Al-Said, S.Findik, O.Assanova, B.Sharmukhanbet, S.Baitemirova, N.2024-09-292024-09-2920242198-4182https://doi.org/10.1007/978-3-031-51997-0_11https://hdl.handle.net/20.500.14619/9658Nowadays, industry 4.0, many new ideas have come up, and one important one is predictive maintenance in modern manufacturing and production systems. This approach capitalizes on the wealth of data generated by sensors and real-time monitoring of machine components. The abundance of this data has paved the way for the application of Deep Learning (DL) techniques, enabling accurate prediction and diagnosis of failures. Consequently, precise prediction and diagnosis of component failures have become imperative for reducing machine downtime, cutting associated costs, extending machine life cycles, enhancing product quality, and fortifying overall reliability. This paper introduces an innovative framework that harnesses a hybrid approach, uniting Convolutional Neural Networks (CNN) and Long Short-Term Memory networks (LSTM), referred to as CNN-LSTM, to address the challenges of predictive maintenance. The performance and accuracy of this novel hybrid model are evaluated using the publicly accessible MetroPT dataset, with the objective of predicting component failures in Air Production Units (APUs) installed in metro vehicles. The experimental results showcase remarkable performance, achieving an F-Score about of 92% for binary classification and an impressive 97% for multiple classifications. Comparative analysis with related studies underscores the superiority of the proposed CNN-LSTM hybrid predictive maintenance approach, emphasizing its enhanced accuracy and prediction capabilities. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.eninfo:eu-repo/semantics/closedAccessCNNDeep learningLSTMMetroPTPredictive maintenanceBook Part10.1007/978-3-031-51997-0_112-s2.0-85189546849153Q4137223