Comparison of Machine Learning Algorithms for Engine Failure Prediction with Local and Cloud Monitoring

The document focuses on the issue of predictive maintenance in industrial environments, specifically aimed at transforming distributions within company departments. The objective is to program predictive maintenance to achieve a comprehensive improvement of processes, using a quantitative methodology based on data filtering and imputation techniques. In this context, various predictive models supported by Machine Learning were explored. Among the models analyzed, the random forest regression model was selected due to its ability to handle complex datasets and multiple variables. This model was preferred for its robustness in predicting patterns in non-linear data and its adaptability to industrial environments that often exhibit non-linear relationships between variables. The model’s choice was based on its demonstrated performance in improving filtering and imputation techniques, contributing to operational efficiency, and reducing economic costs in large-scale implementation. This methodological approach supported by Machine Learning and the specific use of the random forest regression model represent a significant contribution to the field of predictive maintenance. By applying strategies backed by artificial intelligence, the quality of industrial processes is improved, and resources are optimized, yielding promising results for operational efficiency in specific industrial settings.