【作者】 乔宗良；

【导师】 徐治皋；

【作者基本信息】 东南大学 ， 动力工程及工程热物理， 2015， 博士

【摘要】 目前我国燃煤火电机组是大气污染物SO2排放大户,随着环保要求的提高,国家制定了严格的SO2排放标准,在现有的脱硫技术中,石灰石/石膏湿法烟气脱硫系统(WFGD)是当前我国使用最广泛的脱硫工艺,其本身也是一个高能耗的系统,在保证环保达标的同时,增加了火电企业的运营成本,随着电力行业节能减排工作的深入开展,挖掘脱硫系统的优化潜能,提高设计和运行水平具有重要的现实意义。本文针对WFGD系统的若干问题,以优化塔内流动特性、减少“石膏雨”排放和提高脱硫系统运行经济性为目的,对脱硫塔内流动和阻力特性、除雾器叶片结构优化设计、脱硫效率特性模型、基于脱硫系统整体可控成本最小展开的运行优化、以及基于聚类数据挖掘的脱硫系统目标工况获取和状态监测等方面展开研究。WFGD吸收塔内烟气和液滴的两相流动状态直接影响脱硫系统的性能,本文通过数值模拟的方法对某600MW机组配套湿法脱硫吸收塔内流场和阻力特性展开研究,烟气场采用标准κ-ε模型,液滴相采用离散颗粒模型(DPM),模拟结果表明,浆液喷淋对于烟气有良好的整流作用,加入喷淋后,塔内速度场和压力场变得均匀,整体阻力上升。同时,为了使塔内流动更加均匀,运用CFD技术对原有塔内构件进行优化设计研究,提出了优化烟气入口角度和增加托盘这两种可行的改造方案,详细对比分析两种方案实施后对塔内流场和阻力特性的影响。通过数值模拟分析折形和波形两种板型除雾器叶片的流动特性,分析了叶片的结构参数(包括叶片间距、叶片转折角和叶片高度等)和运行工况参数(包括进口烟气速度和液滴粒径等)对除雾性能的影响规律,通过比较显示,波形板相比折形板具有更高的除雾效率和较小压降。另外,提出了基于预数值计算的除雾器叶片结构优化设计方法,建立除雾器特性与叶片结构参数和运行工况参数间的最小二乘支持向量机(LSSVM)关系模型,在此基础上采用根据精英控制策略改进NSGA-Ⅱ算法,求解除雾器效率和压降两个目标Pareto解集,并优选出最佳的结构参数。针对脱硫系统中pH值传感器经常出现因堵塞和磨损造成的故障情况,提出基于改进“稀疏性”LSSVM算法脱硫吸收塔pH值软测量模型,该模型具有较高的精度并适用于在线使用。开展脱硫效率特性模型研究,提出了基于试验数据静态模型和改进AOSVR算法脱硫效率在线自校正模型,该模型具有较强在线自校正能力,敏感性分析研究结果表明所提出的在线建模方法能有效提高实际运行中脱硫效率的预测精度和泛化能力。结合脱硫效率预测模型,构建脱硫系统可控运行成本模型,研究了脱硫系统主要能耗设备的节能优化运行方法。将数据挖掘中的聚类算法引入到脱硫系统目标工况获取和浆液品质状态监测应用中,提出了基于信息熵和K均值算法混合迭代的动态模糊聚类算法(EKFCM),解决了模糊聚类的原型初始化参数问题,用该算法求解不同工况簇内最佳运行目标工况.组成目标工况库,可通过建立的LSSVM目标工况识别模型,确定当前外界运行条件下的最优目标工况。提出了基于浆液品质特征值提取和FCM聚类的浆液品质监测诊断方法,采用FCM聚类方法提取出浆液品质在各种状态的标准模式,然后按照最大隶属原则进行诊断,并提出了一种浆液品质模糊定量监测指数,可对实际运行过程中的浆液品质进行定量监测。更多还原

【Abstract】 At present, coal-fired power plants are the main SO2 emitters in China. With the improvement of environmental concerns, the government formulates the stricter SO2 emission standard. In the existing desulfurization technology, limestone/gypsum wet flue gas desulfurization system (WFGD) is the most widely used desulfurization process in our country. The desulfurization device is a high energy consumption system and increases the operating costs of the power plants while it meets standard of environmental protection. Along with the further work of energy-saving and emission reduction in the power industries, exploiting the optimization potentiality of desulfurization system and improving the design quality and operating level have the vital and practical significance. In this paper, several researches are carried out on a WFGD system to improve the operation reliability and economy, which including the internal flow and resistance characteristics of desulfurization absorption tower, optimization of mist eliminator vanes, characteristics model of desulfurization efficiency and operation optimization, condition monitoring for the desulfurization system based on data mining and so on.Two-phase flow between flue gas and droplets in absorption tower directly affects the function of desulfurization in WFGD. In this paper, the flow and resistance characteristics of a 600MW WFGD are studied. Standard k-ε model and discrete particle model (DPM) are separately selected for the flue gas and droplet phase. Simulation results show that the spray of slurry can rectify the flue gas flow. The velocity field and pressure field become more uniform after adding spray, but the total resistance increases. Furthermore, in order to make the flow field in absorption tower more uniform, some optimization design for original inner structure of tower by CFD are carried out. Two feasible schemes such as changing the entrance angle of flue gas and adding a salver are raised, and the role of two schemes are compared.The flow characteristics of the folded vane and corrugated vane eliminator are analyzed by numerical simulation. The laws of the structure parameters and operational parameters to mist eliminator performance are discussed, and the structure parameters include vane space, vane turning angle and vane length and operational parameters include inlet velocity of flue gas, droplet size. The corrugated plate has higher separation efficiency and lower pressure drop compared to the folded plate. A new optimization design method of the vane structure of eliminator is provided based on numerical calculation and the least squares support vector machine model (LSSVM). A prediction model for removal efficiency and pressure drop is established applying least square support vector regression from the results of numerical computation. Based on the prediction model, a mist eliminator parameters optimization model is created employing improved NSGA-II algorithm. The optimal solution and data analysis showed the model can direct the optimum design of mist eliminator and can overcome the numerical simulation consuming long time.Due to the failure situations caused by blocking and wear phenomenon for the pH sensor of WFGD, the soft sensor model for slurry pH value is proposed based on the improved sparseness LSSVM algorithm which has higher accuracy and is suitable for online measuring requirements. The research on characteristics model of desulfurization efficiency is carried out. An on-line residual prediction model of desulfurization efficiency is established based on the static model by test data and improved AOSVR algorithm, which has strong correction capability. The results of sensitivity analysis show this on-line desulfurization efficiency model can improve the prediction accuracy and generalization ability. Based on the desulfurization efficiency prediction model, the model of operation cost which can be controlled for desulfurization system is proposed and the optimization operation methods of the main energy-wasteful equipment of the desulfurization system are studied.The data clustering algorithm is introduced to the optimization operation and condition monitoring for WFGD. A dynamic fuzzy clustering method is presented based on iterative model of entropy and K-means (EKFCM) to select initialization parameters of the fuzzy cluster.The optimum objective condition is obtained by EKFCM algorithm in a certain operation condition, which composes the objective operation condition database.And the LSSVM model is established to ensure the optimum objective condition in the certain condition. Using FCM clustering algorithm, a model is presented to monitor the slurry quality based on the extraction feature of the slurry quality. The standard modes of the slurry quality are collected in the various states by FCM clustering algorithm and then it’s diagnosed according to the biggest subordinate principle. A fuzzy quantitative detection index of the slurry quality is put forward which can be used for analysis of the slurry quality.更多还原