Ingeniería en Sistemas, Electrónica e Industrial
URI permanente para esta comunidadhttp://repositorio.uta.edu.ec/handle/123456789/1
Examinar
1 resultados
Resultados de la búsqueda
Item type:Ítem, Sistema de preprocesamiento de datos sensoriales utilizando Python en la gestión de mantenimiento predictivo de bombas hidraúlicas en la producción de crudo(2026-01) ;Rea Ramírez Carmen Alexandra ;Álvarez Mayorga Edison HomeroUniversidad Técnica de Ambato. Facultad de Ingeniería en Sistemas, Electrónica e Industrial. Carrera de Tecnologías de la InformaciónPredictive maintenance management in the oil industry heavily depends on the quality, consistency, and reliability of sensor data, which often appear as raw streams with temporal irregularities, information loss, and uncontrolled variations that hinder analysis. To address this challenge, the purpose of this work was to implement a sensor data preprocessing system using Python to support predictive maintenance of hydraulic pumps in crude oil production. The CRISP-DM methodology was adopted, structuring the process from problem understanding to the final preparation of data for integration into predictive models. The developed system comprises three main phases: quality diagnosis and labeling, temporal reconstruction with controlled imputation, and statistical feature generation. Each phase produces and stores its results in independent tables, ensuring traceability and organized use of information. As a result, the system significantly improved the quality and usefulness of sensor data. The impact of data quality was demonstrated by training two models on the Phase 1 dataset, where Logistic Regression achieved an accuracy of 34.92%, showing its limited ability to handle noise, outliers, and temporal variability, while Random Forest reached 99.57%, confirming that higher data quality directly influences model learning and performance. The final system is available for use by the research team and for future extension in subsequent work.
