基于多方向主元分析方法的间歇过程性能监视和故障诊断

将多方向主元分析（ＭＰＣＡ）技术应用于间歇生产过程的建模、过程性能监视和故障诊断，ＭＰＣＡ方法唯一需要的信息是过去成功间歇过程数据集合。作为一种有效的数据压缩和信息提取方法，ＭＰＣＡ方法大大降低数据空间结构的维数，消除变量之间的关联性，去除噪声，提高监视系统的鲁棒性，本文针对半导体生产过程中快速热退火间歇过程进行仿真实验研究。仿真结果表明：ＭＰＣＡ方法能够有效地监视间歇过程性能，及时准确诊断引起产

Batch Process Performance Monitoring and Fault Detection Based on Multiway Principal Component Analysis Method

This paper describes the modeling, process performance monitoring, and fault detection for rapid thermal annealing (RTA) batch processes using multiway principal component analysis (MPCA) method. Only information needed for the approach is historical database of past normal batches. As an efficient means of data compression and information extraction, MPCA method enormously reduces the dimensionality of data space structure, eliminate correlation among original variables and noises, and improve robustness of process monitoring. A RTA batch process in semiconductor manufacturing was studied, simulation results demonstrate that MPCA method is able to efficiently monitor performance of the RTA batch process, exactly detect faults which resulted in change of product quality.