读
为了解决可变学习速率的BP神经网络(VLBP)在训练时容易陷入局部极小的问题,在VLBP的算法规则中引入模拟退火中的metropolis接受准则,使得在均方误差增量超过设定的界限值时,权值更新不总是被取消,而是以一定的概率被接受,构造了一种容易跳出局部极小的VLBP神经网络(MVLBP)。运用MVLBP算法对短时交通流进行预测,仿真结果表明,MVLBP神经网络训练收敛速度更快,且有较好的预测精度。
In order to solve the problem that the BP neural network (VLBP) with variable learning rate is easy to fall into local minima during training, the metropolis acceptance criterion in simulated annealing is introduced into the algorithm rule of VLBP so that when the mean square error increment exceeds the set value When the threshold value is changed, the weight update is not always canceled, but accepted with a certain probability. A VLBP neural network (MVLBP) which is easy to jump out of local minima is constructed. The MVLBP algorithm is used to predict the short-term traffic flow. The simulation results show that the MVLBP neural network has faster convergence rate and better prediction accuracy.