供水管网水力模型校核与漏损定位研究

发布时间：2018-04-26 21:00

本文选题：供水管网 + 水力模型　；参考：《重庆大学》2014年博士论文

【摘要】：管网水力模型的构建与运用是实现供水系统现代化管理的必要手段与途径，它不仅有助于供水调度、优化运营管理，还是开展管网漏损定位、管网水质模拟、预警突发性水质污染事件等相关研究的基础。在构建的管网水力模型中，由于节点流量与管道阻力系数无法直接测量，需要通过实测的水压与流量数据进行校核，在监测点不足的情况下，如何有效校核水力模型并提高校核效率是管网研究领域的难点与热点问题。本文以管网阻力系数校核、节点流量实时校核与漏损定位为主要研究内容，取得了如下几方面的研究成果与结论： 1）管网水力模型校核是复杂的非线性优化问题，其求解方法主要有基于确定性问题的“梯度迭代算法”(Gradient-based algorithm)和基于随机性问题的“随机搜索算法”(Stochastic search algorithm)。雅克比矩阵的计算是利用梯度迭代算法实现参数校核的关键，目前“扰动法”被最广泛用于计算供水管网雅克比矩阵，但该方法计算量大且精度低。本文应用矩阵分析法推导了供水管网雅克比矩阵的解析式，分别包括节点水压与管道流量对节点流量、管道阻力系数、管径的雅克比矩阵解析式。该式的导出不仅能提高管网参数校核程序的运算效率，还有利于参数敏感度分析、漏损定位、水压监测点优化布置等相关研究的开展。 2）提出了基于经验信息的加权最小二乘管网阻力系数校核法。由于实际中监测点个数小于未知量个数，管网阻力系数校核为欠定问题，传统校核方法习惯将管道按管材与管龄进行分组以减小未知量个数，本文将管道阻力系数的经验值作为伪观测值引入目标函数，相较于传统方法，其优点是能充分利用管道阻力系数的经验值方便地将欠定校核问题转化为超定。 3）构建了管网反计算模型并用于节点流量实时校核。考虑到通常将节点流量校核转化为优化问题求解需反复执行管网水力平差计算（正计算），本文构建了一个新的反计算模型，该模型采用矩阵分解法分离管网质量与能量方程中的变量，重新构建管网质量与能量方程组。通过直接求解该模型可同时获得管网中未知的节点流量、节点水压与管道流量，且无需进行传统的管网正计算，为管网水力模拟提供了一个新的算法。 4）渗漏管网水力特性及漏损定位研究。为研究渗漏管网水力特性、模拟管网漏损时水力状态，构建了管道渗漏模型并利用相关实测数据论证了模型的可靠性，将管道破损形式进行分类并提出了不同形式破损口渗漏面积的计算方法。管网漏损可视为节点流量的异常增大，会导致监测值的异常变化，利用一次二阶矩法量化监测值的正常波动范围，实现了管网漏损预警。将漏损定位转化为与节点流量校核相似的优化问题，以最小化监测值与管网水力模型计算值的差构建目标函数，通过逐个校核节点流量比较目标函数残差实现了管网漏损定位。考虑到对大型管网逐个校核节点流量计算量巨大，利用加权最小二乘回归法对目标函数的一阶泰勒展开式进行残差计算，提高了漏损定位效率。 5）考虑管网漏损特征的水压与水质监测点优化布置。定位管网漏损需要利用监测点提供的相关信息，优化水压与水质监测点的布置可使漏损引起的压力变化或水质污染尽可能被及时、准确监测。通过计算管网中各管道与节点的漏损概率，利用Kmeans聚类法对水压敏感度矩阵进行聚类分析，优化了水压监测点的布置。考虑到实际中污染物入侵管网的概率与管道漏损相关，，根据覆盖集法的思想，将管道漏损概率作为权系数引入最短水流时间矩阵中，在给定的监测等级下，通过求解目标函数极值优化了水质监测点的布置。文中所有提出的算法都在Matlab中编写了相应程序，并利用实际管网结合数字仿真阐明了算法的实现过程、测试了算法的收敛速度、验证了算法的正确性与可行性。
[Abstract]:The construction and application of the hydraulic model of the pipe network is the necessary means and way to realize the modern management of the water supply system. It is not only helpful to the water supply scheduling, the optimization of the operation management, but also the foundation of the related research on the leakage location of the pipe network, the water quality simulation of the pipe network, the early warning of the sudden water pollution events and so on. The point flow and pipe resistance coefficient can not be measured directly. It is necessary to check the measured water pressure and flow data. Under the condition of insufficient monitoring points, how to effectively check the hydraulic model and improve the checking efficiency is a difficult and hot issue in the field of pipe network research. This paper checks the pipe network resistance coefficient and the node flow real-time checking and leakage. Positioning is the main research content, and the following research results and conclusions are obtained.
1) the verification of the hydraulic model of the pipe network is a complex nonlinear optimization problem, and its solution methods mainly include the "gradient iterative algorithm" (Gradient-based algorithm) based on the deterministic problem and the random search algorithm (Stochastic search algorithm) based on the randomness problem. The calculation of the Jacobian matrix is to use the gradient iterative algorithm to realize the reference. At present, the "disturbance method" is the most widely used to calculate the Jacobi matrix of water supply network, but this method has large calculation and low precision. This paper derives the analytical formula of Jacobi matrix of water supply pipe network by matrix analysis method, which includes node flow, pipe resistance coefficient and Jacobi diameter of pipe diameter, respectively. The derivation of this formula can not only improve the operation efficiency of the pipe network parameter checking program, but also benefit the analysis of parameter sensitivity, leakage location, and the optimal layout of water pressure monitoring points.
2) a weighted least square pipe network resistance coefficient checking method based on empirical information is proposed. Due to the fact that the number of monitoring points is less than the number of unknown quantities, the resistance coefficient of the pipe network is checked into an underdetermined problem. The traditional checking method is used to group pipes and pipe age to reduce the number of unknown quantities. The empirical value of pipe resistance coefficient is given in this paper. As a pseudo observation value, the objective function is introduced. Compared with the traditional method, the advantage of this method is that it can make full use of the empirical value of the pipe resistance coefficient to convert the underdetermined proofreading problem into overdetermination.
3) the inverse calculation model of the pipe network is constructed and used to check the node flow in real time. Considering that the calculation of the hydraulic adjustment of the pipe network is often carried out, a new inverse calculation model is built in this paper. This model uses the matrix decomposition method to separate the variables in the quality and energy equation of the pipe network. The quality and energy equations of the pipe network are rebuilt. By directly solving the model, the unknown node flow, node water pressure and pipe flow can be obtained at the same time, and the traditional pipe network is not required to be calculated, which provides a new algorithm for the hydraulic simulation of the pipe network.
4) research on hydraulic characteristics and leakage location of leakage pipe network. In order to study the hydraulic characteristics of leakage pipe network, simulate the hydraulic state of pipeline leakage, build the pipeline leakage model and demonstrate the reliability of the model by using the related measured data, classify the pipe breakage form and put forward the calculation method of the leakage area of different forms of damaged mouth. The leakage can be seen as an abnormal increase of node flow, which can lead to abnormal changes in monitoring values. The two order moment method is used to quantify the normal fluctuation range of the monitoring value, and the leakage early warning of the pipe network is realized. The leakage location is transformed into an optimization problem similar to the node flow checking, in order to minimize the difference between the monitoring value and the calculation value of the hydraulic model of the pipe network. The standard function is used to check the leakage location of the pipe network by checking the residual error of the target function by checking the flow rate of the node by one by one. Considering the huge calculation amount of the flow of the large pipe network, the first order Taylor expansion of the target function is calculated by the weighted least square regression method, and the leakage location efficiency is improved.
5) the optimal layout of water pressure and water quality monitoring point considering the leakage characteristics of the pipe network. The leakage loss of the pipe network needs to use the relevant information provided by the monitoring point. The optimization of the layout of water pressure and water quality monitoring points can make the pressure change caused by the leakage or the water pollution as far as possible and accurately. Through the calculation of the leakage of the pipes and nodes in the pipe network The Kmeans clustering method is used to cluster analysis of water pressure sensitivity matrix, and the layout of water pressure monitoring points is optimized. Considering that the probability of the intrusion pipe network is related to the leakage of the pipeline, the leakage probability of the pipeline is introduced into the shortest flow time matrix by the idea of the cover set method, and under the given monitoring grade, the leakage probability of the pipeline is introduced. The layout of water quality monitoring points is optimized by solving the extremum of the objective function.
All the proposed algorithms in this paper have written the corresponding program in Matlab, and use the actual pipe network and digital simulation to illustrate the implementation process of the algorithm, test the convergence speed of the algorithm, and verify the correctness and feasibility of the algorithm.

【学位授予单位】：重庆大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TU991.33

【参考文献】