遗传-神经网络算法在水下隧道涌水量预测中的应用

水下隧道涌水问题受到多种因素的综合影响,具有非线性和高度复杂性。本文应用遗传算法和BP神经网络,结合工程实例,选用隧道围岩裂隙发育情况、上覆含水体富水性、上覆水压、隧道埋深、隧道围岩上覆相对隔水层强度和上覆基岩破碎带与隧道顶板的距离作为影响水下隧道涌水的主要因素,以调查的数据作为训练样本,使用遗传算法优化BP神经网络的初始权值和阈值,建立了水下隧道涌水量的遗传-神经网络预测模型,并进行了计算分析。结果表明:该模型收敛性能好,预测精度高,简单可行。该方法为水下隧道涌水量的预测提供了一条新思路。

Application of genetic-neural network algorithm to forecast water inflow in underwater tunnel

Water inflow problem in underwater tunnel which is affected by many factors possesses highly complexity and nonlinear.The theory of genetic algorithms and BP(back-propagation) neural networks were introduced.Taking a practical engineering as an example,six main control factors were chosen in the analysis,such as the development degree of surrounding rock fissures,water yield property of overlaying aquifer,water pressure of overlaying aquifer,the depth of tunnel,the strength of the overlying impermeable laye...