Predictive Maintenance of a Metro's Air Compressor

A Metro air compressor turns power into potential energy by compressing and storing air in a tank. When the pressure reaches its limit, the compressor shuts down, storing the compressed air until it is needed for various uses. Early prediction enables proactive decision-making, which helps to avoid unfavourable outcomes before they occur. It also optimizes resource allocation to ensure efficiency and effectiveness across several processes. The proposed method is a machine learning-based strategy for predictive maintenance of a metro train's air compressor. The MetroPT -3 dataset was used, which includes parameters for pressure, temperature, motor current, air intake valves, and so on. Seven analogue and eight digital signals. The technique entails data preprocessing, anomaly identification using an exclusion forest classifier, training multiple machine learning models, and integrating the chosen model with the rest of the system. Gradient-boosting models outperformed linear models in terms of accuracy, but they were slower to compute. They conclude that predictive maintenance benefits metro compressors and has the potential to revolutionize equipment maintenance in general.