基于ZOA CNN GRU模型的煤层底板突水等级预测

Prediction of water inrush grade of coal seam floor based on ZOA CNN GRU model

针对传统循环神经网络煤层底板突水等级预测模型存在预测精度低、模型参数过多造成模型训练速率下降和出现过拟合现象等问题,引入斑马优化算法对卷积神经网络和门控循环单元神经网络的组合模型进行优化,建立ZOA CNN GRU神经网络煤层底板突水等级预测模型。为验证模型的可行性,采用九龙矿区煤层底板突水数据对模型进行训练,并将所建模型和CNN GRU神经网络以及GRU神经网络进行对比分析。研究结果表明:与CNN GRU神经网络和GRU神经网络模型相比,ZOA CNN GRU神经网络模型预测准确率最高,达到98%,且ZOA CNN GRU神经网络模型稳定性、泛化能力均优于对比模型。

In view of the problems existing in the coal seam floor water inrush grade prediction model based on traditional circulating neural network,such as low prediction accuracy,decrease of model training rate caused by excessive model parameters and over-fitting phenomenon,zebra optimization algorithm was introduced to optimize the integrated convolutional neural network(CNN)and gated recurrent unit(GRU)neural network model,a ZOA CNN GRU neural network model for predicting the water inrush grade from the coal seam floor was established.To ascertain the model's viability,water inrush data obtained from the coal seam floor in the Jiulong mining area was used to train the model,and the model was compared and analyzed with CNN GRU neural network and GRU neural network.The results indicated that in comparison to the contrasting models,the ZOA CNN GRU neural network model exhibited the highest level of prediction accuracy,reaching 98%,and the ZOA CNN GRU neural network model demonstrated superior stability and generalization capabilities.