爆破振动诱发民房结构损伤识别的随机森林模型

发布时间：2018-03-29 13:51

本文选题：爆破振动　切入点：民房结构损伤　出处：《爆炸与冲击》2017年06期

【摘要】：为快速、准确地评价爆破振动诱发民房结构损伤效应,借鉴随机森林理论并结合工程实际,建立露采爆破振动诱发民房结构损伤识别的随机森林模型;综合考虑爆破参数、爆破振动特征参量及房屋结构动力特性等因素,选取质点峰值振动速度、主频率、主频率持续时间、段药量、爆心距、施工质量参数、场地条件参数、屋盖形式参数、砖墙面积率、民房高度、灰缝强度和圈梁构造柱参数等12个影响因素作为模型输入,将砖混结构建筑物的损害等级作为模型输出;基于多分类器集成的思想,以108组爆破振动实测数据作为学习样本进行训练,建模过程中由多个决策树集成随机森林、用投票的方式实现对民房结构损伤有效识别;用12组现场数据验证模型的有效性;在对样本分类的同时,计算预测变量的重要性值,发现质点峰值振动速度为最重要的评价指标,其后依次为爆心距,主频率持续时间,主频率,圈梁构造柱参数,灰缝强度,屋盖形式参数,民房高度,段药量,施工质量参数,砖墙面积率和场地条件参数。研究结果表明:随机森林模型预测结果学习样本准确度是87.97%,而测试集准确度是91.67%,与实际情况吻合较好,预测精度较高。
[Abstract]:In order to evaluate the damage effect of civil house structure induced by blasting vibration quickly and accurately, using the theory of random forest and combining with engineering practice, a stochastic forest model for structural damage identification of civil house induced by blasting vibration in open pit is established, and the blasting parameters are considered synthetically. The vibration characteristic parameters of blasting and the dynamic characteristics of building structure are selected. The peak vibration velocity of particle, the main frequency, the duration of main frequency, the quantity of explosive, the distance of blasting center, the construction quality parameter, the parameters of site condition, the form parameter of roof are selected. Twelve factors, such as area ratio of brick wall, height of house, strength of ash joint and parameters of ring beam structure column, are taken as model input, and the damage grade of brick and concrete structure building is taken as model output, which is based on the idea of multi-classifier integration. 108 groups of measured data of blasting vibration were used as learning samples to train, in the process of modeling, the random forest was integrated by multiple decision trees, and the damage of civil house structure was effectively identified by voting, and 12 groups of field data were used to verify the validity of the model. At the same time of classifying the samples, the importance of predicting variables is calculated, and it is found that the peak vibration velocity of particle is the most important evaluation index, followed by the burst distance, the duration of the main frequency, the main frequency, the parameters of the structure and column of the ring beam, the strength of the gray joint. Roof formal parameters, building height, section charge, construction quality parameters, The results show that the accuracy of learning sample is 87.97, while the accuracy of test set is 91.67, which is in good agreement with the actual situation, and the prediction accuracy is higher.
【作者单位】：黄淮学院建筑工程学院;中南大学土木工程学院;湖南科技大学能源与安全工程学院;
【基金】：国家自然科学基金项目(11072072)
【分类号】：O38;TU317

【相似文献】