Cigarette Quality Problem Prediction Based on WOA‐BiLSTM‐SPA

ABSTRACT In modern cigarette production, proactive quality prediction is essential for ensuring product consistency and manufacturing efficiency, yet traditional post‐occurrence defect detection methods often lead to delays, inefficiencies, and higher costs. To address these limitations, this paper proposes a deep learning framework (WOA‐BiLSTM‐SPA) integrating the whale optimisation algorithm (WOA), bidirectional long short‐term memory networks (BiLSTM), and sparse attention mechanism (SPA) for proactive quality prediction. The methodology first preprocesses production data using correlation analysis and Kalman filtering to enhance reliability, then employs WOA to optimise BiLSTM‐SPA's hyperparameters, and finally deploys the model for continuous quality prediction. Experimental results demonstrate that WOA‐BiLSTM‐SPA outperforms conventional machine learning and deep learning benchmarks, achieving superior performance in both mean squared error (MSE) and coefficient of determination R 2 . This framework enables early defect detection, significantly improving production efficiency while reducing costs, and offers potential applications across industrial quality control systems.