Development of algorithms of statistical signal processing for the detection and pattern recognitionin time series: application to the diagnosis of electrical machines and to the features extraction in actigraphy signals
- Autores: Miguel Enrique Iglesias Martínez
- Directores de la Tesis: Jose A. Antonino Daviu (dir. tes.)
, Pedro Fernández de Córdoba Castellá (dir. tes.) , José Alberto Conejero Casares (dir. tes.) - Lectura: En la Universitat Politècnica de València ( España ) en 2020
- Idioma: inglés
- Tribunal Calificador de la Tesis: Nicolás Montes Sánchez (presid.) , Elies Fuster García (secret.) , Norge Cruz Hernández (voc.)
- Enlaces
- Tesis en acceso abierto en: RiuNet
- Resumen
Nowadays, the development and application of algorithms for pattern recognition that improve the levels of performance, detection and data processing in different areas of knowledge is a topic of great interest.
In this context, and specifically in relation to the application of these algorithms to the monitoring and diagnosis of electrical machines, the use of stray flux signals is a very interesting alternative to detect the different faults.
Likewise, and in relation to the use of biomedical signals, it is of great interest to extract relevant features in actigraphy signals for the identification of patterns that may be associated with a specific pathology.
In this thesis, algorithms based on statistical and spectral signal processing have been developed and applied to the detection and diagnosis of failures in electrical machines, as well as to the treatment of actigraphy signals.
With the development of the proposed algorithms, it is intended to have a dynamic indication and identification system for detecting the failure or associated pathology that does not depend on parameters or external information that may condition the results, but only rely on the primary information that initially presents the signal to be treated (such as the periodicity, amplitude, frequency and phase of the sample).
From the use of the algorithms developed for the detection and diagnosis of failures in electrical machines, based on the statistical and spectral signal processing, it is intended to advance, in relation to the models currently existing, in the identification of failures through the use of stray flux signals.
In addition, and on the other hand, through the use of higher order statistics for the extraction of anomalies in actigraphy signals, alternative parameters have been found for the identification of processes that may be related to specific pathologies.
