Innovative Development of Hybrid LSTM-CNN Architecture with Optimized Parameters for Enhanced Predictive Maintenance Performance

Maintenance is a significant cost factor in manufacturing, accounting for up to 60 of the total factory manufacturing costs. Optimizing these operations is crucial to minimize costs and avoid the “Run to Failure” approach. This research paper focuses on the practical applications of machine learning, especially using Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, to develop learning models that improve predictive maintenance. Our study uses the C-MAPSS, NCMAPSS, and NASA battery datasets to predict machinery’s Remaining Useful Life (RUL) and the State of Health (SOH) of lithium batteries, making our findings directly applicable to the real-world maintenance prediction scenario, establishing confidence in the practical ability of our research.The results show the potential of Parallel CNN-LSTM models, optimized with Genetic Algorithms (GA), to offer significantly improved prediction accuracy over traditional models. This breakthrough paves the way for enhanced predictive maintenance, reduced maintenance frequency and costs, and improved machine and plant availability, inspiring a new era of maintenance in manufacturing operations. The reductions in RMSE directly translate into improved maintenance scheduling, reduced downtime, and significant cost savings for industries such as aviation, energy, and manufacturing.This work demonstrates a scalable and optimized approach that outperforms traditional predictive maintenance models by integrating spatial and temporal feature extraction with GA optimization. These advancements ensure that the models can handle both high variability in sensor data and complex operational environments.This research paper significantly contributes to the field by showing the practical application of machine learning in predictive maintenance. Our findings offer a tangible solution to the industry’s challenge of reducing costs and increasing efficiency in manufacturing operations, thereby potentially revolutionizing the way maintenance is approached in the future.