基于Elman神经网络的导水裂隙带高度预测模型

Prediction Model for the Height of Water Flowing Fractured Zones Based on Elman Neural Network

准确预测导水裂隙带高度对于矿区煤炭资源安全开采具有重要意义。本文提出一种基于Elman神经网络的导水裂隙带高度预测模型,该方法克服了BP神经网络训练速度慢,稳定性差等问题,通过在结构层中增加一层承接层,作为一步延时算子,达到记忆的目的,增加了全局的稳定性;本文选择开采深度、开采厚度、覆岩结构、工作面斜长、煤层倾角作为影响导水裂隙带高度的主要因素,通过采用43组训练样本和3个测试样本数据建立了基于Elman神经网络的导水裂隙带高度预测模型,并与BP神经网络算法进行了对比。结果表明:BP神经网络与Elman神经网络的最大相对误差分别为16.35%和7.49%,Elman神经网络的预测精度更高。

It is important to accurately predict the height of water flowing fractured zones for the safe mining of coal resources.In this paper,a prediction model for the height of the water flowing fractured zones based on Elman neural network was proposed to solve the problems of BP neural network,such as slow training and poor stability.The model achieved the purpose of memory by adding a succession layer to the structure layer as one-step delay operator,which could increase the global stability.In this study,the prediction model for the height of water flowing fractured zone was established based on Elman neural network and compared with the BP neural network algorithm on the analysis of some factors,including mining depth and thickness,overlying strata structure,inclined length of working face,and dip angle of coal seam,with 43 sets of training sample and 3 sets of testing sample.The results show that the prediction accuracy of the Elman neural network with the maximum relative error of 7.49%is higher than the BP neural network with the maximum relative error of 16.35%.