摘要
针对三维激光扫描技术获取构件空间点云数据时不可避免的测量误差问题,本文研发最小二乘支持向量机-带精英策略非支配排序遗传算法(LSSVM-NSGA-Ⅱ)优化模型,以水平入射角度、倾斜角度、点云密度、测量距离、分辨率以及能见度作为输入指标,对相对误差和扫描时间进行多目标优化。首先利用LSSVM实现三维激光扫描参数对相对误差和扫描时间的高精度预测,得到其非线性映射关系函数后,将其作为目标优化函数,基于NSGA-Ⅱ进行多目标优化。研究表明,基于LSSVM的相对误差和扫描时间预测精度很高,利用遗传算法进行多目标优化后,获得最优的三维激光扫描参数值,被验证效果良好。体现LSSVM-NSGA-Ⅱ模型在寻优中的智能化、精准化,可以达到提高桥梁钢构件三维激光扫描的精准度,从而避免桥梁钢构件误差过大对建筑项目的影响。
For the error during 3 D laser scanning of the spatial point cloud data access components inevitably measurement, this paper develops the least squares support vector machine(LSSVM)-Non-dominated sorting genetic algorithm(LSSVM-NSGA-Ⅱ) optimization model with the level of incidence angle, angle, point cloud density, measuring distance, resolution and visibility as input index, the relative error and scanning time for multi-objective optimization. Firstly, LSSVM is used to realize the high-precision prediction of the 3 D laser scanning parameters on the relative error and scanning time. After obtaining the nonlinear mapping relation function, it is taken as the objective optimization function to carry out the multi-objective optimization based on NSGA-Ⅱ. The results show that the prediction accuracy of relative error and scanning time based on LSSVM is very high. After multi-objective optimization using genetic algorithm, the optimal 3 D laser scanning parameters are obtained, and the results are verified to be good. This proves the intellectualization and precision of LSSVM-NSGA-Ⅱ model in optimization, and the 3 D laser scanning precision of bridge steel components can be improved, so as to avoid the impact of excessive errors of bridge steel components on the construction project.
作者
吴贤国
邓婷婷
黄金龙
王洪涛
王堃宇
陈虹宇
李铁军
WU Xianguo;DENG Tingting;HUANG Jinglong;WANG Hongtao;WANG Kunyu;CHEN Hongyu;LI Tiequn(School of Civil and Hydraulic Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;China Construction Third Engineering Bureau Group Co Ltd,Wuhan 430000,China;School of Civil Engineering and Environment,Nanyang Technological University,Singapore 639798,Singapore;China Communications Construction Group Co Ltd,Beijing 100088,China)
出处
《土木工程与管理学报》
2021年第3期1-7,共7页
Journal of Civil Engineering and Management
基金
国家重点研发计划(2016YFC0800208)
国家自然科学基金(51378235
71571078
51308240)。
作者简介
吴贤国(1964-),女,湖北武汉人,博士,教授,研究方向为土木工程施工与管理(Email:Wxg0220@126.com);通讯作者:黄金龙(1989-),男,湖北武汉人,工程师,研究方向为土木工程施工与管理(Email:870852150@126.com)。
