基于改进粒子群算法的BP神经网络在边坡稳定性评价中的应用

边坡稳定性分析与评价是边坡工程的核心内容，具有高度非线性和不确定性特征。首先，选取了多个边坡工程实例构成学习样本集，以土体重度、内摩擦角、粘聚力、坡角、坡高、孔隙比六个主要影响因素作为土坡稳定性的评价判别指标；然后，采用改进的粒子群算法优化BP神经网络模型，将网络权值和阈值粒子化，通过引入粒子群进化度和粒子群聚合度实现惯性权重的动态变化，利用粒子群算法的全局搜索性实现网络权值和阈值的更新，从而增强算法对非线性问题的处理能力，加快了收敛速度；最后，通过与其它边坡稳定性评价算法进行比较分析，表明了本文研究算法的可行性与合理性。

Application of BP Neural Network in the Slope Stability Evaluation Based on Improved Particle Swarm Optimization Algorithm

Slope stability analysis and evaluation is the core content of slope engineering, it has the characteristics of highly non-linearity and uncertainty. Firstly, the multiple slope engineering instances was selected to constitute a learning sample set, the soil severe, internal friction angle, cohesive force, slope angle, slope height, void ratio, six main influencing factors were slope stability evaluation index. Then, the improved particle swarm algorithm was used to optimize the BP neural network model, the network weights and threshold values were particles, the inertia weight was implemented to dynamic change with the introduction of the particle swarm evolution degree and cluster degree. The network weights and threshold values were updated by use of particle swarm algorithm global searching capability, thus to enhance the algorithm processing capacity for nonlinear problem, and speed up the convergence rate. Finally, by comparing with other slope stability evaluation algorithm analysis, the feasibility and rationality of the research approach in the paper is presented.