混合遗传神经网络在边坡稳定性评价中的应用研究  

Study on the Application of Mixed Genetic-Neural Network in Slope Stability Evaluation

作　　者：杨蕾[1] 林红[2] 

机构地区：[1]中国石油大学(华东)数学与计算科学学院,山东东营257061 [2]中国石油大学(华东)机电工程学院,257061

出　　处：《中国农村水利水电》2006年第7期75-76,79,共3页China Rural Water and Hydropower

摘　　要：将改进的遗传算法与BP神经网络相结合,分析了边坡稳定性的影响因素,建立边坡稳定性评价的神经网络模型。分别用改进的遗传算法和自适应BP算法对网络权值进行全局优化和局部二次优化。选用边坡实例对网络进行训练检验,结果表明,混合遗传神经网络提高了训练速度和泛化能力,对边坡稳定性分析有较好的适用性,为边坡稳定性评价提供了依据。Combining modified Genetic Algorithm and BP Neural Network, the influencing factors for slope stability are analyzed, and a neural network model for the evaluation of slope stability is established. The modified genetic algorithm and self-adaptive BP algorithm are applied in the global optimization and local secondary optimization. Selection of slope is used as an example for the training and verification. The results show that the mixed genetic-neural network speeds up the process of training and popularity capacity, suits for the slope stability analysis, and provides support for the evaluation of slope stability.

关 键 词：混合遗传神经网络 边坡稳定性 遗传算法 自适应BP算法 全局优化 

分 类 号：TU432[建筑科学—岩土工程] TP183[建筑科学—土工工程]