Accurate detection and isolation of an incipient fault is crucial for efficient and optimal operation of industrial processes. In the area of statistical process monitoring (SPM), reconstruction-based methods are widely used for fault isolation purposes. Though possessing several merits, most existing conventional reconstruction methods are not efficient in isolating incipient faults that have small magnitudes. This paper proposes a new method called exponential smoothing reconstruction (ESR) for incipient fault isolation. The proposed ESR introduces the exponentially weighted moving average (EWMA) technique to conventional reconstruction methods. Fault isolability analysis of the ESR is provided, which demonstrates its efficiency in isolating incipient faults, compared with conventional reconstruction-based methods. Several remarks on the use of ESR, such as selection of the smoothing parameter and some practical issues, are discussed as well. Finally, simulation studies on a numerical example and a continuous stirred tank heater (CSTH) process are used to demonstrate the effectiveness of the proposed ESR method, in comparison with conventional approaches.

中文翻译：

---

早期故障隔离的指数平滑重构方法

准确检测和隔离初期故障对于工业流程的高效和最佳运行至关重要。在统计过程监视（SPM）领域，基于重建的方法被广泛用于故障隔离。尽管具有几个优点，但是大多数现有的常规重建方法在隔离具有小幅度的初期断层方面是无效的。本文提出了一种用于初始故障隔离的称为指数平滑重构（ESR）的新方法。拟议的ESR将指数加权移动平均（EWMA）技术引入了传统的重建方法。提供了ESR的故障可隔离性分析，与传统的基于重建的方法相比，该分析证明了ESR在隔离初期故障方面的效率。还讨论了有关使用ESR的一些注意事项，例如平滑参数的选择和一些实际问题。最后，通过对数值示例和连续搅拌釜加热器（CSTH）工艺的仿真研究，与传统方法相比，证明了所提出的ESR方法的有效性。