School of Safety and Emergency Management Engineering, Taiyuan University of Technology

为了提高瓦斯涌出量预测的科学性和准确性,提出一种基于鲸鱼优化算法(WOA)和长短期记忆网络(LSTM)的瓦斯涌出量多步预测模型。该模型首先采用皮尔逊(Pearson)相关系数法进行瓦斯涌出量影响因素的特征分析,筛选了9个主要影响瓦斯涌出量变化的特征作为模型的外部输入特征;其次采用鲸鱼优化算法对LSTM神经网络的隐藏层神经元个数、时间步长、批处理数进行优化;最后,构建WOA-LSTM模型进行瓦斯涌出量预测,实验研究了不同时间步长下模型的预测精度并对比分析了LSTM、RNN、BP模型的预测效果。结果表明:基于WOA-LSTM的瓦斯涌出量多步预测模型在3个时间步长的预测模型误差值达到最小,其平均绝对误差相较于LSTM、RNN和BP神经网络模型分别降低了41.6%、46.6%、65.8%,具有较强的鲁棒性,可为矿井瓦斯的防治提供参考。

In order to improve the scientificity and accuracy of gas emission prediction in coal mine, this paper put forward a multi- step gas emission prediction model of whale optimization algorithm ( WOA) - Long short - term memory network(LSTM). The model first used Pearson correlation coefficient method to analyze the characteristics of the influencing factors of gas emission, and selected 9 main characteristics affecting gas emission as the external input features of the model. Then, the optimal parameters were obtained by using whale optimization algorithm to optimize the number of hidden layer neurons, batch number and time step of LSTM neural network. Finally, a WOA-LSTM model was constructed to predict the value of gas emission. The prediction accuracy of the model under different time steps and the prediction effects of LSTM, RNN and BP model were compared and analyzed experimentally. The experimental results show that the multi-step prediction model of gas emission based on WOA-LSTM reaches the minimum error value in three time steps. Its average absolute error is reduced by 41. 6%,46. 6% and 65. 8% respectively compared with LSTM, RNN and BP neural network models. Moreover, the model has strong robustness and can provide reference for mine gas prevention and control.

张玉财,王毅,郭凯岩. 基于 WOA-LSTM 的工作面瓦斯涌出量预测研究[J]. 矿业安全与环保,2023,50(5):50-55.

ZHANG Yucai,WANG Yi,GUO Kaiyan. Research on prediction of gas emission in working face based on WOA-LSTM[J].Mining Safety & Environmental Protection,2023,50(5):50-55.