The paper introduces an unprecedented technique for enhancing the measurement accuracy of thermal gradients in additive manufacturing by applying Explainable Artificial Intelligence (XAI) combined with Long Short-Term...
详细信息
The paper introduces an unprecedented technique for enhancing the measurement accuracy of thermal gradients in additive manufacturing by applying Explainable Artificial Intelligence (XAI) combined with Long Short-Term Memory (LSTM) networks. Temperature control during fabricating of thermoplastic and thermosetting polymer composites remains fundamental since it dictates the resulting material properties, structural integrity, and layer bonding. Our LSTM model predicts temperature changes during printing operations, providing stable manufacturing processes by reducing common defects, including warping deformation delamination and inconsistent curing. The framework now includes Layer-wise Relevance Propagation (LRP) and SHapley Additive exPlanations (SHAP) techniques to analyse crucial elements affecting temperature distribution. The thermal model shows 0.183degrees C MAE and 0.224 degrees C RMSE values, delivering 46.5 % better precision than previous thermal modelling approaches. Evaluation tests report system adjustments to take place after 86 ms during thermal perturbations and reach full stabilisation by 2.3 s. Its 97.3 % prediction accuracy rate anddecision consistency make it dependable for optimal temperature processing control. The research creates a connection between automated thermal modelling and manufacturing through additive technology which leads to better process management together with material efficiency as well as environmentally responsible creation of polymer parts. The introduced framework shows great promise for use in the aerospace, automotive and biomedical sectors of additive manufacturing, which demand accurate thermal control.
暂无评论