A Data-Driven Human–Machine Collaborative Product Design System Toward Intelligent Manufacturing

In the era of big data, enterprises have accumulated large amounts of valuable data throughout the entire product life cycle (PLC). Such PLC data contains a wealth of design knowledge. Intelligent manufacturing seeks to establish a collaborative platform that integrates advanced data analytics and artificial intelligence into the manufacturing process, providing new opportunities for efficient and intelligent product design. Mining design knowledge from PLC data and applying it to the design stage is a critical issue that urgently needs to be addressed for data-driven product design (DDPD). To enhance the efficiency and adaptability of DDPD, this work proposes a comprehensive framework for extracting design knowledge from PLC data and utilizing the knowledge to inform the design process. A structured storage method is developed to manage PLC data with multi-source and heterogeneous characteristics. Then, human-machine collaborative pattern extraction, deep learning-based relation extraction, and other data mining techniques are used to extract knowledge from PLC data. Moreover, a product design knowledge network is constructed based on knowledge graph to achieve knowledge organization and management. Finally, a novel intelligent push method for product design knowledge, based on context navigation, is proposed as part of the framework. A case study showcases how data-driven human-machine collaborative patterns can be used to improve the flexibility and performance of product design.
Note to Practitioners
—Data-driven method can realize the closed-loop design of products while linking users, products and production processes to improve design efficiency. However, one of the major challenges in DDPD is the need to flexibly extract knowledge from PLC data and push them to designers. In this work, we propose a novel system that leverages human-machine collaboration and deep learning methods to realize DDPD toward intelligent manufacturing. It allows us to extract knowledge from product data, and then proactively push appropriate knowledge to designers for decision-making. The proposed system consists of three main components: product life cycle multi-source heterogeneous data processing, product design knowledge mining, and design knowledge intelligent pushing. Specifically, the human-machine collaboration mechanism improves the system’s capability to address uncertain and complex problems. A case study using shield machine PLC data has demonstrated the feasibility and effectiveness of the proposed framework.