Desarrollo de un modelo de análisis estadístico y de aprendizaje automático de las señales de los sensores de oxígeno en el vehículo Jeep New Compas año 2025

Abstract

The present study aims to develop a statistical analysis and machine learning model applied to the signals generated by oxygen sensors (Lambda sensors) in the 2025 Jeep New Compass vehicle, equipped with a 1.3 Turbo GSE engine. Advanced automotive diagnostic equipment and techniques for the real-time analysis of the PiDs were used, considering the importance of these sensors in the regulation of the stoichiometric mixture and in the verification of the efficiency of the catalyst within the emission control system. The research was carried out by collecting real data under operating conditions, using tools that record signals from sensors located before ("pre") and after ("post") the catalyst. For the analysis of the signals, the machine learning algorithm is applied, including Random Forest (supervised). This method allows for the analysis of short- and long-term fuel trims, providing a more accurate evaluation of the system's performance under different scenarios. The results showed the feasibility of building a predictive model based on machine learning from real-time data from lambda sensors. In addition, a significant correlation was found between signal patterns and engine operating conditions, which can be used to implement preventive diagnostic systems. This research contributes to the development of intelligent solutions aimed at optimizing maintenance and extending the life of components in vehicles with modern electronic control technology such as the 2025 Jeep Compass.