A knowledge graph-based intelligent inference method for sustainable design requirements across the product lifecycle

Intelligent inference of sustainable design requirements can transform ambiguous user demands into measurable product design requirements. However, customers prioritise the use phase, neglecting other lifecycle stages, complicating full lifecycle sustainability optimisation and impeding multi-dimensional design guidance in function, structure, material and process. Thus, a knowledge graph-based reasoning approach for product lifecycle sustainability requirements is proposed. Firstly, an improved fuzzy Kano model is introduced to integrate sustainability demands, employing TF-IDF-BERT to normalise demand and intensity adverbs in component descriptions, establishing a five-level quantification system. It enhances the Kano evaluation mechanism to assess distinctiveness and importance of different stakeholders’ demands. A credibility assessment mechanism, combined with information entropy theory, is further incorporated to analyze the multiple interactive relationships among evaluation results, such as complementarity and conflict, forming a comprehensive demand set that reduces uncertainty and subjectivity. Secondly, a quintuple knowledge ontology – “product-component-sustainability demand-design operation – design requirement” – is constructed by analyzing interactions between demands and multi-dimensional design operations. Mapping relationships are automatically extracted using the LEBERT-BiLSTM-CRF model, building sustainable design requires intelligent reasoning knowledge graphs. Finally, an intelligent inference support system is developed for smartphone screen design, achieving F1 scores of 93.22% and 92.53% for entity recognition and relationship extraction, demonstrating its effectiveness.