A probabilistic performance-based analysis approach for a vibrator-ground interaction system振子-场地相互作用系统的概率性能分析方法
![]()
Peng X, Liu YNW, Hao L, 2024. A probabilistic performance-based analysis approach for a vibrator-ground interaction system. Probabilistic Engineering Mechanics, 76: 103626.DOI: 10.1016/j.probengmech.2024.103626
摘要 | Abstract
输入不确定性对动力系统影响的研究越来越受到关注。振子-场地 (vibrator-ground, VG) 相互作用系统的概率分析较为罕见，且系统不确定性的影响亟需揭示。本研究旨在提出一种多源不确定性条件下对振子-场地系统进行概率性能分析的方法。振子-场地系统的概率模型基于蒙特卡罗 (Monte Carlo, MC) 模拟结合拉丁超立方抽样 (Latin hypercube sampling, LHS) 方法构建，采用遗传算法优化人工神经网络减少蒙特卡罗模拟的大量计算成本。随后，采用优劣解距离法 (technique for order preference by similarity to ideal solution, TOPSIS) 进行多准则灵敏度分析，以评估输入不确定性对振子动力性能的影响。最后，采用所提方法，对振子-场地系统进行概率性模拟分析。结果表明了该概率性能分析方法对振子-场地系统的有效性，并评估了输入不确定性对系统动力性能的影响。关键词: 不确定性分析, 振子-场地系统, 蒙特卡罗模拟, 代理模型, 灵敏度分析There is an increasing interest in investigating the effects of input uncertainties on dynamic systems. The probabilistic analyses for a vibrator-ground (VG) interaction system are rare and the effects of system uncertainties need to be revealed. This study aims to present an approach for the probabilistic performance-based analysis of the VG system under multi-source uncertainties. The probabilistic model of the VG system is constructed on the basis of the Monte Carlo (MC) simulation combined with the Latin Hypercube Sampling (LHS) method, while the artificial neural networks optimized by the genetic algorithms are employed to reduce the large computational expenses in the MC simulation. Then, a multi-criteria sensitivity analysis is presented by using a technique for order preference by similarity to ideal solution (TOPSIS) to evaluate the effects of input uncertainties on the dynamic performance of the vibrator. Finally, a probabilistic simulation analysis of the VG system is conducted by implementing the presented approach. The results demonstrate the effectiveness of the presented probabilistic performance-based analysis approach for the VG system and evaluate the effects of input uncertainties on the dynamic performance of the system.Keywords: Uncertainty analysis; Vibrator-ground system; MC simulation; Surrogate model; Sensitivity analysis![]()
图 1: 模型描述: (a) 振子-场地系统的结构; (b) 动力模型Fig. 1. Model description: (a) Structure of the VG system; (b) Dynamic model
Fig. 2. Flow chart of the soil-baseplate interaction solution scheme
Fig. 3. Nondimensional ground force of the vibrator in one sweep (A = 275 kN, f_s = 3 Hz, f_e = 120 Hz, T = 20 s)
图 4: 人工神经网络的示意图: (a) 基于人工神经网络的代理模型; (b) 单个神经元Fig. 4. Schematic diagram of ANN: (a) ANN-based surrogate model; (b) One neuron
Fig. 5. Illustration of the probabilistic analysis
Fig. 6. Error histogram distribution
图 7: 多源不确定性对系统性能的影响: (a) 共振频率; (b) 无量纲地震动幅值峰值; (c) 无量纲地震动幅值均值 Fig. 7. Effects of multi-source uncertainties on system performance: (a) ω_n; (b) Ā_peak; (c) Ā_mean
图 8: 不同输入参数的灵敏度分析结果: (a) 各性能指标的全局 Sobol 指标; (b) 整体多准则全局灵敏度指标Fig. 8. Sensitivity analysis results of different input parameters: (a) Total Sobol' indices concerning each performance index; (b) Overall multi-criteria global sensitivity index
图 9: 单次分析下不同输入参数的全局 Sobol 指标Fig. 9. Total Sobol’ indices of different input parameters in one sweep
Fig. 10. Overall global sensitivity analysis in one sweep
作者信息 | Authors
彭珣 Xun Peng, 通讯作者 (Corresp.) 西南石油大学 (Southwest Petroleum University) 机电工程学院Email: pengxun@swpu.edu.cn
吉利学院 (Geely University of China)
中国石油集团 (China National Petroleum Corporation) 东方地球物理勘探公司
律梦泽 M.Z. Lyu | 编辑 (Ed)
P.D. Spanos | 审校 (Rev)
陈建兵 J.B. Chen | 审校 (Rev)
彭勇波 Y.B. Peng | 审校 (Rev)
