基于粒子群优化算法的相关向量机边坡稳定性分析模型

为快速获取边坡稳定性系数，及时对边坡进行稳定性评价，提出一种基于粒子群优化算法(particle swarm optimization，PSO)的相关向量机(relevance vector machine，RVM)边坡稳定性分析模型。该模型通过选取影响边坡稳定性安全系数的6个主要因素，并对这6个主要影响因素产生的30组数据进行拟合训练，利用粒子群算法对相关向量机模型参数进行优化，选取最优参数值，根据这30组训练样本来对剩余4组样本进行精准预测。结果表明：与实际值进行对比，基于PSO-RVM模型预测的平均相对误差仅为5.64%，且建立的PSO-RVM预测模型的边坡稳定性安全系数的平均相对误差均明显优于利用BP(back propogation)神经网络和协调粒子群(coordinated particle swarm optimization，CPSO) -BP模型预测得到的平均相对误差，进一步为边坡稳定性预测及评价提供一种新方法。

Slope stability analysis model based on PSO-RVM

In order to obtain the slope stability coefficient quickly and evaluate the slope stability in time, a slope stability analysis model based on particle swarm optimization (PSO) and relevance vector machine (RVM) is proposed. Six main factors were selected as the model selects that affect the safety factor of slope stability, and 30 groups of data generated by these six main influencing factors were fitted. Particle swarm optimization was used to optimize model parameters of relevance vector machine, and the optimal parameter values were selected. The remaining 4 groups of samples were accurately predicted according to these 30 groups of training samples. It is show that the average relative error predicted by PSO-RVM model is only 5.64% comparing with the actual value. The average relative error of the slope stability safety factor for the established PSO-RVM prediction model is significantly better than the average relative error predicted by BP neural network and coordinated particle swarm optimization (CPSO) and BP model, which further provides a new method for slope stability prediction and evaluation.