Thesis PHM-covariables: mapeo de consumo de combustible según visión artificial y despacho
No Thumbnail Available
Date
2025-11-26
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Industrial
Departament
Campus
Campus Santiago Vitacura
Abstract
La condición superficial de los caminos mineros es un factor crítico para la operación de equipos de transporte en faenas de gran escala, ya que influye directamente en su desempeño, seguridad y demanda de mantenimiento. Variaciones en la rugosidad del camino afectan la estabilidad del vehículo, incrementan los esfuerzos mecánicos y pueden provocar consumos anómalos de combustible, repercutiendo tanto en la continuidad operacional como en los costos asociados. A pesar de su relevancia, el monitoreo de la condición del camino suele basarse en inspecciones visuales o mediciones puntuales, las cuales no permiten capturar adecuadamente la dinámica de deterioro ni anticipar escenarios de riesgo operativo. En este contexto surge la necesidad de contar con un modelo capaz de representar la evolución del deterioro en función de variables operacionales reales, integrando de forma explícita la rugosidad del camino, el estado de carga del camión, la pendiente del tramo y el uso acumulado. Este trabajo responde a dicha necesidad mediante el desarrollo de un modelo predictivo basado en Proportional Hazards Model (PHM) con evolución Markoviana de la rugosidad, que permite estimar la confiabilidad condicional y la vida útil remanente (RUL) del camino bajo distintos escenarios operacionales. El objetivo general de esta memoria es formular, implementar y evaluar un modelo PHM Markov que caracterice el deterioro superficial del camino en función de su uso acumulado y severidad operativa, entregando métricas de confiabilidad, vida remanente y fronteras de intervención basadas en riesgo. Los resultados muestran diferencias claras entre escenarios: condiciones benignas presentan mayor estabilidad y RUL prolongado, mientras que escenarios de alta severidad exhiben deterioro acelerado y reducciones significativas en la confiabilidad. Se concluye que (...).
The surface condition of mining haul roads is a critical factor influencing the performance, safety, and maintenance requirements of large-scale truck fleets. Variations in road roughness affect vehicle stability, increase mechanical loading, and may trigger abnormal fuel consumption events, impacting operational continuity and associated costs. Despite its importance, road condition assessment in mining is often based on visual inspections or isolated measurements that do not capture degradation dynamics or anticipate operational risks. This motivates the need for a modeling framework capable of characterizing surface deterioration as a function of real operational conditions, explicitly integrating road roughness, truck loading state, segment slope, and accumulated usage. To address this gap, this thesis develops a predictive framework based on the Proportional Hazards Model (PHM) with Markovian evolution of roughness, enabling the estimation of conditional reliability and Remaining Useful Life (RUL) under heterogeneous operating scenarios. The main objective is to formulate, implement, and evaluate a PHM–Markov model that characterizes road surface degradation in terms of accumulated usage and operational severity, providing reliability metrics, RUL estimates, and risk-based intervention boundaries. The results show clear differences across scenarios: benign conditions exhibit stable reliability and extended RUL, whereas severe conditions—such as uphill loaded travel—display accelerated degradation and sharp reductions in both R(u) and RUL. Additionally, risk-based intervention boundaries derived from reliability thresholds delineate safe operating zones, warning regions, and immediate intervention areas. It is concluded that (...).
The surface condition of mining haul roads is a critical factor influencing the performance, safety, and maintenance requirements of large-scale truck fleets. Variations in road roughness affect vehicle stability, increase mechanical loading, and may trigger abnormal fuel consumption events, impacting operational continuity and associated costs. Despite its importance, road condition assessment in mining is often based on visual inspections or isolated measurements that do not capture degradation dynamics or anticipate operational risks. This motivates the need for a modeling framework capable of characterizing surface deterioration as a function of real operational conditions, explicitly integrating road roughness, truck loading state, segment slope, and accumulated usage. To address this gap, this thesis develops a predictive framework based on the Proportional Hazards Model (PHM) with Markovian evolution of roughness, enabling the estimation of conditional reliability and Remaining Useful Life (RUL) under heterogeneous operating scenarios. The main objective is to formulate, implement, and evaluate a PHM–Markov model that characterizes road surface degradation in terms of accumulated usage and operational severity, providing reliability metrics, RUL estimates, and risk-based intervention boundaries. The results show clear differences across scenarios: benign conditions exhibit stable reliability and extended RUL, whereas severe conditions—such as uphill loaded travel—display accelerated degradation and sharp reductions in both R(u) and RUL. Additionally, risk-based intervention boundaries derived from reliability thresholds delineate safe operating zones, warning regions, and immediate intervention areas. It is concluded that (...).
Description
Keywords
Caminos mineros, Deterioro superficial, Modelo de Riesgos Proporcionales (PHM), Confiabilidad condicional, Mantenimiento predictivo, Riesgo operacional, Gestión de activos, Costos de mantenimiento
