Learning Temporal Patterns for Process Refinement in Smart Manufacturing using ChronGRU-TCN

AI-enabled process optimization is currently a way to achieve Smart Manufacturing, productivity enhancement, cost reduction of operation, and sustainable data-driven decision-making. Because manufacturing systems are dynamic and complex, AI implementation can bring intelligent automation, predictive maintenance, and real-time optimization. Nevertheless, deep learning models of today will be haunted by debilitating limitations in the form of temporal factors modeling abilities, inefficient sequencing relationship handling, inference latency, and failure to learn under dynamic production environments. With an aim to eliminate such constraints, the present paper presents a novel deep learning model known as ChronGRU-TCN, which encompasses the advantages of Gated Recurrent Units (GRUs) and Temporal Convolutional Networks (TCNs). The presented model can learn long-and short-term temporal interactions at low computation. GRU layers maintained temporal dynamics of sequential data, and TCN layers incorporating causal convolution and residual connections added more temporal information to achieve strong features. The above system enhances prediction task performance without side effect on inference speed or interpretability. It is obtained through gigantic experiments on a group of benchmark datasets where the proposed model surpasses its competitor architectures like CNN-LSTM, BiLSTM, Transformer, GRU-Attention, TCN, and ResNet-1D. The model demonstrates its potency and usability in using it in intelligent manufacturing by recording excellent retrievals—98% accuracy, 98.5% recall, 0.48s inference time, and 0.981 ROC-AUC.