论文部分内容阅读
化工过程中缓变故障的存在,会影响装置长周期稳定的运行,严重的直接造成装置生产能力下降、生产成本增加。针对化工过程中难以监测的缓变故障,提出1种新的多变量统计过程的监测方法。把传统的单变量累计和控制图(CUSUM)扩展为多变量的形式,通过累计作用提取过程的微小变化,并与小波变换在定尺度下提取测量变量决定性特征的特性,以及传统的主元分析(PCA)去除变量间关联的优势相结合,构成新的MCUSUM-MSPCA方法。通过仿真研究TE过程,证明此方法可行和有效,极大地改善了监测过程缓变故障的效果。与PCA方法相比,MCUSUM-MSPCA方法能在不同频率范围内,有效、及时地监测到过程中的缓变故障,提高了过程监测的灵敏性,为操作人员在线排除故障提供了可能,从而可降低操作成本,保证产品质量。
The existence of slow failure in the chemical process will affect the stable operation of the plant for a long period of time, resulting in a serious decrease in the production capacity of the plant and an increase in the production cost. Aiming at the slowly changing faults which are difficult to be monitored in the chemical process, a new monitoring method of multivariate statistical process is proposed. The traditional univariate cumulative control chart (CUSUM) is extended to a multivariate form. Through the small changes of the cumulative extraction process and the characteristics of the deterministic features of the measured variables extracted from the wavelet transform, the traditional principal component analysis (PCA) to remove the association between variables to form the new MCUSUM-MSPCA method. Through the simulation study of the TE process, this method is proved to be feasible and effective, which greatly improves the effect of slowing down the monitoring process. Compared with the PCA method, the MCUSUM-MSPCA method can effectively and timely monitor the gradual faults in the process in different frequency ranges, improves the sensitivity of the process monitoring and provides the possibility for the operator to troubleshoot on-line so that Reduce operating costs and ensure product quality.