基于混合PSO-RBF神经网络的铁路隧道岩爆分级预测

岩爆是铁路隧道建设中主要灾害之一。为了准确预测铁路隧道岩爆烈度等级,以岩石应力系数σ_θ/σ_c、岩石脆性系数σ_c/σ_t以及弹性能量指数W_et作为岩爆烈度评价指标,提出一种基于混合粒子群优化算法优化的径向基(RBF)神经网络岩爆预测模型。首先在国内外研究成果基础上,选取80组已有岩爆实例作为模型基础数据;然后运用结合了模拟退火算法的粒子群算法(混合PSO)改进径向基神经网络,通过训练数据选取最优的权值W和基函数标准差σ,得到混合PSO-RBF神经网络岩爆烈度预测模型;最后将模型应用于实际铁路隧道工程进行验证。研究结果表明:该模型兼顾个体最优和全局最优,能够正确、有效的对铁路隧道岩爆等级做出预测,为铁路隧道岩爆预测提供了一种新方法。

Prediction of rockburst classification of railway tunnel based on hybrid PSO-RBF neural network

Rock burst is one of the main disasters in railway tunnel construction.In order to accurately predict the rockburst intensity grade of railway tunnel,a rockburst prediction model based on radial basis(RBF)neural network optimized by hybrid particle swarm optimization algorithm was proposed in this paper,which stress coefficientσ_θ/σ_c,brittleness coefficienσ_c/σ_tand elastic energy index of rockW_etwere chosen as the rockburst prediction indexes.Firstly,based on the research results at home and abroad,80 groups of existing rock burst cases were selected as the basic data of the model.Then,the particle swarm optimization algorithm combined with simulated annealing algorithm(Hybrid PSO)was used to improve the radial basis function neural network,and the optimal weight(W)and the basis function standard deviation(σ)were selected by training the data,and the prediction model of rock burst intensity based on hybrid PSO-RBF neural network was obtained.Finally,the model was applied to the actual railway tunnel engineering for verification.The case study shows that the model takes into account both individual optimization and global optimization,can correctly and effectively predict the rockburst level of railway tunnels,and provides a method and approach for rockburst prediction of railway tunnels.