当前位置: X-MOL 学术Water Res. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Uncertainty analysis method for diagnosing multi-point defects in urban drainage systems
Water Research ( IF 11.4 ) Pub Date : 2024-11-24 , DOI: 10.1016/j.watres.2024.122849
Chutian Zhou, Pan Liu, Xinran Luo, Yang Liu, Weibo Liu, Huan Xu, Qian Cheng, Jun Zhang, Kunming Wu

Urban drainage system (UDS) plays a key role in city urbanization, where defective pipes can lead to seepage. Previous studies have identified the locations of defects in UDS using inverse optimization models. However, the unique optimal solution neglects uncertainty analysis, which may lead to misdiagnosis. In addition, the multi-point defect diagnosis has heavy computational burden due to high dimensional parameters space. To address these issues, this paper proposes a hybrid method that leverages the genetic algorithm (GA) to identify probable space, and then utilizes the adaptive Metropolis (AM) to provide an estimation of the posterior probability distribution (PPD). Firstly, a multi-population GA is employed for the maximum exploration within the model space. Then, AM algorithm is used to explore the final PPD of each pipe defect parameter. The metrics accuracy (ACC), Matthews correlation coefficient (MCC) and mean absolute error (MAE) are used to evaluate the diagnosis performance. A synthetic UDS case with randomized multi-point seepage scenarios is used to validate the method. Results indicate that the proposed hybrid method is effective in diagnosing multi-point defect, with 0.91, 0.78 of the hybrid method and 0.87, 0.69 of the DiffeRential Evolution Adaptive Metropolis method for the ACC and MCC, respectively. Meanwhile, the diagnosis speed has increased by 32 %. The result PPD passes the 90 % confidence interval validation. The proposed method can provide effective uncertainty analysis to reduce misdiagnosis of the traditional method.

中文翻译:


城市排水系统多点缺陷诊断的不确定性分析方法



城市排水系统 (UDS) 在城市化中起着关键作用,有缺陷的管道会导致渗漏。以前的研究已经使用逆向优化模型确定了 UDS 中缺陷的位置。然而,独特的最优解忽略了不确定性分析,这可能会导致误诊。此外,由于高维参数空间,多点缺陷诊断具有较重的计算负担。为了解决这些问题,本文提出了一种混合方法,该方法利用遗传算法 (GA) 来识别可能的空间,然后利用自适应大都会 (AM) 提供后验概率分布 (PPD) 的估计。首先,采用多种群 GA 在模型空间内进行最大程度的探索。然后,采用增材制造算法探究各管道缺陷参数的最终 PPD。指标准确性 (ACC) 、 马修斯相关系数 (MCC) 和平均绝对误差 (MAE) 用于评估诊断性能。使用具有随机多点渗流情景的合成 UDS 案例来验证该方法。结果表明,所提出的混合方法对多点缺陷的诊断是有效的,ACC 和 MCC 分别为 0.91、0.78 和 0.87、0.69 的 DiffeRential Evolution Adaptive Metropolis 方法。同时,诊断速度提高了 32%。结果 PPD 通过 90% 置信区间验证。所提方法可提供有效的不确定性分析,以减少对传统方法的误诊。
更新日期:2024-11-24
down
wechat
bug