【作者】 黄良沛；

【导师】 刘义伦；

【作者基本信息】 中南大学 ， 机械设计及理论， 2005， 博士

【摘要】 随着我国城市人口的快速增长和人民生活水平的不断提高,对城市供水系统的服务要求也越来越高,供水系统的规模在不断扩大,复杂性随之提高。城市供水系统是极其重要而复杂的网络系统,确保其安全可靠的运行和正确有效的管理具有重大意义。为了提高系统的社会效益和经济效益,采用现代化的技术手段、先进的控制理论来提高供水企业的生产水平和生产效率,运用现代化的管理理论、优化方法以及计算机技术对城市供水系统进行管理、监控、预测和优化调度都是势在必行的。论文对城市供水系统中管网预测、优化调度、智能控制以及大功率电机平稳切换等问题进行深入系统的研究和探讨。 城市供水管网预测是优化调度的前提和基础。选取BP神经网络作为建模基础,研究探讨城市供水系统神经网络预报的原理和机制,阐述了建立基于BP网络的城市供水时序预测模型方法。分别以管网的节点压力、管道流量和用水负荷的过去某一历史时期数据纪录,建立基于BP神经网络的管网预测模型,对未来某一时段的节点压力、管道流量和用水负荷进行预测。运用BP神经网络对城市供水的变化规律进行识别的实质,就是通过选择适当的神经网络模型逼近实际系统的动态过程。人工神经网络模型在已有的历史数据的基础上,一旦通过学习识别出城市供水的变化规律,网络模型通过联想,就可以预报下一个时段的城市供水需求。 从供水管网系统整体优化和最小供水费用的角度出发,建立了城市供水系统管网优化调度的数学模型。重点研究在用户节点压力已知条件下,复杂供水系统整体优化调度数学模型求解问题。对城市供水系统管网优化调度的遗传算法进行深入的研究,并给出了详细的求解步骤。文中还对某市供水系统的优化调度进行仿真研究,根据不同时段的各用户节点压力服务质量要求,采用MATLAB编写基于遗传算法的优化调度程序,对供水系统进行管网整体优化调度。采用优化调度程序进行生产调度可以节省运行费用。 根据供水厂二级泵房的水泵机组生产调度特点,建立了水泵机组组合优化的数学模型,对水泵机组优化组合遗传算法进行了认真的研究和探讨。文中还结合某供水厂二级泵房机组情况,对水泵机组组合优更多还原

【Abstract】 With the quickly increase of city population in our country, and with the gradually improvement of people life, the service of city water distribution systems are brought up higher and higher requests, the scale of city water distribution systems gradually expand and water distribution systems become much more complex. Due to importance and complexity of water distribution systems, it is important significance to ensure water distribution systems to safety work and effective management. In order to enhance economic benefits and society benefits of water distribution systems, it is necessary to improve production level and efficiency of water plants by using modern industry technology and advanced process control theory, and it is indispensable to realize water distribution systems optimal schedule, intelligence prediction, supervision and management by modern management theories, optimal methods automation and computer technology. In this dissertation pipe network prediction, optimal schedule, intelligence control and the steady switch of high power electromotor are deeply researched and discussed for water distribution systems.Water supply pipe network prediction is premise base of optimal schedule. As the foundation of establishing model, BP neural networks is selected to research and discuss the principles and mechanisms of neural networks water supply pipe network prediction, and the method is presented how to establish water supply time serial prediction model based on BP neural networks. On the ground of the history datum of node pressure, pipe flow, water charge of water distribution systems, water supply pipe network prediction model based on BP neural networks is respectively established, and node pressure, pipe flow, water charge in next period of time are predicted by their respective BP neural networks model. Making use artificial neural networks recognize principles of water supply is to select proper neural networks model to approach practical systems. Based on these history datum, once BP neural networks model recognize principles of water supply through study, water supply request next time would be predicted.Proceeding from view of overall optimum and minimum water cost for water distribution systems, mathematic model of optimum schedule is built for complex water distribution systems. Genetic algorithms for optimum schedule of complex water distribution systems are presented, then code regular, chromosome evaluation and genetic operation are deeply studied, furthermore solution steps for optimum schedule are given in detail. Matlab is adopted to write optimum control program based on genetic algorithms, and the program is applied to simulate production control for certain city water distribution systems. Simulation result shows that overall optimum solution is obtained when genetic algorithms are used to optimum schedule of complex water distribution systems, and that production by optimum schedule program contributes to economize water-supply cost.According to practical production characteristic of water supplying plant second grade stations, mathematic model is formulated about optimizing combination of pump units. On account of model optimal solution, the pump units optimizing combination genetic algorithm is presented, then code regular, chromosome evaluation and genetic operation are deeply studied and discussed. In terms of pump units condition of some water plant second grade stations, Simulation research is carried out about pump units combination optimization. It is indicated that optimizing scheduler of pump units with this mathematic model acquires much satisfying available values and great economic benefits.Aiming at characteristics of vvvf constant pressure water-supplying system such as multi-parameters, strong coupling, nonlinear, long time delay, the self-adaptive Fuzzy-PID controller based neural networks is designed. The controller can make online self-regulation of PID parameters, according to different working condition, proper parameters are selected to effectively adjust water pressure constant. It is especially discussed about system identification NN1 and self-adaptive NN2,furthermore solution steps is given out about Fuzzy-PID control algorithm based on BP neural networks, simulation program is written in Matlab environment to make simulation research on controller,, Simulation result and field application show that after the Fuzzy-PID control algorithm based on BP neuralnetworks being used, both the static characteristic of control system and dynamic characteristic of that are improved, system output datum accord with referenced model output datum.According to electrical equivalent circuit and vector graph of induction electromotor, discussion is made in theory about the problem ,that exists in the course of transformation from variable frequency to work frequency. Furthermore it is pointed out that whether the high-power electromotor can be switched from VF to WF lies on phase consistency between work frequency phase voltage and corresponding work frequency phase voltage. At the same time switch method are studied and discussed. Frequency and phase discriminator is adopted in the industry test to successfully complete the transformation from VF to WF. As the switch electrical current is small, non-impact switch of the high-power electromotor is realized. Furthermore it is also presented that PLL is integrated inside VWF, when the frequency of PLL input sine wave is equal to the VCO base frequency, PLL can eliminate initiative phase errors, thus PLL can rapidly lock frequency and phase of VWF output voltage, and keep phase consistency between WF voltage and corresponding VF output voltage.更多还原