【作者】 陈荣保；

【导师】 费敏锐；

【作者基本信息】 上海大学 ， 控制理论与控制工程， 2010， 博士

【摘要】 电力系统是国家建设的保障体系,在以可持续发展的各项发电技术中,煤电充当着发电行业的主要角色。燃煤发电技术从能耗角度分析,低能耗零排放和等量煤耗下的高发电效率,使亚临界机组不断地被超临界机组和超超临界机组所替代。但从发电工艺角度分析,都是燃煤发电。燃煤发电包括了给粉环节、燃烧环节、发电环节和辅助环节,整个发电过程是一个大容量、大滞后、非线性的理化特性型工艺参数的多能量转换过程,由电厂分散控制系统(DCS)实时过程监控。电厂DCS在电厂生产中发挥着重要作用,是电厂生产普遍选用的监控系统。但在DCS对电厂的监控过程中,控制策略的依据并非来自于发电过程中最重要的工艺对象——炉膛火焰,而是间接测量与之相关的延伸参数,如主汽温度、主汽压力等。事实上,炉膛火焰关联着所有工艺参数,关联着发电过程的各个环节和生产的安全、稳定和可靠。针对炉膛火焰的研究及其对炉膛火焰的燃烧监控,涉及到热动力学、燃烧学、图型学、信号处理、控制科学与技术等多学科领域,而且对发电本身也具有直接正向效应。对大型燃煤锅炉而言,炉内悬浮燃烧状态的火焰,是一种非常复杂的悬浮燃烧,它的工况是不稳定的。锅炉燃烧的安全性取决于火焰燃烧的稳定性,如果燃烧不稳定,炉内温度场不均匀,容易出现重大事故。因此开展炉膛火焰的研究具有重要的学术意义和显著的应用价值。炉膛火焰的信号获取是基于CCD传感器的二维视频信号,本文在了解和掌握图像处理方法,分析炉膛火焰的基本信息和图像预处理算法的基础上,运用图像处理技术对炉膛火焰展开全面的特性研究,并基于研究结果探索了对炉膛火焰的诊断技术和实时监控方法,具体开展了以下研究工作:(1)分析炉膛火焰图像的噪声和抑制技术,采取算术平均滤波和中值滤波算法有效实现图像去噪;研究了炉膛火焰图像的灰度特征、温度特征和相关的煤粉特性;全面研究炉膛火焰的温度和温度场测量技术,提出了基于景深温度场的误差修正方法;划分了火焰特征区域,确定了区域边缘特征及其特征区域所呈现的各个图型特征,全面研究了基于炉膛火焰特征的主要诊断技术,通过对比分析得出特征区温差、火焰平均温度、高温面积率、高温区圆度率和火焰质心偏移率等的实时数值有助于炉膛火焰的监控。(2)研究了火焰探测技术及其控制。全面研究炉膛火焰的图像处理算法,为控制策略提供有效的、准确的特征数值;研究了炉膛火焰燃烧控制的策略方法,包括控制对象模型识别、控制策略及其实现,提出了将基于二维图像的火焰特征、聚类特征和基于时基的火焰探测器燃烧状态进行融合,以此共同决策炉膛火焰的实时控制方法。(3)总结了炉膛火焰图像的操作、预处理、转换、处理技术、特征计算、策略研究和温度—色彩变换、伪彩色显示技术等,设计了炉膛火焰图像信息平台。基于图像采集和图像处理、计算、判断和控制,研制了实时监控DCS和底层的炉膛火焰测控基站,能完成无线图像传送功能并确保传送的实时性。采用本文技术构成的DCS能满足基于炉膛火焰的实时控制需求,使电厂生产更加安全、稳定和可靠。更多还原

【Abstract】 Power system is the security system of nation-building, and in the sustainable power generation technologies, coal power generation plays a major role. Coal-fired power generation technologies from the perspective of energy consumption, sub-critical unit is constantly replaced by supercritical units and ultra-supercritical unit because of low-power zero-emission and high power generation efficiency on equivalent coal consumption. However, in the angle of power generation technology, they are all coal-fired power generation. Coal-fired power generation includes the "link of the provide-powder," "burning part," "power generation" and other auxiliary links. The all power generation process is a Multi-energy conversion process with high capacity, large delay, physical and chemical nonlinear model parameters, which is real-time monitored by a power plant distributed control system ( DCS ).Plant DCS plays an important role in the plant production, is a widely used monitoring system in plant production. However, in the monitoring process of the power plant DCS, the basis of control strategy does not come from the most important technological object of generation- the furnace flame, but the relevant extension parameters, such as main steam temperature, main steam pressure, which are indirectly measured. In fact, the furnace flame is linked to all the technical parameters and associated with all aspects of power generation and the security, stability and reliability of production.Research of furnace flame and control of flame combustion involve a lot of technologies, such as thermal dynamics, combustion, graphics, signal processing and control. In the large coal-fired boilers, state of the furnace flame burning is a very complex suspension combustion, and its condition is unstable. Combustion flame safety depends on the stability of combustion. If the combustion is unstable, the furnace temperature field would be uneven, which may cause some major accidents. Therefore, the research of furnace flame has important significance and significant academic value.Flame signal acquisition is based on the two-dimensional CCD sensor video signals. In this paper, on the basis of grasping the image processing methods and analyzing the flame basic information and preprocessing algorithm, conducted a comprehensive study of flame image features by using image processing technology. Meanwhile, explored the diagnosis technology and real-time monitoring method of Flame, the specific tasks are as follows:Analyzed the flame image noise and the suppression, and took the arithmetic mean filter and median filter algorithm to realize the image denoising; studied the intensity characteristics, temperature characteristics and related coal properties of furnace flame. Comprehensively studied the temperature and temperature field measurement technique of furnace flame, proposed the error correction method based on the depth of field temperature field. Divided the flame region based on characteristics, determined the features of region edge and image characteristics presented in feature area and did a comprehensive study of main diagnosis based on furnace flame characteristics. Got the temperature of characteristics areas, the average flame temperature, high temperature area ratio, roundness rate of high temperature and the rate of flame centroid through Comparative Analysis, and the real-time data could be used to do the furnace flame monitoring.Researched the fire detection technology and its control. Comprehensively studied the furnace flame image processing algorithms and provided effective and accurate Numerical to control strategies. Studied the flame combustion control strategies, which includes the control object identification model, control strategy and its implementation. Proposed the integration method of flame characteristics based on two-dimensional image, cluster characteristics and time-based flame detector combustion state.Summarized the operation, preprocessing, conversion, processing, feature calculation, strategy and temperature - color transform, pseudo-color display technology of the furnace flame image, designed flame image information platform. Based on image acquisition and image processing, calculation, determine and control , developed the real-time monitoring DCS and the underlying flame monitoring base stations, which can complete the wireless image transmission function and ensure the real-time delivery. DCS constructed by the technology of this paper can meet the real-time control of furnace flame, and make plants produce more secure, stable and reliable.更多还原