【作者】 曹宏；

【导师】 唐胜利；

【作者基本信息】 重庆大学 ， 动力工程， 2002， 硕士

【摘要】 清洁、高效是人们衡量发电技术先进与否的两条最重要标准。以天然气、煤层气为燃料的联合循环——“燃气联合循环”发电集清洁(能源)、高效(发电方式)于一身，成为当今世界愈加受到青睐的实用发电技术。本文以华能重庆燃机电厂（108.25MW）为研究对象，应用理论分析和人工神经网络相结合方法，对蒸汽--燃气轮机联合循环进行变工况性能计算，并以此为基础开发出一套适用于工况监测的软件系统，以提高联合循环电站的营运水平，获取最大经济效益。本文进行了以下的主要工作：①建立理论模型，根据对象的实际设备的特点和理论关系推导压气机、燃烧室、燃气轮机、余热锅炉、汽轮机的静态数学模型；根据本研究的实际情况，在建立各部件模型时，对已有特性曲线的部件，采用多项式拟合的方法，得到较为准确的特性，而对部分没有曲线的特性，则采用神经网络的方法建立模型。②进行了大量的现场实验，测试在不同条件下系统各节点的参数，并以此作为修正理论模型的依据；③根据现场实测数据和理论模型，计算理论模型中的各种参数；④分析并确定了本软件从已有网络设备获取信息的方法。通过本文研究取得了如下成果：①在阅读了大量文献资料的基础上，对以燃气轮机为核心的多种发电厂系统结构进行了分析，并分析了我国电力工业结构调整的方向；②根据研究对象的实际系统结构，给出了系统中各设备的效率计算关系，并根据实际设备的性能曲线，拟合了相应的计算公式，其离线计算结果正确，并为实现在线计算奠定了基础；③首次将人工神经网络的理论及其基本特征引入并应用于燃气--蒸汽联合循环电站的系统故障诊断和变工况性能分析和计算中。以此为理论基础，结合实际系统和文献资料，给出了影响联合循环电站经济性的各主要设备的常见故障原因、故障现象，构造了相应的故障诊断模型，应用神经网络方法率先提出并建立了相应的故障诊断系统和软件系统。更多还原

【Abstract】 Cleanness and high efficiency are the two important standards by which people determine whether a power generation technology is advanced or not. The combined cycle power generation burning natural gas and coal-based gas has become a practical power generation technology which is more and more popular world-wide for its both advantages of cleanness (energy) and high efficiency ( method of power generation).This essay, which researches into Huaneng Chongqing Gas Turbine Power Plant (108.25MW) as its object of study, carries out calculations on the variable rated performance of the steam-gas turbine combined operation, utilizing an integrated method of theoretic analytics and artificial neural net. A set of software system applicable to monitoring of plant performance has been developed based on the calculations, which is used to raise the efficiency of a combined cycle plant to achieve largest possible economic profit.This essay has touched upon the following work:1. Has built up theoretic models including static mathematic models of air compressor, combustion chamber, gas turbine, waste heat recovery boiler and steam turbine which are deduced according to the feature and theory of the equipment of the object of study. Upon establishment of each model based on the actual conditions, a method of multiple imitation and combination is applied to parts which have already got characteristic curves to have achieved quite accurate curves whilst the method of neural net is applied to the parts which have no characteristic curves for establishment of models.2. Has carried out a lot of on-site tests to test parameters of each system section under different conditions, which is used as a basis for correction of theoretic models.3. Has calculated various parameters in the theoretic models according to the tested datum on site and the theoretic models.4. Has analyzed and determined the method for this software system to obtain information from existing network equipment. This essay has achieved the following findings via its research:Has analyzed various power plant structures centered by a gas-steam turbine unit based on its references to large amount of technical documentation, it has also<WP=8>1. analyzed possible ways of restructuring of China’s power industry.2. Has worked out the relationship of efficiency calculation of each equipment in the system according to the actual system structure of the object of study. Corresponding calculation formulas is imitated and combined in accordance with the characteristic curves of the actual equipment, and the 0ff-site calculation appear to be correct which plays a good foundation for on-site calculation.3. Has for first time applied the theory of artificial neural net and its basic characteristics to the calculation on system fault detection and variable rated performance of a gas-steam turbine combined cycle power plant, Based on this theory while incorporating the information from the actual system and technical documentation, the essay has also outlined causes and phenomenon of normal faults of the main equipment which influence the efficiency of a combined cycle power plant, and has structured fault detection model accordingly. Thus corresponding fault detection system and a software system are proposed and established for the first time via utilization of the method of neural net.更多还原