摘要
本文首先介绍基于PSO优化的BP神经网络算法;并在此基础上,考虑影响边坡稳定性分析的包括岩石容重、粘聚力、内摩擦角、边坡角、边坡高度、空隙水压力等各种自然因素,以文献中土质边坡工程的数据为例,用基于PSO优化的BP神经网络算法对边坡稳定性进行评价。最后将评价结果与传统BP方法计算结果进行逐一对比,结果证明该方法不仅能满足边坡稳定性评价的精度要求,而且还可以加快训练收敛速度,其泛化性能也比较好,使得边坡稳定性评价更为便捷迅速,从而证明该方法具有一定的推广应用价值。
Neural network based on particle swarm algorithm is introduced in this paper. Based on the data collected from projects in typical areas, the stability of typical slopes is assessed with consideration of various natural factors such as rock density, soil cohesion , soil inner friction angle, slope grade, slope height and pore water pressure. Based on the data gained from the literature, using neural network on the basis of Particle Swarm Algorithm to evaluate slope stability. Comparing the assessment results with traditional BP algorithm's results, it suggests that this algorithm can reduce number and error of training obviously and has better generalization than traditional BP algorithm, and worth being popularized.