【作者】 李海滨；

【导师】 刘彬；

【作者基本信息】 燕山大学 ， 电路与系统， 2006， 博士

【摘要】 能源与环境面临的严峻现状给水泥行业提出了新的要求。提高水泥熟料的冷却控制技术是实现节能降耗的关键,准确的篦冷机熟料控制模型又是提高冷却控制技术的前提。本课题在分别研究并实现熟料温度、厚度在线测量的基础上,从分析熟料的高温流动特性入手,以熟料微元换热方程为纽带,建立篦冷机熟料控制模型,通过实验验证控制模型的正确性,针对实验结果的误差设计了补偿算法,提高了控制模型的精度。具体工作包括:提出基于CCD比色测温原理的非接触测量篦冷机熟料温度的方法,建立篦冷机熟料温度测量模型。以提高测温模型在被测对象温度范围内的平均灵敏度为目标,解决了测温模型中波长与带宽最优解问题,给出了确定波长与带宽的约束方程,并针对某水泥厂的3500t/d篦冷机设计了实验系统,结果验证了测温系统的有效性。采用基于仿生原理的“双目”立体视觉测量理论实现篦冷机熟料厚度测量,建立篦冷机熟料厚度测量模型。设计基于二进小波变换的子线段匹配方法,通过求解样条二进小波变换系数在不同尺度下的模极大点确定不同尺度下信号急速变化点位置,采用由粗到细的匹配策略匹配端点,再对端点之间的每一个子线段采用顺序一致的原则,进行点对点匹配。仿真结果证明该方法可比原有的归一化协方差相似性测度匹配方法具有更高的准确性与快速性。从篦冷机熟料的高温流动特性入手,确定熟料在流动过程中的填充状态,分析熟料厚度与颗粒直径波动对控制模型影响,提出对篦冷机熟料采用三维分割的思想进行建模;设计了基于龙贝格变步长的三维分割算法,对z方向的分割次数进行优化;分析流动状态下熟料微元的质量平衡与热平衡关系,分别沿x、y、z方向递推,建立篦冷机熟料整体控制模型。针对控制模型进行实验分析,实验结果验证了控制模型的正确性,并指出控制模型在x方向存在误差积累效应。提出了一种基于自适应局部微调遗传算法(Genetic algorithm with adaptive local adjustment)优化的RBF网络进行控制模型误差补偿的方法;设计了一种自适应局部微调遗传算法,提高优化过程收敛速度与解的精度;针对熟料温度误差问题构建了优化的RBF网络,利用实际数据来学习训练网络权值,进而补偿篦冷机熟料控制模型,实验结果证明,补偿算法可以将控制模型误差积累问题彻底解决。更多还原

【Abstract】 It poses new challenge for cement industry in the current serious energy and environment requirements. Increasing cooling control technology of cement clinker is the key for saving energy sources and reducing pollution. Furthermore, the accurate clinker control model is the precondition of improving the coolling techonology. Firstly, the onliner measurement methods for clinker temperature and thickness are studied in this dissertation. Then, explointing the fluxion characteristic under high temperature, the clinker control model for grate cooler with the clinker’s differential heat exchange equation is constructed. Finally, the correctness of our proposed model is valiated by experiment. Furthermore, a compensation algorithm for the error of experiment result is proposed, which improves the accuracy of control model. These work can be detailed as the following:A measurement scheme using CCD noncontacting in the clinker temperature field measurement is proposed and CCD colorimetric temperature measurement model is constructed. The objective is to improve the average sensitivity in the temperature range of measured object for the temperature measurement model. The restriction equation between wavelength and bandwidth is presented, which solves the wavelength and bandwidth optimal problem in the temperature measurement model. Furthermore, an experimental system for one cement factory’s 3500t/d grate cooler is designed. The experiment’s result has proved the validity of the temperature measurement system.The thickness measurement is implemented using“binocular”stereo measurement theory based on bionics. Subline matching method based on binary system small wave transform is proposed, which expliots the character that the model maximum point of spline binary small wave transform coefficient at different scales can provide position information of signal’s fast changing point. The matching scheme from thick to thin for point matching is adapted. Then, the point to point matching is calculated with the sequence consistent principle in each subline. The simulation result validates that the proposed method has higher accurateness and faster than that of unitary covariance similar measure matching method.Exploiting the high temperature fluxion character of grate cooler clinker, the filling state during entire colling process can be determined. Then impact on the control model can be analyzed introducing by the fluctuating of the clinker’s thinkness and diameter. a modeling method with three dimensions division is proposed. The Longbeige step-changing dichotomy is used to optimize the division time toward z direction. Then the mass balance and heat balance analysis for every cell derived from clinker’s three dimentions division is presented. Finally, the clinker temperature field’s unitary control model with deducing in x, y, z direction separately is constructed.With the experiment analysis for the proposed control model, the accuracy of the proposed control model is validated. Furhermore, the existing of the accumulating error problem in the x direction is pointed out. A compensate scheme to the accumulating error is proposed, which is based on the genetic algorithm with adaptive local adjustment usig RBF network. A genetic algorithm with adaptive local adjustment to improve the convergence speed of optimal process and solution accuracy is designed. With the algorithm, a RBF network with optimize structure for the problem of clinker temperature error is given. The weight setting of RBF network can be learned by the practical data. Then the designed RBF network can be used to compensate the clinker control model of grate cooler. The experiment results validate that the compensation algorithm can solve the accumulating error problem in control model fully.更多还原