Optimizing Manufacturing Process With Knowledge Graph-Based Adaptive Neural Network: Approach to Industry 5.0 Consumer Electronics

作者：Chang Su, Yong Han, Qi Jiang, Tao Wang, Qingchen He · 发表于：IEEE Transactions on Consumer Electronics · 年份：2024 · DOI：10.1109/tce.2024.3513393 · 被引用次数：4 · 研究领域：Digital Transformation in Industry、Manufacturing Process and Optimization

In the era of Industry 5.0, where smart systems and individual consumer needs are paramount, this study addresses the critical issue of precision product manufacturing, particularly in the aerospace sector. Traditionally, process parameter decision relies on workers’ knowledge or data-driven AI, each with its pros and cons. Minor deviations in process parameters can significantly impact the final product’s conformity, leading to increased waste and unsustainable practices. This research explores the integration of human expertise and machine intelligence to optimize parameters effectively. Our novel approach, Knowledge Graph-Based Adaptive Neural Network Optimization (KG-ANN), integrates knowledge graph-based feature selection with neural network models. It uses an enhanced sparrow search algorithm (SSA) for weight initialization in a backpropagation (BP) neural network, improving prediction and optimization of crucial parameters. Tested within the manufacturing framework of precision electronic components for aero-engine enterprises, KG-ANN showed a 50% improvement in prediction accuracy and a 45% increase in precision compared to traditional BP models. The product conformity rates increased to 82.6%-88.4%. These findings underscore the potential of advanced technologies in fulfilling Industry 5.0’s vision of high-quality, human-centric outputs and support sustainable manufacturing goals through reduced waste and improved resource efficiency.