نوع مقاله : علمی-پژوهشی
نویسنده
دانشکده فنی و مهندسی، دانشگاه لرستان، خرم آباد، ایران
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسنده [English]
This paper presents an innovative online approach for univariate fault detection and alarm generation using the power of spectral analysis. The new approach utilizes a Spectral Stability Index (SSI) for simultaneous and sensitive monitoring of small and incipient changes in both the mean and variance of a process signal, which are common precursors to industrial alarm systems. The SSI framework accomplishes this by quantifying the deviation of the real-time power spectral density of the process signal from a reference model of normal operation. The method inherently carries an important advantage: it can detect low-level, developing faults much earlier than methods based on time-domain analysis. Its superior effectiveness is demonstrated through numerical and industrial case studies, where it significantly outperforms conventional methods, such as CUSUM, deadbands and delay timers, and more sophisticated advanced methods that have been recently introduced, such as serial and cascaded alarm systems. The results confirm the method's potential for substantially improving the responsiveness and accuracy of industrial alarm systems.
کلیدواژهها [English]
)