Artificial intelligence (AI) has been gaining traction in various industries, including manufacturing. The combination of zero-defect manufacturing (ZDM) and AI provides new opportunities for advanced problem-solving and quality management in manufacturing. By using machine learning algorithms, AI can process complex datasets and make optimal decisions in real-time, enabling smart manufacturing.
However, implementing AI in manufacturing is not a straightforward process and requires overcoming several challenges, such as data quality and the lack of a standardized approach to collecting data from production systems. The consequences of false positives and false negatives should also be carefully considered when building an AI solution.
A case study was conducted at a manufacturing company producing heavy-duty vehicles to apply AI and identify insights contributing to ZDM. The study used various data collection techniques, including interviews, observations, and technical documentation, to gain knowledge of the production process and the specific issue at hand. The Cross-Industry Standard Process for Data Mining (CRISP-DM) process was followed to develop an AI model to support defect detection. The study found that by detecting defects early in the production process, the manufacturing company was able to save time and money by avoiding the need for replacements and rework.
The study also highlights the importance of data availability and quality for AI-assisted ZDM strategy development. To label axles as defective or approved, semi-supervised learning was used to create a more accurate dataset. In the second phase, features were extracted from the vibration curves using sampling and binning strategies. In the third phase, a supervised ML algorithm was trained to estimate the defect similarity ratio based on the extracted features. The AI solution enables defect detection at the component level, contributing to zero-waste strategies and supporting traditional quality methods such as DOE.
Another scientific paper emphasized the importance of implementing a ZDM strategy in industrial settings, specifically in the automotive industry. The paper suggests using AI to develop a detection strategy that can identify defects early in the manufacturing process and prevent them from occurring. The authors acknowledge the limitations of their study and the need for further testing and calibration of the AI solution.
Implementing AI in manufacturing can be a useful tool for defect detection and continuous quality improvement. However, it is important to consider the challenges and limitations of AI and ZDM strategies. A standardized and stable assembly process is necessary to enable data-driven methods and make valid statistical conclusions. Furthermore, companies need to experiment with AI to develop their data science know-how and integrate it into their business models and processes. Overall, the combination of ZDM and AI has the potential to transform manufacturing and improve business performance by preventing defects and errors in production processes.
Authors
Leberruyer, N., Bruch, J., Ahlskog, M., & Afshar, S. (2023). Toward Zero Defect Manufacturing with the support of Artificial Intelligence—Insights from an industrial application. Computers in Industry, 147, 103877. https://doi.org/10.1016/j.compind.2023.103877