Data-Driven Process Management for High-Quality Hip Joint Implants—A Case Study in Investment Casting
针对熔模铸造企业，提出一个质量4.0部署框架，通过分析髋关节植入物铸造过程中的质量数据和流程日志，利用机器学习模型预测缺陷和机械性能，实现零缺陷生产，大幅降低废品率。
Quality 4.0 aims to make zero-defect manufacturing possible across industries through automating and digitizing quality functions. Casting companies, primarily, are small and medium scale and suffer huge production losses from the rejection of cast components in the postproduction quality checks. While meeting the clients' mechanical strength, surface finish, and dimensional accuracy specifications is vital, casting companies should upgrade their production processes digitally and adhere to industry-specific regulatory mandates in order to compete in the international markets. This article proposes a Quality 4.0 deployment framework for investment casting companies. We adopt a case-based research methodology, and the case company makes metallic hip joint implants using the conventional investment casting process. Analyzing the quality data and process logs obtained from the case company's archives, we characterize each defect type and mechanical property, exploring various machine learning algorithms, and the best-fit models are those with high predictive performance. We perform within-case analysis to quantify the influence of potential causal variables and show the causal relationships among the implants' quality characteristics. Furthermore, the proposed quality management system with real-time process sensing capability involves a repetitive quality inferencing scheme to predict the quality of the items being cast, which enables process regulation and further leads to the production of zero-defective castings. Our quality inference models detect defects with an overall average balanced accuracy of 0.88% and predict mechanical properties with an average root mean squared error (RMSE) of 0.09, thereby drastically reducing the postproduction rejection rates in the case company. Hence, this case study fortifies the causal relationship of data-driven process management with the process outcomes of zero-defective casting of parts. The proposed Quality 4.0 theoretical framework, hence, emphasizes the feedback intervention and self-regulation capabilities of the casting companies for effective process management and encourages future researchers to investigate the role of data-driven process management in enhancing customer satisfaction and driving operational performance.