摘要
为实现复杂地质条件中深基坑变形的精确预测,提出了一种动态惯性权重粒子群算法改进支持向量机的基坑变形预测模型。引入遗传算法改进的支持向量机模型和标准BP神经网络模型作为横向对比验证了预测效果。结果表明:动态惯性权重对支持向量机核函数参数的寻优速度更快,收敛精度更高,采用改进粒子群算法优化的支持向量机模型预测的平均相对相对误差仅为5.46%,拟合精度相较其他算法更高,预测效果良好,可较为准确的实现深基坑的变形预测。
In order to achieve accurate prediction of deep foundation pit deformation in complex geological conditions,a dynamic inertia weight particle swarm optimization algorithm improved support vector machine(SVM)based prediction model for foundation pit deformation was proposed.The improved support vector machine model based on genetic algorithm and the standard BP neural network model are introduced as lateral comparisons to verify the prediction effect.The results show that the dynamic inertia weight has a faster optimization speed and higher convergence accuracy for the kernel function parameters of support vector machines.The average relative error predicted by the support vector machine model optimized by the improved particle swarm optimization algorithm is only 5.46%,and the fitting accuracy is higher than other algorithms.The prediction effect is good,and it can more accurately predict the deformation of deep foundation pits.
作者
蔡群群
CAI Qun-qun(Shanghai Urban Construction Municipal Engineering Group Co.,Ltd.,Shanghai 200065,China)
出处
《黑龙江交通科技》
2023年第5期97-99,103,共4页
Communications Science and Technology Heilongjiang
关键词
基坑工程
变形预测
机器学习
支持向量机
粒子群算法
foundation pit works
deformation prediction
machine learning
support vector machine
particle swarm optimization
