人工神经网络在供水优化调度中的应用研究

发布时间：2018-10-12 18:06

【摘要】：供水管网优化调度的目标是在满足用户对供水压力、流量、水质要求的前提下,尽可能降低生产的直接成本和间接成本,并提高生产和输送过程的安全性、稳定性。管网建模仿真是预测供水管网系统动态工况的有效办法,有助于实现对管网的科学化、现代化管理,有助于实现供水系统的科学、调度优化,有助于实现按需供水、减少损耗和漏损。传统的供水管网建模采用微观模型,但在实际应用中存在着一些缺点,导致了在供水管网系统的调度优化中无法充分有效地利用该模型。本文首先利用BP神经网络建立了城市用水量的预计模型。将临近的24小时历史用水量数据作为模型输入,计算下一小时的用水量,作为优化模型中的用水量约束参数。城市供水管网模型采用宏观模型,模型将各个水厂、泵站的流量作为输入,测压点压力及水厂泵站的出口压力作为输出。通过BP神经网络的训练,可以得到供水管网的模型。利用历史数据来验证上述两个模型,结果表明模型具有较高的精度。在得到供水量预计模型和管网模型后,建立了优化方程。以供水总成本(电耗、制水成本等)最小作为优化目标,并以供水量、测点水压等作为约束。采用启发式算法解决该优化问题,首先将优化问题通过罚函数法转化为无约束问题,然后通过遗传算法求解。最后,通过实际数据来测试优化模型,结果表明,在能够满足供水量、水压的前提下,优化的结果可以很好降低供水总成本。因此,本文设计的供水优化调度模型可以有效的调整供水方案,为水厂、泵站的运行调度提供理论指导。该模型在未来也可以与SCADA系统有机的结合,形成高效的自动供水调度系统。
[Abstract]:The aim of optimal dispatching of water supply network is to reduce the direct and indirect cost of production and to improve the safety and stability of production and transportation process under the premise of satisfying the demand of users for water supply pressure, flow rate and water quality. The modeling and simulation of water supply network is an effective method to predict the dynamic working conditions of water supply network system. It is helpful to realize the scientific and modern management of the water supply network, to realize the science of the water supply system, to optimize the dispatching, and to realize the water supply on demand. Reduce loss and leakage. The traditional water supply network model adopts micro model, but there are some shortcomings in practical application, which leads to the failure to make full use of the model in the optimization of water supply network system. In this paper, the prediction model of urban water consumption is established by using BP neural network. The approaching 24-hour historical water consumption data is used as the input of the model and the next hour water consumption is calculated as the water consumption constraint parameter in the optimization model. The model of urban water supply network adopts macroscopic model, which takes the flow rate of each water plant and pump station as input, the pressure of measuring point and the outlet pressure of pump station of water plant as the output. Through the training of BP neural network, the model of water supply network can be obtained. The two models are verified by historical data, and the results show that the model has high accuracy. After obtaining the water supply prediction model and the pipe network model, the optimization equation is established. The minimum total cost of water supply (electricity consumption, water production cost, etc.) is taken as the optimization objective, and the water supply and water pressure are taken as constraints. The heuristic algorithm is used to solve the optimization problem. Firstly, the optimization problem is transformed into an unconstrained problem by penalty function method, and then solved by genetic algorithm. Finally, the optimization model is tested by the actual data. The results show that the optimization results can reduce the total cost of water supply well under the premise of satisfying the water supply quantity and water pressure. Therefore, the water supply optimal scheduling model designed in this paper can effectively adjust the water supply scheme and provide theoretical guidance for the operation and scheduling of water plants and pumping stations. The model can also be combined with SCADA system in the future to form an efficient automatic water supply dispatching system.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TU991;TP183

【参考文献】