基于LSTM神经网络的煤矿突水预测

Coal mine water inrush prediction based on LSTM neural network

  • 摘要: 针对煤层底板突水预测问题,在总结现有突水预测方法和理论的基础上,通过特征选择实验得出水压、距工作面距离、砂岩段厚度、煤层厚度、煤层倾角、断层落差、裂隙带、开采面积、采高、走向长度是影响突水发生的主要因素,这些因素具有复杂、非线性的特点。提出基于长短时记忆(LSTM)神经网络构建的突水预测模型,将煤矿突水实例的数据作为样本数据对模型进行训练。最后,将LSTM神经网络模型与遗传算法-反向传播(GA-BP)神经网络模型和反向传播(BP)神经网络模型进行对比实验。实验结果表明,LSTM神经网络模型在测试集上的预测正确率更高,稳定性更好,更适用于煤层底板突水预测。

     

    Abstract: According to the prediction of water inrush from coal seam floor, based on the summarization of existing water inrush prediction methods and theories, the feature selection experiment shows that water pressure, distance from the working surface, sandstone section thickness, coal thickness, coal seam inclination, fault throw, fissure zone, mining area, mining height and strike length are the main factors affecting the occurrence of water inrush. These factors are complex and non-linear. A water inrush prediction model based on long short-term memory(LSTM) neural network was proposed. The data of the coal mine water inrush case was used as sample data to train the model. Finally, the LSTM neural network model is compared with the genetic algorithm-back propagation(GA-BP) neural network model and back propagation(BP) neural network model. The experimental results show that the LSTM neural network model has higher prediction accuracy, better stability, and is more suitable for coal seam floor water inrush prediction.

     

/

返回文章
返回