Title：AI-Based Industrial Data Analytics: A Case Study in Metal Additive Manufacturing

Abstract：Metal Additive manufacturing (MAM) is a popular manufacturing technique which is broadly exploited in rapid prototyping and fabricating components with complex geometries. To ensure the stability of the MAM process, it is of critical importance to carry out data analytics on MAM process by monitoring the sensor data collected and detecting potential defects/outliers. This goal of the data analytic leads to the development of a knowledge-based system which is to readapt Product engineering stages: 1) Building AI model to detect future deviations caused by complex geometries to propose alternative geometry changes; and 2) Modifying the manufacturing strategy based on trained AI algorithm to avoid deposition paths that cause final distortions or heat accumulation. In this talk, we focus on the defect detection of thermal image data and outlier detection of welding sensor data based on artificial intelligence techniques. In the first part, a novel image processing method, an image-enhancement generative adversarial network, with aim to improve the contrast ratio of the thermal images for image segmentation will be discussed. In the second part, a novel clustering-based outlier detection method for anomaly detection will be introduced. The proposed methods are exploited in analyzing the real-world industrial data collected from a wire arc MAM pilot line in Sweden.

Speaker：Weibo Liu，Brunel University, Doctor.

Date：3:30pm-5:30pm 2023-7-14（Friday）.

Venue: 208,School of Mathematical Science

Organizer：School of Mathematical Science

Students and teachers are welcome.