Feature Selection Based on Genetic Programming for Fault Prognosis in Dynamical Systems

This study explores the application of Genetic Programming (GP) for feature selection in the context of fault prognosis, with an emphasis on predicting the Remaining Useful Life (RUL) of a Computer Numerical Control (CNC) milling machine. The research focuses on improving the efficiency of RUL prediction, a crucial challenge for modern industry. Utilizing real data from the 2010 PHM Society Challenge, which provides a dataset from sensors monitoring the operational state of the machine, the research aims to develop an innovative and effective approach. To achieve this goal, a hybrid approach was developed that combines particle filters, fuzzy systems, and a feature selection process based on Genetic Programming. This technique stands out for its ability to identify the most relevant features for the predictive model, thereby increasing the accuracy and robustness of the predictions. The effectiveness of the proposed technique was evaluated through comparisons with traditional methods in the literature, using metrics such as monotonicity, trend, prognosability, and prediction accuracy, highlighting the superiority of GP in feature selection. The results show that the technique is capable of detecting complex patterns and signs of machine wear, contributing to fault prognosis and more accurate RUL prediction. The application of GP, therefore, proves promising for optimizing predictive maintenance and reducing costs associated with machine failure. The research highlights the potential of GP as a tool for feature selection in industrial scenarios, emphasizing its importance in the development of more efficient predictive maintenance systems. These advances are crucial for optimizing machine operation, reducing maintenance costs, increasing the reliability of industrial processes, and promoting the safety and sustainability of operations. The research contributes to the advancement of knowledge in fault prognosis and facilitates new innovative and practical solutions for the industry.