| 摘要: |
| 现有的码头结构评价方法具有滞后性，无法持续的获取结构的安全信息。本文基于损伤诱因反演研究，综合码头常见的两种失效模式，引入承载力富余系数作为码头结构评价指标，依托各机器学习方法开展损伤诱因强度、位置、承载力富余系数的预测。结果表明：PSO-BP(粒子群优化神经网络)模型对强度反演与承载力富余预测效果最好，均方误差为0.6812；PSO-SVM(粒子群优化支持向量机)优化模型对损伤诱因位置识别精度最高，准确率为0.9844。该方法模型预测精度较高，为基于损伤诱因反演的码头健康监测方法提供了状态评价指标，可为码头结构长期、动态评价提供思路与指导。 |
| 关键词: 架空直立式高桩码头 机器学习 损伤诱因反演 承载力富余估计 |
| DOI： |
| 分类号: |
| 基金项目:重庆交通科技项目(CQJT2022ZC03); 重庆市教委重点科技项目(KJQN201800739)； |
|
| Structural Health State Assessment Method for Overhead Vertical High Pile Wharf based on Damage Induced Inversion |
|
ZHOU Shi-liang1, TANG Meng-han1, SUN shiquan2, XU ying2, XIE lei2
|
|
1.The Southwestern Research Waterway Engineering Institute,Chongqing Jiaotong University;2.Chongqing Shipping Constion Development Co,Ltd
|
| Abstract: |
| Existing methods for evaluating dock structures are lagging and do not allow continuous and dynamic access to the safety information of the structure. In this paper, based on the study of damage causation inversion, two common failure modes of wharves are synthesized, and the load carrying capacity surplus coefficient is introduced as the evaluation index of wharf structure, and the prediction of damage causation strength, location and load carrying capacity surplus coefficient is carried out by relying on various machine learning methods. The results show that: PSO-BP (particle swarm optimization neural network) model has the best effect on the strength inversion and load carrying capacity surplus prediction, with the mean square error of 0.6812; PSO-SVM (particle swarm optimization support vector machine) optimization model has the highest accuracy in the damage causative factor location identification, with an accuracy of 0.9844. This method has high prediction accuracy, and it provides a good solution for the health monitoring of wharf based on the damage causative factor inversion. This method has high model prediction accuracy and provides a state evaluation index for the dock health monitoring method based on damage causation inversion, which can provide ideas and guidance for the long-term and dynamic evaluation of dock structure. |
| Key words: Overhead vertical high pile wharf Machine Learning Damage induced inversion Bearing capacity surplus estimation |
