Optimization of Big Data-Driven Enterprise Production Scheduling and Decision Support Systems Utilizing Deep Learning and Visualization Technologies

Abstract

Modern manufacturing systems face escalating complexity in balancing dynamic production demands with supply chain volatility under Industry 4.0 paradigms. In response to this, we propose a three-stage neural architecture combining channel-wise feature recalibration, cross-dimensional attention mechanisms, and self-supervised temporal learning to optimize adaptive scheduling decisions. Validation across automotive and semiconductor manufacturing scenarios achieved 89.6% resilience to supply disruptions and 47.8% faster policy convergence than conventional methods. This work establishes a physics-aware framework for intelligent scheduling that bridges multi-source data fusion with operational constraints, offering transformative potential for future smart factory ecosystems.