水泥篦冷机高温渗流换热规律及温度预测模型研究

发布时间：2019-01-30 13:49

【摘要】：水泥篦冷机是水泥生产线中冷却高温水泥熟料和回收热量的关键设备,其换热效率的高低直接影响熟料质量和水泥生产能耗。因此研究篦冷机内高温熟料层的气固换热机理及温度预测模型,对篦冷机稳定运行、提高热回收效率、降低熟料生产能耗、减少尾气排放,具有重要的理论意义和社会价值。本文针对目前篦冷机气固换热研究的局限性,建立了考虑高温熟料颗粒间热辐射效应的篦冷机非稳态高温渗流换热模型,并提出了相应流热耦合求解算法,给出了熟料颗粒层非均匀分布下的高温水泥熟料渗流换热规律。在此基础上,提出了双并联双隐含层快速学习机(Double Parallel and Double Hidden Layers Fast Learning Machine,DDFLM),建立了融合高温渗流换热机理与DDFLM的篦冷机混合温度预测模型。具体研究工作如下:首先,将多孔介质渗流换热理论引入到水泥熟料层换热研究中,并通过实验给出料层内颗粒直径与孔隙率的分布情况;在熟料颗粒辐射传热研究中,将高温水泥熟料颗粒等效为光学厚介质,推导热传导与热辐射综合作用下的颗粒间传热系数;基于非线性达西定律与局部非热平衡理论,建立考虑堆积颗粒间辐射传热效应的一维、二维篦冷机非稳态高温渗流换热模型。其次,基于有限差分法针对所建立一维模型与二维模型,分别进行求解区域与数学模型的离散化,建立各维度的流热耦合求解算法,完成篦冷机非稳态高温渗流换热模型的流热耦合求解;给出非均质高温水泥熟料层的气固温度分布,并验证所建模型与所提出算法的正确性;通过数值模拟研究料层颗粒特性与篦冷机主要工作参数对高温熟料冷却过程的影响,获得篦冷机非均质高温水泥熟料层的渗流换热规律。再次,为了克服非稳态高温渗流换热模型对篦冷机温度预测所产生的误差,在超限学习机的基础上提出了双并联双隐含层快速学习机,通过典型的回归问题测试所提出的DDFLM神经网络的有效性,并基于DDFLM进行非稳态高温渗流换热模型的误差辨识研究,将DDFLM作为非稳态高温渗流换热模型误差的估计器,建立融合高温渗流换热机理与DDFLM的篦冷机混合温度预测模型。最后,利用篦冷机实际生产数据与高温熟料堆积体冷却实验平台,分别进行篦冷机内测点温度的预测实验与高温水泥熟料堆积体强制风冷的熟料温度预测实验,验证本文所建立的融合高温渗流换热机理与DDFLM的混合温度预测模型对于篦冷机内测点温度预测与料层温度直接预测的可行性。
[Abstract]:Cement grate cooler is the key equipment for cooling high temperature cement clinker and recovering heat in cement production line. Its heat transfer efficiency directly affects clinker quality and cement production energy consumption. Therefore, it is of great theoretical significance and social value to study the gas-solid heat transfer mechanism and temperature prediction model of high temperature clinker layer in grate cooler, which is of great theoretical significance and social value for steady operation of grate cooler, improvement of heat recovery efficiency, reduction of energy consumption in clinker production and reduction of tail gas emission. In view of the limitation of the research on gas-solid heat transfer of grate cooler at present, a non-steady heat transfer model of grate cooler considering the effect of heat radiation between high temperature clinker particles is established, and the corresponding fluid-heat coupling algorithm is proposed. The percolation heat transfer law of high temperature cement clinker under non-uniform distribution of clinker granular layer is given. On this basis, a double parallel double hidden layer fast learning machine (Double Parallel and Double Hidden Layers Fast Learning Machine,DDFLM) is proposed, and a hybrid temperature prediction model of grate cooler is established, which combines the heat transfer mechanism of high temperature seepage flow and DDFLM. The specific research work is as follows: firstly, the theory of percolation heat transfer in porous media is introduced into the heat transfer study of cement clinker layer, and the distribution of particle diameter and porosity in the layer is given through experiments. In the study of radiative heat transfer of clinker particles, the high-temperature cement clinker particles are equivalent to optical thick medium, and the heat transfer coefficient between particles under the combined action of heat conduction and heat radiation is deduced. Based on the nonlinear Darcy's law and the local non-thermal equilibrium theory, a one-dimensional and two-dimensional heat transfer model of unsteady high temperature percolation of grate cooler considering the radiation heat transfer effect between stacked particles is established. Secondly, based on the finite difference method, the solution region and mathematical model are discretized respectively for the one-dimensional model and the two-dimensional model, and the fluid-heat coupling algorithm for each dimension is established. The fluid-heat coupling solution of the unsteady high temperature percolation heat transfer model of grate cooler is completed. The gas-solid temperature distribution of heterogeneous high-temperature cement clinker is given, and the correctness of the established model and the proposed algorithm is verified. The effects of particle characteristics and main working parameters of grate cooler on the cooling process of high temperature clinker were studied by numerical simulation, and the seepage heat transfer law of heterogeneous high temperature cement clinker layer of grate cooler was obtained. Thirdly, in order to overcome the error caused by the unsteady heat transfer model of high temperature seepage to the temperature prediction of grate cooler, a double-parallel double-hidden layer fast learning machine is proposed based on the over-limited learning machine. The validity of the proposed DDFLM neural network is tested by a typical regression problem, and the error identification of the unsteady high temperature seepage heat transfer model is studied based on DDFLM. DDFLM is used as the estimator of the error of the unsteady high temperature seepage heat transfer model. The mixing temperature prediction model of grate cooler combined with high temperature percolation mechanism and DDFLM was established. Finally, based on the actual production data of grate cooler and the cooling test platform of high temperature clinker, the prediction experiment of temperature at the test point inside the grate cooler and the prediction experiment of clinker temperature by forced air cooling of high temperature cement clinker accumulator are carried out, respectively. The feasibility of combining the heat transfer mechanism of high temperature percolation with the mixed temperature prediction model of DDFLM for the temperature prediction of test points in grate cooler and the direct prediction of material layer temperature is verified in this paper.
【学位授予单位】：燕山大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：TQ172.622.4

【相似文献】