化工过程强制去噪小波滤波最优分解层数的选取

Choice of the Optimal Decomposition Level of Wavelet Forced De-noising in Chemical Industry

强制去噪小波滤波能有效降低化工过程测量数据的误差,然而该滤波对误差的校正效果与分解层数有很大关系。因此提出用高频系数和低频系数的相关系数和相邻两层高频系数方差之比共同判断最优分解层。通过总结2个化工数据校正例子的计算结果,得出了相关系数和方差比的阈值。在庚烷催化反应模型中数据校正的结果表明,所提出的2个判断参数能有效确定强制去噪滤波的最优分解层。

Wavelet forced de-noising have a good performance on chemical measurements error reduction,and the error reduction effect is related to the decomposition levels.In the paper,two judgement parameters,the ratio of variance of high frequency coefficients and the correlation coefficients between high frequency coefficients and low frequency coefficients,are proposed to judge the optimal decomposition level,and the thresholds of the two judgement parameter are given by summary the calculation results from two chemical data reconciliation examples.The performance of two judgement parameters is illustrated by applied in a heptane catalytic reaction model,results show that two judgement parameters give promising results for the choice of the optimal decomposition level.