基于改进BP算法的煤体氧化速率预测研究

The Application and Research of Predicting Coal Oxidation Velocity based on Ameliorative BP Algorithms

BP网是一种单向传播的多层前向网络,可用于最优预测。本文介绍了如何建立BP网络结构模型,包括如何确定隐含层层数以及隐含层节点数目,从而构造一个三层的9-20-1型神经网络,利用此网络预测煤体氧化速率的定量变化,分析得到的预测结果可用来研究煤体氧化性规律,对考察煤体自然发火倾向性具有指导意义。

Back-Propagation Network is a kind of multilayer back-propagation networks which are unilateral propagation and it can be applied on optimization predicting. In this article, how to found structure model of BP network is introduced, in which how to choose layer number of erytie layer and node number of erytie layer is explored. At the same time, a three-layer type of 9-20-1 network is constructed, with this network we can predict the quantitative change of coal oxidation velocity, via analyzing the result of predicting we can study the rule of coal oxidation and it can direct us to research the orientation of coal spontaneous detonability.