考虑参数相关性的给水管网模型不确定性分析

Uncertainty analysis of water distribution network considering correlation of parameters

摘要:

给水管网运行环境的随机性与复杂性,导致管网水力模型参数存在不确定性.为分析模型参数的不确定性对模拟结果的影响,基于Monte Carlo随机模拟方法,提出考虑参数相关性的给水管网模型不确定性分析方法.将存在相关性的参数视为具有一定相关系数的随机变量,在Monte Carlo随机抽样阶段,先进行独立正态随机数抽样；当计算随机样本对应的管网模型结果时,通过逆-正交变换和逆-Nataf变换,把独立正态随机数转换为相关随机数,将相关随机数作为管网模型参数值进行水力计算.以管道摩阻系数的相关性和不确定性对模型计算结果中的节点压力和管道流量的影响为重点进行探讨.算例分析结果表明,管道摩阻系数的相关性对不确定性分析结果有显著的影响,考虑相关性与不考虑相关性的计算结果相比,节点压力、管道流量的平均波动值相差比率分别可达70%和60%以上.所以,在给水管网模型的不确定性分析中应该考虑模型参数间的相关性问题.

Abstract:

Due to the randomness and complexity of the actual operating environment, the parameters of water distribution network (WDN) hydraulic model are uncertain. Based on the Monte Carlo stochastic simulation method, this paper proposes an uncertainty analysis method for WDN hydraulic model considering the correlation of parameters. Correlated parameters were regarded as random variables with specified correlation coefficients. In the Monte Carlo random sampling, independent normal random number sampling was performed, and then the independent normal random number was converted into the correlated random number by inverse-Orthogonal transformation and inverse-Nataf transformation. The correlated random number was used as the model parameter value for hydraulic calculation. The effect of the correlation and uncertainty of pipe roughness on the uncertainty of the node pressure and the pipe flow in the WDN hydraulic model was studied. Case analysis showed that the correlation of pipe roughness had a significant impact on the uncertainty analysis results. The differences of the average fluctuation values of node pressure and pipe flow were more than 70% and 60% separately in correlated and independent conditions. Therefore, the correlation between model parameters should be considered in the uncertainty analysis of WDN.