基于GRA-SSA-Elman的隧道施工瓦斯安全性预测评价

Gas safety prediction and evaluation of tunnel constructionbased on GRA-SSA-Elman

为准确预测隧道施工瓦斯灾害,有效评估瓦斯安全,建立基于灰色关联分析(grey relational analysis,GRA)与麻雀搜索算法(sparrow search algorithm,SSA)优化Elmen神经网络的隧道施工瓦斯安全性预测评价模型。首先从水文地质条件、隧道设计、煤层结构状况及施工管理四个方面考虑,遴选12个主要影响因素,采用GRA法量化分析选取的影响因素与评价对象间的关联性;然后引入SSA优化Elman模型中的初始权值和阈值建立模型,采用留一交叉验证法验证模型性能;最后为进一步验证模型性能,对选取的实例瓦斯隧道待测样本进行瓦斯灾害预测评价,同时与PSO-Elman、Elman、Fisher判别分析(fisher discriminant analysis,FDA)模型预测结果与现场实际结果对比分析。结果表明:SSA-Elman预测结果与实际工程结果一致性更高,优化后的模型精确度和稳定性更好,可操作性强,对隧道施工瓦斯安全性评估具有良好的适用性。

In order to accurately predict tunnel construction gas disasters and effectively evaluate gas safety,a prediction and evaluation model of tunnel construction gas safety based on grey relational analysis(GRA)and sparrow search algorithm(SSA)is established.Firstly,considering the hydrogeological conditions,tunnel design,coal seam structure and construction management,12 main influencing factors are selected,and the correlation between the selected influencing factors and the evaluation object is quantitatively analyzed by GRA method.Then,SSA is introduced to optimize the initial weight and threshold in Elman model,and the model performance is verified by leaving one cross validation method.Finally,in order to further verify the performance of the model,the selected samples of gas tunnel to be tested are evaluated for gas disaster prediction,and compared with the prediction results of PSO-Elman,Elman,Fisher discriminant analysis(FDA)model and the field actual results.The results show that the SSA-Elman prediction results are more consistent with the actual engineering results,and the optimized model has better accuracy and stability,strong operability,and good applicability to the gas safety evaluation of tunnel construction.