This study presents a comprehensive evaluation of various machine learning algorithms for Predictive
Maintenance (PdM) in industrial equipment, focusing on their effectiveness in predicting equipment
failures and maintenance needs. Supervised learning techniques such as Linear Regression, Decision
Trees, Random Forest, Support Vector Machines (SVM), and Neural Networks are analyzed alongside
unsupervised methods like K-Means and DBSCAN, as well as deep learning models like Recurrent
Neural Networks (RNN) and Long Short-Term Memory (LSTM). The results demonstrate that while
traditional algorithms like Linear Regression and Decision Trees provide a baseline performance,
advanced models such as LSTM and ensemble methods like Random Forest significantly enhance
predictive accuracy, precision, and recall. The study underscores the importance of selecting the
appropriate algorithm based on the specific characteristics of the data and the industrial context, with
deep learning and ensemble approaches emerging as the most effective for complex, high-dimensional
data environments.
