This paper presents a feature-based paradigm for intelligent manufacturing, highlighting the operational influence of critical variables temperature, operating mode, power consumption, network latency, production speed, and error rate on system efficiency. Four hybrid models were created by integrating Stacking Classification (SC) and Gaussian Process Classification (GPC) with Artificial Rabbit Optimization (ARO) and Coronavirus Herd Immunity Optimizer (CHIO) to elucidate intricate feature connections and improve prediction accuracy. The models—STCO, STAO, GPCO, and GPAO were evaluated using metrics such as Accuracy, F1-score, and Matthews Correlation Coefficient. STAO attained the greatest test accuracy (0.981) and MCC (0.972), thereby validating its exceptional performance. Feature importance analysis indicated that production speed and error rate are the most significant variables. SHAP and FAST studies provided additional insights, indicating that interaction effects among characteristics predominantly influence model behavior. The findings indicate that hybrid intelligent models utilizing feature-level input priority provide enhanced predicted accuracy and increased explainability, rendering them appropriate for real-time industrial application.
Mingyu Li (Sun,) studied this question.