【作者】 黄群星；

【导师】 池涌； 严建华； 岑可法；

【作者基本信息】 浙江大学 ， 工程热物理， 2005， 博士

【摘要】 随着世界经济的发展,能源与环境的问题越来越严峻,四角切圆煤粉燃烧锅炉作为我国电力供应的主力机组,是国内石化燃料的主要消耗者和大气污染物的重点排放者,保证炉内燃烧过程的安全性,稳定性和经济性对于国民经济的可持续发展有着重要的现实意义。然而由于锅炉燃烧过程的复杂性以及现场测量条件的限制,导致常规的接触式测量方式无法用于炉内燃烧过程的分析和诊断。近几年来,随着光电检测技术的发展,基于炉内燃烧介质本身辐射特性的辐射能监测与分析技术成为目前国内外炉内燃烧诊断的一个新的研究方向。本课题的研究目的是在详细考察炉内介质辐射特性的基础上,利用辐射能与锅炉燃烧过程参数的对应关系,结合人工智能技术为优化炉内燃烧提供可视化的指导依据和可行的解决方案。 基于瑞利和Mie散射理论,对炉内不同成分的固体颗粒辐射特性进行了研究,利用逐线计算模型和谱带模型对炉内主要气体成分的吸收系数进行了计算讨论,并考察了气体吸收谱带的压力和温度增宽效应。同时我们还计算了固体粒子云和气固混合介质的辐射特性,并通过实验分析了基于多波长消光法的遗传算法粒度谱重建和基于可调谐激光器的气体浓度和温度测量技术。 在介质辐射特性计算的基础上,对炉内辐射传递进行了数值模拟研究。利用FLUENT等CFD数值模拟软件计算得到炉内介质浓度与温度的分布,建立了三维逆向Monte Carlo辐射传热计算方法,并模拟计算了四角切圆煤粉炉内不同高度上CCD探测器所接受辐射能的变化特性。 提出了基于神经网络的快速映射算法和基于插值滤波反投影的快速算法来满足炉内燃烧过程分析的实时性要求。 研究了人工智能推理技术在炉内燃烧诊断分析中的应用,建立了燃烧诊断专家知识库,对智能推理和搜索技术在燃烧诊断的应用进行了探讨,研究了炉内燃烧过程安全性,稳定性和经济性的分析方法。 针对目前燃烧过程燃料控制中存在的时滞性问题,对基于辐射能的燃烧优化控制进行了深入实验和可行性研究。通过回归分析得到了辐射能与锅炉运行参数之间的对应关系。计算了燃烧过程传递函数的时间常数,并给出了以辐射能作为燃料调整提前反馈量的逻辑控制方案。更多还原

【Abstract】 As the world economic growing, energy and environment problems have been more serious than ever especially for our developing countries. As the primary coal-firing power plant boiler model, 300MWe corner-firing tangential boilers have been the mainly fossil fuel consumer and air contaminations producer. The stability, efficiency and safety operation of the boiler combustion process is all-important for our country’s continual economic development. During the last several years many optical measurement and analysis techniques have been introduced for non-invade boiler internal combustion process diagnostic, such as boiler-section-temperature distribution reconstruction based on flame images. And also some artificial intelligence techniques have been used for combustion optimizing such as neural networks and etc. The object of this paper is to establish an expert system for boiler combustion operation optimizing based on the radiative property of the boiler inside particulate media integrated with boiler operation parameters such as boiler load, drum pressure and also.The radiative property of the solid particles inside the boiler were calculated and discussed based on Rayleigh and Mie scattering theories. The affection of the complex refractive index for different radius distribution particle clouds was analysis. The band model and line-by-line model have been used to calculated the absorption coefficient of molecular gases, and also the temperature and pressures band widen effect were discussed. Based on the discussed results, multi-wavelength extinction method integrated with modified evolution program was provided for particles radius distribution sizing. And a diode tunable laser detector was used to measure the gas concentration by gas molecular spectrum-line broaden mechanism.In this paper , the backward Monte Carlo method was improved to analysis the radiative intensity received by the detector located at the boiler water cooler walls based on the output of the CFD simulationThe BP neural network’s fast mapping techniques and interpolated filtered back projection algorithm were introduced for on-line section temperature-distribution diagnostic.In this paper an expert system was provided for combustion optimizing and diagnostic based on boiler flame image information integrated with boiler operation parameters through object oriented method and self-organization neural networks. The implementation of the expert system on 300MWe boiler was discussed.The last part of this paper discussed boiler fuel control optimizing strategy based on the radiative intensity detected by photon-electronic sensor. And some in situ experiments were done to evaluate the feasibility of the optimizing strategy.更多还原