改进的神经网络算法在瓦斯涌出量预测中的应用

针对传统神经网络存在的收敛速度慢、容易陷入局部最小等缺陷,采用改良的BP算法———双权值迭代优化法,提高神经网络传统BP算法的训练速度。以三层神经网络为例,对权值进行优化。实验对比表明：双权值迭代优化法应用于瓦斯涌出量的预测,比一般BP网络有更高的预测精度和程序运行速度。

Application of estimation based on improved BP algorithm

This paper is aimed at an alternative to conventional neural network which suffers from a lower convergence rate and higher tendency to fall into partially smallest,by developing the improved BP algorithm,namely double weight iteration optimization method,to enhance the training speed of the neural conventional BP algorithm network.The method consists of taking three layer neural networks as the example,and,through the analysis and the optimization for weight value,applying the algorithm into predicting the mine gas emission.The experimental comparison shows that the double weight iteration optimization method gives a greater forecast accuracy and faster program operation than conventional BP algorithm network in predicting the mine gas emission.