论文部分内容阅读
许多工业过程中获得的输入输出数据在时间、空间上是高度相关的,且含有测量噪声。针对此类数据,提出采用动态PLS方法,最大化输入和输出变量矩阵之间的协方差,保留大部分有用信息,去除测量噪声,把高维数据空间降维,建立较为精确的工业过程数学模型。提出对MIMO系统进行研究,推导了可直接用于控制的动态PLS模型数学公式。对一个工业过程实例进行仿真,分别应用动态PLS回归和线性回归MLR方法,得出的结果经分析验证了此算法的有效性。
The input and output data obtained in many industrial processes are highly correlated in time and space, and contain measurement noise. For such data, the dynamic PLS method is proposed to maximize the covariance between the input and output matrix, retain most of the useful information, remove the measurement noise, reduce the dimension of high-dimensional data space, and establish a more accurate mathematical model of industrial process . The research of MIMO system is proposed and the mathematical formula of dynamic PLS model which can be directly used for control is deduced. An example of an industrial process was simulated. The dynamic PLS regression and linear regression MLR method were respectively used. The results obtained were verified by the analysis of the effectiveness of this algorithm.