AI-driven Life Cycle Assessment for sustainable hybrid manufacturing and remanufacturing

A novel AI-based hybrid approach is presented to predict sustainable performance in hybrid manufacturing and remanufacturing processes. It is an ensemble AI model that features Convolutional Neural Networks (CNN), Random Forests, Long Short-Term Memory (LSTM) networks, and Gradient Boosting Machines (GBM) integrated with a hybrid Life Cycle Assessment (LCA) framework to perform a dynamic and complete sustainability assessment. The model was trained, validated, and tested for over 18 months using data collected from three international manufacturing facilities that specialise in producing vehicle parts. Compared with traditional LCA methods, results show a 23% increase in the accuracy of environmental impact prediction, an 18% reduction in Root Mean Square Error (RMSE), and a 31% reduction in assessment time. It achieves an R2 value of 0.89, whereas that of the conventional LCA method is only 0.44. Integrating AI with hybrid LCA can effectively address the two main limitations of traditional LCA methods: uncertainty and granularity, which are improved by 15% and 20%, respectively. Therefore, this AI-based hybrid approach with real-time sustainability optimisation can be applied to a manufacturing context with principles of Industry 4.0 and the circular economy.