【作者】 王滨；

【导师】 胡林献；

【作者基本信息】 哈尔滨工业大学 ， 电气工程， 2008， 硕士

【摘要】 热电厂具有供电和供热的双重功能,比普通电厂拥有更高的能源利用率。目前,热电厂在电力、石化、冶金、造纸和热电联产等企业之中广泛存在。热电厂的运行方式普遍采用母管制。母管蒸汽压力的自动控制对热电厂的安全经济运行至关重要,而负荷的优化分配则是实现母管蒸汽压力控制的基础。因此,研究母管蒸汽压力控制的负荷优化具有重要意义。本文针对热电厂采用不同厂家分布式控制系统(Distributed Control System,简称DCS)的情况,就实现母管蒸汽压力自动控制时所遇到的负荷分配方面的问题进行了深入研究。首先,针对不同厂家的DCS之间不能联网这一问题,设置了一套母管蒸汽压力控制系统,规划了它与各炉DCS之间的信息交换。在详细地分析了各种通信方式的特点后,提出了信息交换采用硬接线的通信方式进行。其次,研究了母管系统的压力负荷特性,建立了压力负荷特性模型,提出了一种基于现场数据的BP神经网络模型参数辨识方法。在此模型的基础上推导出了根据母管蒸汽压力变化求解母管蒸汽总负荷需求的计算公式。实例计算表明参数辨识方法的有效性和模型的正确性。母管蒸汽总负荷需求的求解是设置的母管蒸汽压力控制系统的一项重要功能,是实现母管蒸汽压力控制的先决条件。母管系统压力负荷特性模型的建立有助于推导母管蒸汽总负荷需求的计算公式。然后,为了实现母管蒸汽总负荷需求的分配,针对已有控制方案固定调压锅炉、母管蒸汽压力难以长期自动控制的弊端,提出了调压锅炉组在线辨识、调压锅炉数量的实时统计方法及调压锅炉组蒸汽总负荷的计算公式。最后,针对调压锅炉组蒸汽总负荷的优化分配,提出了一种负荷优化分配新方法——自适应Hopfield神经网络法。算例证明该方法具有和常规的等微增率法一样良好的优化效果,且没有使用条件的限制。更多还原

【Abstract】 The thermal power plants supply us with electrical power and heat, and have more efficient in energy usage than common power plants. So the thermal power plants exist in enterprises numerously such as power plants, petrochemical enterprises, metallurgical enterprises, papermaking factories and CHP enterprises. The operation mode of thermal power plants commonly is main-pipe scheme. The automatic control of main-pipe steam pressure is very important to safe and economic operation of thermal power plants, and the optimal assignment of load is the basis of control of main-pipe steam pressure. So the research on load optimization of main-pipe steam pressure control is of great significance.This paper conducts in-depth research on the problems of load assignment for main-pipe steam pressure control against the thermal power plants using distributed control system (DCS) of different companies.Firstly, against the problem that different kinds of DCS can’t communicate, this paper designs a main-pipe steam pressure control system (MSPCS), plans the message exchange between the MSPCS and every boiler DCS. Besides, after analyzing the features of different communication ways, this paper presents the hard-wiring communication way to exchange the messages.Secondly, this paper does research on the pressure-load character of main-pipe system, models the character, and presents a BP neural network method basing on history data to identify the model parameters. Besides, this paper drives the calculation formula which can calculate the total steam load demand according to the variety of main-pipe steam pressure on the basis of model of pressure-load character. The calculation results of an example demonstrate that the identifying method is available and the model proposed is correct. The calculation of main-pipe total steam load demand is an important function of main-pipe steam pressure control system designed and the predetermination of main-pipe steam pressure control. The building of model of pressure-load character of main-pipe system is helpful to derive the calculation formula of main-pipe total steam load demand.Thirdly, in order to assign the main-pipe total steam load demand, this paper presents the mind of pressure-adjustment boilers’online identifying, the real-time quantity calculation of pressure-adjustment boilers and the calculation formula of steam total load of pressure-adjustment boilers against the defects that the main-pipe steam pressure control plans existed fix pressure-adjustment boiler which steam pressure can’t be controlled effectively in a long run.At last, against the optimal assignment of steam total load of pressure-adjustment boilers, this paper presents a new optimization method which is adaptive Hopfield neural network method. The calculation results of an example demonstrate that the adaptive Hopfield neural network method is as good as common equal micro-increase rate method in optimal results. Besides, the adaptive Hopfield neural network method has no using restrictions.更多还原