改进PSO-BP算法在隧道洞室地基稳定性中的应用

Application of Improved PSO-BP Algorithm in the Stability of the Tunnel Cavern Foundation

隧道洞室稳定性问题是一个复杂的非线性力学问题,常规的方法很难描述这种复杂的非线性关系。为及时评价隧道洞室地基的稳定性,以便采取合理的开挖方案,提出了改进PSO-BP算法对其稳定性进行预测的方法。粒子群算法具有搜索速度快、效率高、算法简单等优点。BP算法有很强的非线性映射能力、泛化能力等功能,但其容易陷入局部最优。采用PSO算法克服BP神经网络的缺陷,提高BP网络模型预测精度。以重庆小什字车站洞室为例,验证了改进PSO-BP算法能够快速、准确地获取不同方案下的洞室地基安全系数,且预测结果比模糊神经网络预测结果要好,证明了该方法的可行性。

Tunnel cavern stability problem is a complex nonlinear mechanical issue, which is difficult to describe this complex nonlinear relationship through conventional method. In order to evaluate the stability of the tunnel cavern and take reasonable foundation excavation scheme timely, a method of improved PSO-BP algorithm is put forward to predict the stability of the tunnel cavern. PSO algorithm has such advantages of quick search speed, high efficiency, and simple arithmetic. BP algorithm has a strong nonlinear mapping ability, generalization ability and other functions, but it is easy to fall into local optimum. PSO algorithm is used to overcome BP neural network defects and improve its prediction accuracy. Chongqing Xiaoshizi station cavern is taken as example, and it is verified that the improved PSO- BP algorithm can quickly and accurately obtain ground safety factor cavern under different scenarios, and the forecast results is better than fuzzy neural network, which proves the method feasible.