Condition‐based monitoring (CBM) as a new control scheme suggests characterizing the machine condition and triggering the corresponding control actions. CBM includes prognostic and diagnostic modules. In this study, the framework of equipment deterioration modeling and monitoring for batch processes is proposed with two objectives in the semiconductor industry. The first one is to characterize equipment behavior by exploiting the temporal data of batch processes. The second one is to model the deterioration trend with the most related causes. With the best‐fitted mother wavelet, wavelet packet decomposition transforms the temporal data into macro and micro level domains to identify two types of deterioration. The determinant of the correlation matrix of the decomposed signals is calculated as the equipment condition, and the factors that account for the deterioration are identified through a stepwise searching algorithm. A case study shows that the proposed methodology can identify influencing factors and model deterioration.

中文翻译：

---

半导体制造中的设备退化建模和原因诊断

基于状态的监视（CBM）作为一种新的控制方案，建议表征机器状态并触发相应的控制动作。CBM包括预后和诊断模块。在这项研究中，针对半导体行业中的两个目标，提出了用于批处理过程的设备劣化建模和监控框架。第一个是通过利用批处理过程的时间数据来表征设备行为。第二个是用最相关的原因来模拟恶化趋势。利用最适合的母子波，小波包分解将时间数据转换为宏观和微观域，以识别两种类型的恶化。计算分解信号的相关矩阵的行列式作为设备条件，并通过逐步搜索算法确定导致变质的因素。案例研究表明，所提出的方法可以识别影响因素并建立模型恶化模型。