Marcos GuillénJosé A. ParedesÓscar Camacho2026-03-222026-03-222010http://erevistas.saber.ula.ve/index.php/cienciaeingenieria/article/view/1123https://andeanlibrary.org/handle/123456789/61441This paper presents an approach for fault detection and diagnosis based on the wavelet transform. The proposed method is used to detect faults in chemical process instrumentation (valves and transmitters), but it can be extended to other kinds of processes. The proposed approach applies the wavelet transform on the online-monitored signal to map it into the wavelet domain. A feature extraction algorithm followed by a pattern recognition algorithm performed on the wavelet coefficients allows us to detect and classify faults present in the process instrumentation (valve or transmitter). Furthermore, having defined the faulty component and the kind of fault, two methods to estimate the percentage of breakdown are developed. Thus, the approach not only detects and classifies the fault but also yields an estimate of how serious the problem is.esWaveletFault detection and isolationWavelet transformPattern recognition (psychology)Artificial intelligenceDomain (mathematical analysis)Fault (geology)Computer scienceStationary wavelet transformWavelet packet decompositionarticle