水泥回转窑工作点的多变量系统辨识建模

摘要

The optimization modeling of the cement rotary kilns was investigated, as the precise mathematical model was needed in the realization of automatic control and optimization to the cement rotary kilns. As a high-temperature thermal equipment, the cement rotary kiln requires precise modeling because of its complex nonlinearity system. In order to solve the above problems, the time series models were introduced, and the Least Squares Method was applied to identify the parameters. The linear system models at and around the operating point were obtained, and the system modeling was realized though expert intelligent control software. Experimental results show that the normalized outputs of identified system fitts the real system outputs well, which demonstrates that the model can describe this multi-variable system at the operating point precisely and proves the effectiveness of the time series model identification algorithm.The optimization modeling of the cement rotary kilns was investigated, as the precise mathematical model was needed in the realization of automatic control and optimization to the cement rotary kilns. As a high-temperature thermal equipment, the cement rotary kiln requires precise modeling because of its complex nonlinearity system. In order to solve the above problems, the time series models were introduced, and the Least Squares Method was applied to identify the parameters. The linear system models at and around the operating point were obtained, and the system modeling was realized though expert intelligent control software. Experimental results show that the normalized outputs of identified system fitts the real system outputs well, which demonstrates that the model can describe this multi-variable system at the operating point precisely and proves the effectiveness of the time series model identification algorithm.%研究水泥回转窑优化建模问题,水泥回转窑自动控制优化的实现,需要以准确的数学模型作为基础.由于水泥回转窑为高温热工设备,其系统是复杂的非线性系统,要求准确建模.为解决上述问题,提出采用时间序列模型,应用最小二乘的参数寻优方法进行系统辨识,得到某生产线工作点及其附近的线性化模型,用专家智能控制系统软件实现建模.实验结果可见,系统的实际输出与辨识所得模型的输出曲线归一化后拟合良好,表明所得模型在工作点附近可以准确地描述多变量系统,验证了所提时间序列模型辨识建模算法的有效性.